{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":216,"total_is_capped":false,"direct_labels_cover":0,"predictions_cover":216,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12","author_layer_release":"2026-06-26"},"query_hash":"8e06b9c4de61","filters":{"topic":"Authorship Attribution and Profiling"}},"results":[{"id":"W1014449310","doi":"10.1609/icwsm.v6i1.14340","title":"Homophily and Latent Attribute Inference: Inferring Latent Attributes of Twitter Users from Neighbors","year":2021,"lang":"en","type":"article","venue":"Proceedings of the International AAAI Conference on Web and Social Media","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":319,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Homophily; Assortativity; Inference; Computer science; Context (archaeology); Instant messaging; Machine learning; Artificial intelligence; World Wide Web; Psychology; Social psychology; Geography; Complex network","authors":[{"name":"Faiyaz Al Zamal","is_ca":true},{"name":"Wendy Liu","is_ca":true},{"name":"Derek Ruths","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06759890956918559,"gpt":0.2797267150628564,"spread":0.2121278054936708,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003514486,0.0005996496,0.0009368378,0.002604025,0.0009200519,0.001744895,0.001123294,0.001049822,0.001755642],"category_scores_gemma":[0.02083715,0.000359541,0.0007960451,0.00198272,0.000602923,0.004436338,0.001558434,0.001253901,0.0007937892],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004712521,"about_ca_system_score_gemma":0.0005113673,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002779505,"about_ca_topic_score_gemma":0.003089232,"domain_scores_codex":[0.997376,0.001409536,0.0001247499,0.0005574913,0.0003691568,0.0001630671],"domain_scores_gemma":[0.9836556,0.01122292,0.00135453,0.002821418,0.0006234149,0.0003222213],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001272531,0.0008673458,0.423631,0.0003777755,0.001184216,0.0004786409,0.002620172,0.1274503,0.005159604,0.03439084,0.003889416,0.3986781],"study_design_scores_gemma":[0.00002570468,0.00008802312,0.02643022,0.0000473708,0.000114257,0.000258814,0.0006773386,0.9131336,0.003915716,0.05387377,0.001389259,0.00004590936],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.482888,0.0007303031,0.5093342,0.0007365027,0.00005925149,0.00009768425,0.001146313,0.0006892769,0.004318489],"genre_scores_gemma":[0.9743229,0.0001076929,0.02443077,0.00004943309,0.00005735159,0.00002444163,0.0004899382,0.00001783433,0.0004996461],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003514486,"threshold_uncertainty_score":0.01858658,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2170770919","doi":"10.1109/cmpsac.2004.1342667","title":"N-gram-based detection of new malicious code","year":2004,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":281,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"","keywords":"Computer science; Code (set theory); Signature (topology); Computer virus; Software; Malware; Data mining; Artificial intelligence; Programming language; Computer security","authors":[{"name":"Tony Abou-Assaleh","is_ca":true},{"name":"N. Cercone","is_ca":true},{"name":"Vlado Kešelj","is_ca":true},{"name":"Ray Sweidan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03563911430917237,"gpt":0.2790708994228318,"spread":0.2434317851136594,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007903779,0.0007409422,0.0006025247,0.00309685,0.0005946964,0.0005242432,0.0005681038,0.0007991585,0.0009834934],"category_scores_gemma":[0.00530362,0.0001153199,0.0003742852,0.001183059,0.0003147877,0.0009447788,0.000669959,0.0005630562,0.001253864],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003562329,"about_ca_system_score_gemma":0.0005809945,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003568069,"about_ca_topic_score_gemma":0.009542163,"domain_scores_codex":[0.9986603,0.0002582237,0.0001290504,0.0002606887,0.0005244445,0.0001672951],"domain_scores_gemma":[0.9959438,0.001533422,0.0005265342,0.000444021,0.001225943,0.0003262037],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001594139,0.0007770638,0.1077979,0.0004659638,0.000159736,0.00125631,0.000592344,0.02022168,0.2060589,0.001186234,0.01270902,0.6471807],"study_design_scores_gemma":[0.00002171022,0.0003156546,0.05985561,0.00003350445,0.00004701627,0.001254218,0.0001834035,0.8244221,0.107278,0.001791702,0.004721113,0.00007604991],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8939402,0.0005851144,0.09068161,0.0003601181,0.0001692218,0.0001992765,0.003257146,0.006731578,0.004075688],"genre_scores_gemma":[0.9198371,0.0001095977,0.07438412,0.0000569337,0.0000486381,0.00006781598,0.003275309,0.00009787403,0.002122597],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003568069,"threshold_uncertainty_score":0.007094622,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2025882237","doi":"10.1093/llc/fqm023","title":"Bigrams of Syntactic Labels for Authorship Discrimination of Short Texts","year":2007,"lang":"en","type":"article","venue":"Literary and Linguistic Computing","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":226,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Bigram; Library science; History; Computer science; Media studies; Artificial intelligence; Sociology","authors":[{"name":"Graeme Hirst","is_ca":true},{"name":"Olga Feiguina","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03613429507286367,"gpt":0.3221275605100855,"spread":0.2859932654372218,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001250786,0.000822917,0.0006374476,0.007263493,0.0009023172,0.001277055,0.000798577,0.000958774,0.003384542],"category_scores_gemma":[0.009121668,0.000282069,0.0004195137,0.004150359,0.0005722203,0.002260718,0.001020239,0.00121426,0.00304708],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002592934,"about_ca_system_score_gemma":0.0005707532,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005190025,"about_ca_topic_score_gemma":0.001324931,"domain_scores_codex":[0.9986418,0.0004535987,0.0001337437,0.0002236856,0.0004291409,0.0001180817],"domain_scores_gemma":[0.9905875,0.005664883,0.0008117433,0.001184386,0.001259158,0.0004923127],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0005522504,0.000362426,0.02143077,0.0005938265,0.0001121744,0.0003604423,0.001017566,0.002243269,0.08248636,0.005700402,0.005188175,0.8799524],"study_design_scores_gemma":[0.0002026741,0.0009334332,0.1142479,0.0005148273,0.000375125,0.003041518,0.002585597,0.5718406,0.191836,0.07950472,0.03446101,0.0004566762],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3452111,0.002196694,0.6283318,0.0006810051,0.0004495041,0.0004274306,0.003670903,0.005651139,0.01338038],"genre_scores_gemma":[0.6679199,0.0004450398,0.3239074,0.00008117477,0.0002560789,0.0003016604,0.003426128,0.0003879197,0.003274735],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007263493,"threshold_uncertainty_score":0.01132238,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2284070151","doi":"","title":"What's in a Name? Using First Names as Features for Gender Inference in Twitter","year":2013,"lang":"en","type":"article","venue":"National Conference on Artificial Intelligence","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":168,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Classifier (UML); Inference; Computer science; Artificial intelligence; Social media; Natural language processing; Baseline (sea); Information retrieval; Machine learning; World Wide Web","authors":[{"name":"Wendy Liu","is_ca":true},{"name":"Derek Ruths","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2696613820475069,"gpt":0.4140428855734447,"spread":0.1443815035259378,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001765912,0.0005129256,0.0004586785,0.003410674,0.00087775,0.001315651,0.0004942392,0.0007485378,0.002592571],"category_scores_gemma":[0.01205645,0.0001923624,0.0003527796,0.002051606,0.0003733035,0.003477206,0.0007762527,0.0008380301,0.002564924],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006499884,"about_ca_system_score_gemma":0.0005700513,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003952281,"about_ca_topic_score_gemma":0.007285049,"domain_scores_codex":[0.998578,0.0005653019,0.0001082671,0.0003256454,0.0002644286,0.0001583873],"domain_scores_gemma":[0.9936317,0.00323388,0.001397221,0.0005491881,0.0009381375,0.0002499779],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001682214,0.0002341257,0.5202211,0.0004952151,0.0001230053,0.0007208169,0.00358485,0.005801283,0.01566314,0.008284723,0.03373238,0.4094571],"study_design_scores_gemma":[0.00007417639,0.0003580031,0.4110777,0.000451867,0.0002041952,0.001704688,0.009817742,0.3839645,0.0484373,0.04299179,0.1006497,0.0002683571],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8143128,0.002474348,0.1309502,0.00316158,0.0006328753,0.0001881737,0.02254044,0.002355075,0.0233846],"genre_scores_gemma":[0.9593914,0.0004815685,0.02997141,0.0002127437,0.0002519008,0.00008355264,0.006246823,0.0001054,0.003255283],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003952281,"threshold_uncertainty_score":0.009339094,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2125962012","doi":"10.1016/j.diin.2010.03.003","title":"Mining writeprints from anonymous e-mails for forensic investigation","year":2010,"lang":"en","type":"article","venue":"Digital Investigation","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":146,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Exploit; Cluster analysis; Stylometry; Anonymity; Identification (biology); Classifier (UML); Cybercrime; Focus (optics); Writing style; Information retrieval; World Wide Web; Data science; Data mining; Artificial intelligence; Computer security; The Internet","authors":[{"name":"Farkhund Iqbal","is_ca":true},{"name":"Hamad Binsalleeh","is_ca":true},{"name":"Benjamin C. M. Fung","is_ca":true},{"name":"Mourad Debbabi","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04413737110212299,"gpt":0.2634196282676218,"spread":0.2192822571654988,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002736854,0.0007698136,0.000697794,0.01348977,0.001373236,0.002439719,0.0008922673,0.001409183,0.003144258],"category_scores_gemma":[0.02955322,0.0003559741,0.0004821336,0.008154401,0.0005021053,0.002376371,0.00164747,0.0008320834,0.005449221],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000383343,"about_ca_system_score_gemma":0.001221121,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000434737,"about_ca_topic_score_gemma":0.0009384266,"domain_scores_codex":[0.994091,0.001228553,0.0009818757,0.0007595256,0.002370002,0.0005690475],"domain_scores_gemma":[0.9504859,0.01921589,0.01176661,0.008107142,0.008720583,0.001704051],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001261488,0.0006632371,0.3534002,0.001039289,0.0001781805,0.004060037,0.003596753,0.003100144,0.03670503,0.006260365,0.02593942,0.563796],"study_design_scores_gemma":[0.0001058108,0.001097546,0.4096667,0.001001475,0.0007170813,0.01817424,0.01643811,0.1300568,0.203346,0.03048776,0.1886107,0.0002977234],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9057133,0.001623818,0.06044876,0.001000734,0.0004663545,0.0004299869,0.01647409,0.002294646,0.01154833],"genre_scores_gemma":[0.9261526,0.0008444351,0.0465415,0.0001481763,0.000454444,0.0002153361,0.01505832,0.0002181424,0.01036699],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01348977,"threshold_uncertainty_score":0.01447403,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2250194349","doi":"10.18653/v1/d13-1114","title":"Gender Inference of Twitter Users in Non-English Contexts","year":2013,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":133,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"Social Sciences and Humanities Research Council of Canada; Natural Sciences and Engineering Research Council of Canada; Public Safety Canada","keywords":"Inference; Computer science; Orthography; Natural language processing; Social media; Artificial intelligence; Data science; Linguistics; World Wide Web","authors":[{"name":"Morgane Ciot","is_ca":true},{"name":"Morgan Sonderegger","is_ca":true},{"name":"Derek Ruths","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04556533857021437,"gpt":0.2934247514939073,"spread":0.2478594129236929,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003245334,0.0003673969,0.000418167,0.001139047,0.0006124938,0.001409627,0.0004517835,0.0005741806,0.004013023],"category_scores_gemma":[0.02160423,0.000197603,0.0004769192,0.001113663,0.0004001763,0.002706019,0.00106195,0.0008982324,0.001693227],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004454716,"about_ca_system_score_gemma":0.0003816479,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006907291,"about_ca_topic_score_gemma":0.009454398,"domain_scores_codex":[0.9980797,0.0009908347,0.0001183961,0.0003470621,0.0002200308,0.0002440581],"domain_scores_gemma":[0.9915684,0.005205333,0.00086044,0.0009960046,0.001057472,0.00031246],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001167914,0.0002032788,0.8532783,0.0001609648,0.000148225,0.0004031429,0.006216015,0.004010012,0.004864778,0.008059033,0.006013799,0.1154745],"study_design_scores_gemma":[0.0000730692,0.0002584507,0.6230617,0.0002628474,0.000234982,0.0009503006,0.01884604,0.2667872,0.02074713,0.04776133,0.02083877,0.0001781422],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9428589,0.0004909392,0.03957625,0.001415924,0.000107107,0.00004805966,0.004320519,0.0002132812,0.0109691],"genre_scores_gemma":[0.9929678,0.0001117557,0.004011413,0.0000864138,0.00004031244,0.00001342173,0.00143553,0.000023853,0.001309394],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.006907291,"threshold_uncertainty_score":0.01716322,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2123407642","doi":"10.1016/j.diin.2008.05.001","title":"A novel approach of mining write-prints for authorship attribution in e-mail forensics","year":2008,"lang":"en","type":"article","venue":"Digital Investigation","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":131,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Authorship attribution; Suspect; Computer science; Attribution; Decision tree; Support vector machine; Quality (philosophy); Information retrieval; Data mining; Data science; Computer security; Artificial intelligence; Psychology","authors":[{"name":"Farkhund Iqbal","is_ca":true},{"name":"Rachid Hadjidj","is_ca":true},{"name":"Benjamin C. M. Fung","is_ca":true},{"name":"Mourad Debbabi","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1382981898189501,"gpt":0.2774152743910965,"spread":0.1391170845721464,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001634958,0.000696854,0.001029288,0.006485504,0.0009063369,0.001676474,0.001412654,0.001477657,0.0007330155],"category_scores_gemma":[0.009291749,0.0003982355,0.000802186,0.004088483,0.0006974593,0.00242967,0.001044763,0.0009566247,0.0005856178],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000350546,"about_ca_system_score_gemma":0.000889021,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007252286,"about_ca_topic_score_gemma":0.00113516,"domain_scores_codex":[0.9974917,0.0005328101,0.0003192521,0.0007190591,0.0007566535,0.0001805432],"domain_scores_gemma":[0.9945996,0.0019844,0.001152905,0.0008477832,0.001200244,0.0002150589],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003824673,0.001132561,0.06326223,0.000430859,0.0002197101,0.0006944423,0.0009081998,0.02191262,0.0218417,0.006278247,0.003272608,0.8796643],"study_design_scores_gemma":[0.00006647349,0.0003894044,0.02734105,0.00009325144,0.0001384559,0.002219816,0.0006872435,0.9183375,0.02710585,0.01631269,0.007204607,0.0001036541],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.2062648,0.0007668533,0.7882147,0.000609567,0.0001499778,0.0002528717,0.00075967,0.001386467,0.001595044],"genre_scores_gemma":[0.6039976,0.000315137,0.3925392,0.00008255549,0.0001730242,0.0001752187,0.0008403155,0.00004245376,0.001834571],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006485504,"threshold_uncertainty_score":0.008646548,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2041413176","doi":"10.3115/1073445.1073470","title":"Language and task independent text categorization with simple language models","year":2003,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":112,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Natural language processing; Categorization; Artificial intelligence; Task (project management); Language identification; Simple (philosophy); Language model; Variety (cybernetics); Independence (probability theory); Selection (genetic algorithm); Identification (biology); Feature (linguistics); Character (mathematics); Text categorization; Feature selection; Natural language; Linguistics","authors":[{"name":"Fuchun Peng","is_ca":true},{"name":"Dale Schuurmans","is_ca":true},{"name":"Shaojun Wang","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01593452888467338,"gpt":0.2515908326777858,"spread":0.2356563037931124,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002337298,0.001509378,0.001167884,0.002045991,0.0008253244,0.001713808,0.002123871,0.001941533,0.004718918],"category_scores_gemma":[0.007915046,0.0005162295,0.001770063,0.001522449,0.0008695837,0.004034683,0.002580086,0.001773308,0.005482635],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004869291,"about_ca_system_score_gemma":0.001305411,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001182922,"about_ca_topic_score_gemma":0.002875791,"domain_scores_codex":[0.9976329,0.0007488006,0.0001570341,0.0005941322,0.0006986551,0.0001685146],"domain_scores_gemma":[0.9966821,0.001340087,0.0002556035,0.0009581537,0.0005762494,0.0001878363],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004319178,0.0006311986,0.002153967,0.0004250611,0.0003145985,0.0002250059,0.0003941887,0.02889732,0.07072942,0.009486802,0.01196257,0.874348],"study_design_scores_gemma":[0.00009453876,0.0003941922,0.002208825,0.00004735649,0.0001077648,0.0004538406,0.0001713372,0.9080892,0.04037616,0.03721933,0.01070181,0.0001356991],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01173601,0.0001923304,0.9823314,0.0001633732,0.0001183709,0.0001893619,0.0002968798,0.003727485,0.001244913],"genre_scores_gemma":[0.1596035,0.0001849785,0.830322,0.0003318561,0.0002609791,0.000650864,0.001963203,0.0003170249,0.00636557],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004718918,"threshold_uncertainty_score":0.01578641,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2251433042","doi":"10.3115/v1/w14-5904","title":"Automatic Identification of Arabic Language Varieties and Dialects in Social Media","year":2014,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":90,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Arabic; Computer science; Identification (biology); Natural language processing; Social media; Language identification; Linguistics; Artificial intelligence; Natural language; World Wide Web; Botany","authors":[{"name":"Fatiha Sadat","is_ca":true},{"name":"Farzindar Kazemi","is_ca":false},{"name":"Atefeh Farzindar","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01806990496924178,"gpt":0.2669399753216908,"spread":0.248870070352449,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008752725,0.0005792789,0.0003818967,0.004159898,0.000706622,0.0012393,0.0003410477,0.0005291293,0.001470246],"category_scores_gemma":[0.004146681,0.0001582098,0.0003540178,0.001660319,0.0002497002,0.001869543,0.0009432223,0.0005094813,0.002242686],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000355879,"about_ca_system_score_gemma":0.0003263717,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002623127,"about_ca_topic_score_gemma":0.004596252,"domain_scores_codex":[0.9991266,0.0003011423,0.00007853698,0.0002122599,0.0001793822,0.0001021814],"domain_scores_gemma":[0.9975266,0.0009359519,0.0003846111,0.000273519,0.0007195079,0.0001598517],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001025122,0.0004986657,0.2708468,0.000606929,0.00014157,0.0008975473,0.003802615,0.003560876,0.05054202,0.003435937,0.02145712,0.6431848],"study_design_scores_gemma":[0.00006310661,0.0003746186,0.4543249,0.0003347096,0.0001885753,0.002926136,0.01418138,0.4061882,0.05882713,0.0143239,0.04807148,0.0001958421],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9358654,0.001292436,0.0455699,0.0006704699,0.0002643844,0.000269839,0.006366311,0.001561825,0.008139399],"genre_scores_gemma":[0.9479809,0.0003899872,0.04172231,0.00008631561,0.0001005767,0.00009137209,0.006239462,0.00004986307,0.003339113],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.004159898,"threshold_uncertainty_score":0.005215764,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2085509246","doi":"10.1016/j.diin.2009.01.004","title":"Towards an integrated e-mail forensic analysis framework","year":2009,"lang":"en","type":"article","venue":"Digital Investigation","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":86,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Phishing; Spamming; Law enforcement; Harassment; Computer security; Context (archaeology); Child pornography; Cybercrime; Digital forensics; The Internet; Internet privacy; World Wide Web; Data science","authors":[{"name":"Rachid Hadjidj","is_ca":true},{"name":"Mourad Debbabi","is_ca":true},{"name":"Hakim Lounis","is_ca":true},{"name":"Farkhund Iqbal","is_ca":true},{"name":"Adam Szporer","is_ca":true},{"name":"Djamel Benredjem","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03798295588258519,"gpt":0.2880873846024712,"spread":0.250104428719886,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007012168,0.001231271,0.001531556,0.005885012,0.001747084,0.009114946,0.003617381,0.002551292,0.002163807],"category_scores_gemma":[0.009473628,0.0009438525,0.002195224,0.002596828,0.001499407,0.008689633,0.006469801,0.002790518,0.001679728],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001248939,"about_ca_system_score_gemma":0.004266826,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002839852,"about_ca_topic_score_gemma":0.004614846,"domain_scores_codex":[0.9947795,0.001201555,0.0005785894,0.0007977075,0.00231806,0.0003247407],"domain_scores_gemma":[0.9933124,0.001494894,0.0006290728,0.001873494,0.002244938,0.0004451262],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003024941,0.001209923,0.008035474,0.0005452886,0.0005741271,0.0009185857,0.001408703,0.06595964,0.02392375,0.3295059,0.01425098,0.5533652],"study_design_scores_gemma":[0.00003292141,0.00009135954,0.001324461,0.0002020215,0.0002877275,0.0005841674,0.0005655433,0.7630568,0.0236903,0.1817719,0.02830014,0.00009264102],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00212757,0.00006980805,0.9940034,0.0001734799,0.00001541514,0.0001065855,0.00008485108,0.002716714,0.0007022418],"genre_scores_gemma":[0.07274595,0.0001687316,0.924455,0.0001349725,0.00004995695,0.0001421494,0.0005141214,0.0002475034,0.001541615],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.009114946,"threshold_uncertainty_score":0.03708434,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2914961791","doi":"10.1145/3292577","title":"Code Authorship Attribution","year":2019,"lang":"en","type":"review","venue":"ACM Computing Surveys","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":86,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of New Brunswick","funders":"New Brunswick Innovation Foundation","keywords":"Computer science; Malware; Attribution; Authorship attribution; Code (set theory); Field (mathematics); Cryptovirology; Categorization; Data science; Artificial intelligence; Computer security; Programming language; Set (abstract data type); Psychology","authors":[{"name":"Vaibhavi Kalgutkar","is_ca":true},{"name":"Ratinder Kaur","is_ca":true},{"name":"Hugo Gonzalez","is_ca":true},{"name":"Natalia Stakhanova","is_ca":true},{"name":"Alina Matyukhina","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2631097808434538,"gpt":0.4126861586154646,"spread":0.1495763777720108,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01039562,0.0009543793,0.001112093,0.01238217,0.002499507,0.005991335,0.001828333,0.002811765,0.02738067],"category_scores_gemma":[0.09908952,0.0003659549,0.0006513151,0.01234018,0.002692925,0.007091108,0.003899678,0.001720704,0.01628689],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002024719,"about_ca_system_score_gemma":0.005342808,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008827788,"about_ca_topic_score_gemma":0.0009773043,"domain_scores_codex":[0.9827711,0.005709412,0.001900804,0.001969503,0.006777773,0.000871282],"domain_scores_gemma":[0.9019348,0.04851487,0.01297933,0.01278805,0.02153509,0.002247824],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00006181095,0.00003486279,0.003571104,0.004735036,0.00005956007,0.0001491094,0.0007100572,0.0003330017,0.0003871475,0.05749892,0.1070481,0.8254113],"study_design_scores_gemma":[0.00001298113,0.00003189122,0.003154316,0.006423503,0.00006236751,0.001059132,0.0007263776,0.0005689332,0.001085902,0.03930755,0.947522,0.0000449404],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.01391977,0.5176393,0.08781175,0.0258873,0.02314328,0.001117861,0.005872788,0.002408712,0.3221993],"genre_scores_gemma":[0.261682,0.5146638,0.04336286,0.009669034,0.01374459,0.001513608,0.01076194,0.001117317,0.143485],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.02738067,"threshold_uncertainty_score":0.0915975,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2129364433","doi":"10.1016/j.diin.2014.03.012","title":"OBA2: An Onion approach to Binary code Authorship Attribution","year":2014,"lang":"en","type":"article","venue":"Digital Investigation","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":73,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Authorship attribution; Code (set theory); Binary number; Attribution; Information retrieval; World Wide Web; Programming language; Natural language processing; Arithmetic; Mathematics","authors":[{"name":"Saed Alrabaee","is_ca":true},{"name":"Noman Saleem","is_ca":true},{"name":"Stere Preda","is_ca":true},{"name":"Lingyu Wang","is_ca":true},{"name":"Mourad Debbabi","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07222085393703048,"gpt":0.2798793971174146,"spread":0.2076585431803841,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005638445,0.002344837,0.0019313,0.01347886,0.002167991,0.005595443,0.003432627,0.002495188,0.01061486],"category_scores_gemma":[0.02030686,0.001037473,0.002515738,0.005578621,0.001900244,0.009082415,0.009773344,0.00299408,0.005661183],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001559914,"about_ca_system_score_gemma":0.002383099,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003724997,"about_ca_topic_score_gemma":0.006826123,"domain_scores_codex":[0.9919528,0.001628775,0.0006049283,0.001755853,0.003197209,0.0008602796],"domain_scores_gemma":[0.9883562,0.003182909,0.001566611,0.004519201,0.001800593,0.0005744665],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0009929063,0.0005501937,0.02241216,0.0003882967,0.0003887706,0.001081411,0.0016366,0.01940351,0.009106222,0.06278166,0.04051956,0.8407387],"study_design_scores_gemma":[0.00009099083,0.0002228888,0.005759291,0.0002429025,0.0001644539,0.001197484,0.0008105217,0.7247509,0.02076602,0.1558158,0.09002583,0.0001528805],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02230486,0.0007239009,0.9254867,0.001162862,0.0005043003,0.0006452354,0.00226637,0.03005123,0.01685464],"genre_scores_gemma":[0.2817112,0.000617424,0.6764323,0.00099894,0.0005968587,0.0007754001,0.00717754,0.003853606,0.02783695],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01347886,"threshold_uncertainty_score":0.03551018,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2035395173","doi":"10.1145/1774088.1774428","title":"e-mail authorship verification for forensic investigation","year":2010,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":65,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Suspect; Phishing; Spamming; NIST; Matching (statistics); Context (archaeology); The Internet; Plagiarism detection; Sample (material); Computer security; Data mining; Information retrieval; World Wide Web; Natural language processing; Statistics","authors":[{"name":"Farkhund Iqbal","is_ca":true},{"name":"Liaquat Ali Khan","is_ca":true},{"name":"Benjamin C. M. Fung","is_ca":true},{"name":"Mourad Debbabi","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04174213818809263,"gpt":0.2857227118872353,"spread":0.2439805736991427,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["research_integrity"],"consensus_categories":[],"category_scores_codex":[0.00442196,0.0005827644,0.000802446,0.003400527,0.000805009,0.001883062,0.001042887,0.001565357,0.002649853],"category_scores_gemma":[0.02214236,0.0002248301,0.0004106894,0.001463046,0.0006515865,0.002548552,0.001315224,0.0008934839,0.002412226],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005144682,"about_ca_system_score_gemma":0.001058756,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000311104,"about_ca_topic_score_gemma":0.0005394816,"domain_scores_codex":[0.9951984,0.002795489,0.0002524371,0.0004509053,0.00113517,0.0001676227],"domain_scores_gemma":[0.9813387,0.007789937,0.0026749,0.004458705,0.003267242,0.0004704968],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004782413,0.0004402598,0.01670294,0.0002792333,0.00006539181,0.0002608297,0.0002712753,0.0126973,0.03371095,0.01352623,0.006322636,0.9152448],"study_design_scores_gemma":[0.00007012601,0.0005068544,0.02700053,0.0002940086,0.0001459792,0.003336979,0.001002023,0.7586728,0.1403062,0.04390131,0.0246408,0.0001222739],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1592423,0.001960182,0.8250788,0.001230604,0.0001848594,0.0003129296,0.0005731552,0.004161901,0.00725517],"genre_scores_gemma":[0.7575407,0.000628142,0.2389315,0.00007873555,0.0001235732,0.0001003664,0.0005887616,0.00005971747,0.001948535],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9984347,"threshold_uncertainty_score":0.02338588,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2576057564","doi":"10.1002/dac.3259","title":"Authorship verification using deep belief network systems","year":2017,"lang":"en","type":"article","venue":"International Journal of Communication Systems","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":63,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Toronto Metropolitan University; University of Victoria","funders":"","keywords":"Computer science; Merge (version control); Bernoulli's principle; Gaussian; Word error rate; Basis (linear algebra); Artificial intelligence; Block (permutation group theory); Natural language processing; Data mining; Theoretical computer science; Algorithm; Information retrieval; Mathematics","authors":[{"name":"Marcelo Luiz Brocardo","is_ca":true},{"name":"Issa Traoré","is_ca":true},{"name":"Isaac Woungang","is_ca":true},{"name":"Mohammad S. Obaidat","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09828633259928508,"gpt":0.3644836674045816,"spread":0.2661973348052965,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002607798,0.0006136798,0.000853806,0.001411948,0.0005266399,0.001761263,0.001153409,0.0012975,0.00189863],"category_scores_gemma":[0.0102581,0.0004407985,0.0006242468,0.0007919673,0.00090871,0.002495167,0.001386208,0.001701346,0.000489743],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001666809,"about_ca_system_score_gemma":0.001121508,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007227893,"about_ca_topic_score_gemma":0.005414616,"domain_scores_codex":[0.998419,0.0006802437,0.00009098513,0.0003175042,0.0003115574,0.0001806998],"domain_scores_gemma":[0.9938079,0.003823316,0.0006513781,0.0006360002,0.0009458056,0.0001357623],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002897356,0.000112359,0.003666638,0.00008467699,0.0001033956,0.00009855064,0.0001170372,0.8429381,0.001568035,0.01346111,0.001244742,0.1363156],"study_design_scores_gemma":[0.000002524149,0.000006067558,0.0001058602,0.00000394804,0.000003032959,0.000004756045,0.000004199157,0.9949734,0.0003631443,0.004453633,0.00007667381,0.000002700094],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1082774,0.0007803927,0.8843976,0.001164168,0.00009561376,0.00006991197,0.0002344416,0.001458696,0.003521834],"genre_scores_gemma":[0.9554681,0.0001370455,0.04245707,0.00009533272,0.00003706071,0.00003253576,0.0001524606,0.00002257193,0.001597747],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007227893,"threshold_uncertainty_score":0.01437163,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2045244401","doi":"10.1016/j.jcss.2014.12.019","title":"Authorship verification of e-mail and tweet messages applied for continuous authentication","year":2014,"lang":"en","type":"article","venue":"Journal of Computer and System Sciences","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":60,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Toronto Metropolitan University; University of Victoria","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Stylometry; Word error rate; Support vector machine; Set (abstract data type); Feature selection; Feature (linguistics); Authentication (law); Writing style; Information retrieval; Selection (genetic algorithm); Block (permutation group theory); Artificial intelligence; Natural language processing; Data mining; Computer security; Mathematics","authors":[{"name":"Marcelo Luiz Brocardo","is_ca":true},{"name":"Issa Traoré","is_ca":true},{"name":"Isaac Woungang","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02652990645237236,"gpt":0.2698352296278049,"spread":0.2433053231754325,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004767463,0.0005042456,0.0009009118,0.002197944,0.001101878,0.002029434,0.0008013888,0.001437916,0.002725327],"category_scores_gemma":[0.03081246,0.0002219122,0.0004024087,0.001355648,0.0005612005,0.002321547,0.001511363,0.001043639,0.002001038],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005612471,"about_ca_system_score_gemma":0.001303172,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000605089,"about_ca_topic_score_gemma":0.0005849405,"domain_scores_codex":[0.9936911,0.002276492,0.0006293337,0.0008724642,0.001852822,0.0006777945],"domain_scores_gemma":[0.9591695,0.01981974,0.004948449,0.007420041,0.00752463,0.001117607],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.006215017,0.001107153,0.07369824,0.0007344205,0.0002963533,0.001288553,0.001425277,0.03399746,0.1013377,0.01895233,0.006762632,0.7541848],"study_design_scores_gemma":[0.0001063418,0.0007803144,0.02509939,0.0000991594,0.0001245555,0.0009985809,0.0007300801,0.8666978,0.08878075,0.01244889,0.004031383,0.0001027549],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6457821,0.0007510817,0.3432447,0.0006555188,0.0007066503,0.0004463355,0.001058802,0.002334178,0.005020681],"genre_scores_gemma":[0.9702556,0.0000776673,0.02785976,0.00002390733,0.00007869899,0.00003829495,0.0002642541,0.00001748904,0.001384342],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.004767463,"threshold_uncertainty_score":0.02521306,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2251483280","doi":"10.18653/v1/d15-1088","title":"Mr. Bennet, his coachman, and the Archbishop walk into a bar but only one of them gets recognized: On The Difficulty of Detecting Characters in Literary Texts","year":2015,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":59,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Character (mathematics); Archbishop; Computer science; Bar (unit); State (computer science); Artificial intelligence; Natural language processing; History; Mathematics; Algorithm; Physics","authors":[{"name":"Hardik Vala","is_ca":true},{"name":"David Jurgens","is_ca":true},{"name":"Andrew Piper","is_ca":true},{"name":"Derek Ruths","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06396051724858276,"gpt":0.2615950096518417,"spread":0.197634492403259,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002761591,0.001117562,0.0006308522,0.004826061,0.003365069,0.003647454,0.001113653,0.002055708,0.01047466],"category_scores_gemma":[0.01604456,0.0004406537,0.0004359698,0.003534223,0.0009580225,0.005553405,0.002659669,0.002807271,0.01725922],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008679103,"about_ca_system_score_gemma":0.0009461151,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003874379,"about_ca_topic_score_gemma":0.01937169,"domain_scores_codex":[0.9973798,0.0006148397,0.0001915596,0.0006944467,0.0009595206,0.0001598044],"domain_scores_gemma":[0.9899673,0.003162258,0.0006760785,0.002536406,0.002791465,0.0008664445],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0005927154,0.0001736402,0.03094416,0.0003640252,0.00007079679,0.001148703,0.002053243,0.002306619,0.0145085,0.01502109,0.2756159,0.6572007],"study_design_scores_gemma":[0.00003202046,0.0001661045,0.02244587,0.0005555462,0.00005343902,0.006041253,0.005543214,0.05542721,0.05148065,0.03914984,0.8188731,0.0002317133],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3115061,0.01685191,0.3901984,0.105176,0.01614972,0.00047865,0.01503196,0.02090503,0.1237023],"genre_scores_gemma":[0.5025133,0.004655847,0.2989618,0.006847105,0.001763149,0.0002465022,0.01572971,0.002424044,0.1668585],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01047466,"threshold_uncertainty_score":0.03504121,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2103195385","doi":"10.1145/2632188.2632207","title":"Automatic identification of arabic dialects in social media","year":2014,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":58,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Computer science; Natural language processing; Artificial intelligence; Arabic; Classifier (UML); Hidden Markov model; Naive Bayes classifier; Modern Standard Arabic; Social media; Language identification; Context (archaeology); Probabilistic logic; n-gram; Language model; Linguistics; Natural language; Support vector machine; World Wide Web","authors":[{"name":"Fatiha Sadat","is_ca":true},{"name":"Farnazeh Kazemi","is_ca":false},{"name":"Atefeh Farzindar","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03165082704362224,"gpt":0.2818454130505005,"spread":0.2501945860068783,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008734317,0.0006868309,0.0004299079,0.004536909,0.000746098,0.001469307,0.0003643997,0.0006106441,0.00232403],"category_scores_gemma":[0.003553954,0.0001848973,0.0004024227,0.001637361,0.0002741011,0.001743803,0.001002388,0.0005318662,0.003852466],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004524586,"about_ca_system_score_gemma":0.0004185596,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003409623,"about_ca_topic_score_gemma":0.004765383,"domain_scores_codex":[0.9990131,0.0003586413,0.0000828594,0.0002269874,0.0002045674,0.0001138389],"domain_scores_gemma":[0.9974664,0.0008349915,0.0003437668,0.000274662,0.0009409044,0.0001392497],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001085771,0.0003988144,0.1164774,0.0007309649,0.0001566386,0.001437275,0.004093414,0.003169986,0.07751209,0.004073606,0.03771956,0.7531444],"study_design_scores_gemma":[0.00007399413,0.0004074427,0.2723301,0.0004146614,0.0002284531,0.003213195,0.01536625,0.4871541,0.1051296,0.01567825,0.09974212,0.0002619048],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.866618,0.001873542,0.0951743,0.001342794,0.0006337159,0.0004541253,0.01267966,0.006055143,0.01516864],"genre_scores_gemma":[0.9024004,0.0005617341,0.07971725,0.0001848336,0.0002078554,0.0001427005,0.009659707,0.0001506278,0.006974859],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.004536909,"threshold_uncertainty_score":0.007774711,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4220816919","doi":"10.1126/sciadv.abm2463","title":"Based on billions of words on the internet, <scp>people</scp> = <scp>men</scp>","year":2022,"lang":"en","type":"article","venue":"Science Advances","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":53,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"York University","keywords":"Similarity (geometry); Context (archaeology); Measure (data warehouse); The Internet; Linguistics; Computer science; Psychology; Similarity measure; Social psychology; Natural language processing; Sociology; World Wide Web; Artificial intelligence; History","authors":[{"name":"April H. Bailey","is_ca":false},{"name":"Adina Williams","is_ca":false},{"name":"Andrei Cimpian","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02511934123538057,"gpt":0.2815690368670433,"spread":0.2564496956316628,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0009531016,0.0008055745,0.0004801397,0.009848638,0.001248601,0.002782255,0.0003730898,0.0007881697,0.008189142],"category_scores_gemma":[0.008345485,0.0002222881,0.0003108279,0.01746688,0.001599964,0.005566981,0.001703399,0.0008407736,0.004422661],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007669192,"about_ca_system_score_gemma":0.0007492275,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003211865,"about_ca_topic_score_gemma":0.005509136,"domain_scores_codex":[0.9985548,0.00031956,0.0003220722,0.0002848404,0.0004631329,0.00005552548],"domain_scores_gemma":[0.9940262,0.003027087,0.001403603,0.0004415509,0.0008836435,0.0002179677],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0007836614,0.0001482975,0.09401266,0.007702911,0.0005074582,0.002723788,0.02838877,0.001577211,0.0190203,0.09488308,0.2522082,0.4980437],"study_design_scores_gemma":[0.00004149637,0.000135113,0.1863835,0.001595921,0.0001996414,0.004658591,0.01528652,0.003708792,0.004502331,0.02772379,0.755583,0.0001812272],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5711943,0.03906748,0.03582389,0.01359357,0.004206842,0.0007260188,0.2051091,0.002327774,0.127951],"genre_scores_gemma":[0.8279544,0.01562109,0.05258653,0.001991472,0.001522654,0.0006805063,0.08012147,0.000694797,0.01882706],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9990469,"threshold_uncertainty_score":0.02739543,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2891582949","doi":"10.18653/v1/d18-1294","title":"Syntax Encoding with Application in Authorship Attribution","year":2018,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":46,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa; National Research Council Canada","funders":"National Key Research and Development Program of China; Beijing Advanced Innovation Center for Big Data and Brain Computing; State Key Laboratory of Software Development Environment; National Natural Science Foundation of China","keywords":"Computer science; Syntax; Natural language processing; Sentence; Encoding (memory); Artificial intelligence; Abstract syntax; Embedding; Parsing; ENCODE; Abstract syntax tree; Word (group theory); Representation (politics); Parse tree; Tree (set theory); Lossless compression; Linguistics; Data compression; Mathematics","authors":[{"name":"Richong Zhang","is_ca":false},{"name":"Zhiyuan Hu","is_ca":false},{"name":"Hongyu Guo","is_ca":true},{"name":"Yongyi Mao","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02970367137682244,"gpt":0.2849840804760073,"spread":0.2552804090991848,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001883016,0.001094659,0.000814672,0.002771507,0.0007201408,0.002116267,0.001609157,0.001745608,0.005557085],"category_scores_gemma":[0.01650161,0.0004500731,0.001024017,0.003185831,0.001290813,0.009693616,0.003089441,0.002641221,0.002131433],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00102766,"about_ca_system_score_gemma":0.001447355,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001214652,"about_ca_topic_score_gemma":0.001333929,"domain_scores_codex":[0.9980837,0.0007646951,0.0002196287,0.0004899232,0.0003157928,0.0001263051],"domain_scores_gemma":[0.9938486,0.002817108,0.0006110903,0.00182159,0.0007388659,0.0001627509],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000368459,0.000305621,0.004252607,0.0004915275,0.00009409347,0.0003921291,0.0008380603,0.0867096,0.01152768,0.1221491,0.01576093,0.7571102],"study_design_scores_gemma":[0.00006175503,0.0001114187,0.0009462601,0.0001193144,0.00007206241,0.000342382,0.0002141633,0.5458046,0.01514004,0.422738,0.01435462,0.00009546827],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02031086,0.0004691565,0.9684189,0.0009009428,0.0002101089,0.0001032393,0.001572603,0.005841017,0.002173244],"genre_scores_gemma":[0.5213835,0.0006335392,0.4684877,0.0004707898,0.0002944492,0.0003048842,0.003735293,0.0007247025,0.003965071],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005557085,"threshold_uncertainty_score":0.01859033,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2250174871","doi":"","title":"Classifying English Documents by National Dialect","year":2013,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":42,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"British National Corpus; Linguistics; Variety (cybernetics); Variation (astronomy); Computer science; Natural language processing; Intuition; Artificial intelligence; Psychology","authors":[{"name":"Marco Lui","is_ca":false},{"name":"Paul Cook","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02166335923556349,"gpt":0.2665022127154754,"spread":0.2448388534799119,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001458323,0.0004795828,0.0005140118,0.009016614,0.001003089,0.002346317,0.0003741366,0.000420092,0.002317403],"category_scores_gemma":[0.009384949,0.0001542377,0.0003753477,0.007450703,0.000448046,0.001583727,0.0009061313,0.0004206231,0.001324794],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001059711,"about_ca_system_score_gemma":0.001141982,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02838944,"about_ca_topic_score_gemma":0.05389735,"domain_scores_codex":[0.9988298,0.0002227694,0.000208686,0.0002739732,0.0003136721,0.0001511622],"domain_scores_gemma":[0.9906883,0.00442545,0.0008761577,0.000863582,0.00282074,0.0003258167],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0005742231,0.0003762207,0.4971851,0.00153329,0.0003176052,0.0006198373,0.005889709,0.002906125,0.03021995,0.003983611,0.02769412,0.4287003],"study_design_scores_gemma":[0.00006755948,0.0001807093,0.7812376,0.0003810341,0.0003823051,0.001347601,0.01077878,0.04765545,0.03640736,0.004752112,0.116669,0.0001405085],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.9383478,0.003576119,0.01245818,0.0005826675,0.0001328736,0.000263947,0.0269882,0.0009031879,0.01674714],"genre_scores_gemma":[0.914391,0.001611522,0.03912862,0.0001163819,0.000101218,0.000175538,0.0373931,0.0001306404,0.006951964],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.02838944,"threshold_uncertainty_score":0.0564484,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1543116020","doi":"","title":"Web-based inference detection","year":2007,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":41,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Inference; Salient; Information retrieval; Identification (biology); Web application; Data science; Data mining; Artificial intelligence; World Wide Web","authors":[{"name":"Jessica Staddon","is_ca":false},{"name":"Philippe Golle","is_ca":false},{"name":"Bryce Zimny","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03315902922342352,"gpt":0.3023352610639297,"spread":0.2691762318405062,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01429224,0.001190541,0.001682869,0.00969508,0.001427167,0.005859466,0.00372874,0.002221305,0.00554397],"category_scores_gemma":[0.09779859,0.0008605681,0.001259746,0.004239505,0.001395974,0.00709903,0.003349358,0.002730297,0.002809496],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001643687,"about_ca_system_score_gemma":0.002448516,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004468136,"about_ca_topic_score_gemma":0.003992239,"domain_scores_codex":[0.9745339,0.005423717,0.0022902,0.004396348,0.01256462,0.0007912283],"domain_scores_gemma":[0.8664294,0.08622798,0.01254468,0.0206467,0.01308846,0.001062846],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0009511001,0.0005093159,0.06002844,0.0009614106,0.0004613363,0.001492337,0.002032486,0.02356987,0.02318838,0.06880824,0.02927141,0.7887257],"study_design_scores_gemma":[0.00008441694,0.0001301088,0.01403872,0.0001563424,0.0002058056,0.00141948,0.0003974907,0.7648615,0.09124108,0.08214045,0.04515519,0.0001694419],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03564144,0.0005536286,0.923868,0.000713699,0.0001628184,0.0005473198,0.002803654,0.02843341,0.007275953],"genre_scores_gemma":[0.4399524,0.0004272469,0.5419997,0.0006459935,0.0003146054,0.0006410272,0.006592902,0.002058989,0.007367108],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01429224,"threshold_uncertainty_score":0.07558548,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2251370914","doi":"","title":"Robust, Lexicalized Native Language Identification","year":2012,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":40,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Artificial intelligence; Natural language processing; Classifier (UML); Training set; Task (project management); Identification (biology); Set (abstract data type); Language identification; Test set; Natural language","authors":[{"name":"Julian Brooke","is_ca":true},{"name":"Graeme Hirst","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0716369537548322,"gpt":0.3186020371403878,"spread":0.2469650833855556,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006571268,0.001225091,0.001496895,0.002713992,0.00118603,0.003971785,0.001905279,0.002122818,0.006804669],"category_scores_gemma":[0.02881436,0.0004464128,0.0005445026,0.001253026,0.000922802,0.005986751,0.005574277,0.001662388,0.0103405],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004909578,"about_ca_system_score_gemma":0.001208031,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001288412,"about_ca_topic_score_gemma":0.001943539,"domain_scores_codex":[0.9901154,0.00337622,0.0006865365,0.003076663,0.002252609,0.0004925265],"domain_scores_gemma":[0.9843875,0.005176863,0.0008599926,0.004834547,0.004300371,0.0004406546],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0008668639,0.000781297,0.01802954,0.0009459354,0.0002756432,0.0006974015,0.00245416,0.01209038,0.144017,0.00706928,0.02851664,0.784256],"study_design_scores_gemma":[0.0002445332,0.001101195,0.07847854,0.0003571915,0.0002459796,0.003776429,0.00569536,0.5561177,0.2518595,0.05046263,0.05107416,0.0005868498],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.4496232,0.00142132,0.4876615,0.00073397,0.0007829505,0.0009320139,0.006463311,0.02199879,0.03038307],"genre_scores_gemma":[0.78747,0.0002248843,0.1937339,0.0003024923,0.0001682323,0.0005018577,0.008319804,0.001238031,0.0080408],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006804669,"threshold_uncertainty_score":0.03475255,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1595390009","doi":"","title":"Cybergenre: automatic identification of home pages on the web","year":2004,"lang":"en","type":"article","venue":"Journal of Web Engineering","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":40,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"","keywords":"Web page; Home page; Computer science; World Wide Web; Classifier (UML); Information retrieval; Static web page; Web development; The Internet; Artificial intelligence","authors":[{"name":"Michael Shepherd","is_ca":true},{"name":"Carolyn Watters","is_ca":true},{"name":"Alistair Kennedy","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01742234240319396,"gpt":0.2319652182478575,"spread":0.2145428758446635,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006991696,0.0004712734,0.0004645302,0.00547107,0.0003238225,0.001083506,0.0003929017,0.000555839,0.002562352],"category_scores_gemma":[0.002589651,0.0001480045,0.0002600815,0.001659679,0.0002016907,0.001966214,0.0006500983,0.0003932257,0.002060993],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002910169,"about_ca_system_score_gemma":0.0002781237,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001675951,"about_ca_topic_score_gemma":0.00305488,"domain_scores_codex":[0.9994608,0.0001415848,0.00004599652,0.0001287626,0.0001592258,0.00006368572],"domain_scores_gemma":[0.9980952,0.0006870924,0.0003027863,0.0003786519,0.0003919767,0.0001443538],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0008828549,0.0005534642,0.1017389,0.000482791,0.0001170419,0.0005770517,0.0006385991,0.004080236,0.03143503,0.001799649,0.01772431,0.8399702],"study_design_scores_gemma":[0.00008390744,0.0004569212,0.2794515,0.0001743153,0.0001130429,0.002444292,0.001547172,0.5982065,0.08206464,0.006963381,0.02838984,0.000104475],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8749635,0.001252102,0.08725727,0.0002586371,0.0001400272,0.0003959782,0.006961149,0.01734345,0.0114278],"genre_scores_gemma":[0.8783457,0.0004575335,0.1015119,0.0000532489,0.00007881867,0.0001156066,0.01053847,0.0002402934,0.008658449],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.00547107,"threshold_uncertainty_score":0.008571923,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2155085300","doi":"10.1109/hicss.2008.53","title":"An Examination of Genre Attributes for Web Page Classification","year":2008,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"","keywords":"Computer science; Web page; Set (abstract data type); Information retrieval; Identification (biology); World Wide Web; Recall; Precision and recall","authors":[{"name":"Lei Dong","is_ca":true},{"name":"Carolyn Watters","is_ca":true},{"name":"Jack Duffy","is_ca":true},{"name":"Michael Shepherd","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1191895850919988,"gpt":0.3147126104816039,"spread":0.1955230253896051,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02758641,0.001110414,0.001501405,0.004288444,0.001723798,0.00393315,0.0008136441,0.0009738601,0.001250474],"category_scores_gemma":[0.1126928,0.0003740002,0.001280372,0.004867392,0.0008324605,0.004370969,0.0009079208,0.00245849,0.0007330229],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009368391,"about_ca_system_score_gemma":0.0009086823,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002358591,"about_ca_topic_score_gemma":0.003266965,"domain_scores_codex":[0.9806874,0.01189356,0.001719689,0.001632005,0.003632244,0.0004349641],"domain_scores_gemma":[0.8187288,0.1564519,0.005588732,0.009028865,0.007904597,0.002297011],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.003782719,0.002671527,0.2883328,0.001233948,0.0009941119,0.000323703,0.001382782,0.01431471,0.02202571,0.001236077,0.008835657,0.6548662],"study_design_scores_gemma":[0.0005543825,0.009446396,0.4723636,0.0008851315,0.00157567,0.00318239,0.00351305,0.413423,0.05145427,0.00966251,0.03330858,0.0006310703],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9305342,0.005001844,0.04941183,0.001374005,0.0007084468,0.0007615857,0.003265453,0.001767573,0.007174913],"genre_scores_gemma":[0.9124268,0.001005101,0.08129075,0.0002402466,0.000274891,0.0002012951,0.003274521,0.0001867525,0.001099628],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02758641,"threshold_uncertainty_score":0.1458927,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2527936838","doi":"10.63317/4e2puhxuby8y","title":"Discriminating Similar Languages: Evaluations and Explorations","year":2016,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"National Research Council Canada","funders":"","keywords":"Computer science; Natural language processing; Artificial intelligence; Task (project management); Oracle; Digital subscriber line; Annotation; Machine learning; Programming language","authors":[{"name":"Cyril Goutte","is_ca":true},{"name":"Serge Léger","is_ca":true},{"name":"Shervin Malmasi","is_ca":false},{"name":"Marcos Zampieri","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05894784463913573,"gpt":0.3522310635662659,"spread":0.2932832189271302,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02576143,0.004389539,0.003231489,0.006945222,0.001886513,0.004616154,0.002964584,0.003693311,0.002191132],"category_scores_gemma":[0.08654507,0.0007557407,0.001788765,0.005445751,0.001937109,0.006718908,0.004313369,0.002526149,0.002034829],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001870571,"about_ca_system_score_gemma":0.001330283,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006661866,"about_ca_topic_score_gemma":0.006729185,"domain_scores_codex":[0.9582401,0.02457889,0.003253248,0.004968994,0.006813744,0.002144941],"domain_scores_gemma":[0.900428,0.07429386,0.002353644,0.01172543,0.009033212,0.002165768],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.007581749,0.003888005,0.0602864,0.003227945,0.002147167,0.001466282,0.003687308,0.1359321,0.02103069,0.003962701,0.04290391,0.7138858],"study_design_scores_gemma":[0.0006966834,0.003284429,0.03617486,0.0006697603,0.00102927,0.002607484,0.005197299,0.8647785,0.0443079,0.02266376,0.01812195,0.0004680755],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8927265,0.01749084,0.06058958,0.001680552,0.0004931168,0.0005379546,0.003722861,0.008498242,0.01426031],"genre_scores_gemma":[0.9315621,0.001480379,0.05352144,0.0004541501,0.0001907402,0.0002256325,0.008559514,0.001360175,0.002645771],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02576143,"threshold_uncertainty_score":0.1362411,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2963515066","doi":"10.1145/3209978.3210008","title":"SynTF","year":2018,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"European Commission","keywords":"Computer science; Information retrieval; Authorship attribution; Vector space model; Suspect; Task (project management); Term (time); Obfuscation; Identity (music); Stylometry; Contrast (vision); Data mining; Artificial intelligence; Computer security","authors":[{"name":"Benjamin Weggenmann","is_ca":false},{"name":"Florian Kerschbaum","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02695464514839649,"gpt":0.2845048336781714,"spread":0.257550188529775,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001468568,0.001751888,0.001253496,0.002638384,0.001995416,0.003901599,0.003057522,0.002557835,0.2365586],"category_scores_gemma":[0.004999822,0.0009918597,0.001388051,0.002443913,0.000842322,0.004441543,0.003899152,0.002185554,0.2268308],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001471984,"about_ca_system_score_gemma":0.002154326,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003218113,"about_ca_topic_score_gemma":0.003428758,"domain_scores_codex":[0.9979028,0.0002864616,0.0001750736,0.0005276644,0.0007875783,0.0003204142],"domain_scores_gemma":[0.9979261,0.0003863216,0.0001193573,0.0008197071,0.0006042066,0.0001442691],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0005822465,0.0001566644,0.002199659,0.0008813133,0.00006980707,0.0005281602,0.0003771,0.002164926,0.0109697,0.06438972,0.6255641,0.2921166],"study_design_scores_gemma":[0.00007769296,0.0000743986,0.0007697984,0.0001122678,0.00002908023,0.0006256024,0.0001421998,0.007166673,0.008815902,0.02559859,0.9565353,0.0000524367],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"other","genre_scores_codex":[0.01210255,0.002310691,0.3067466,0.003063491,0.002634084,0.000720971,0.1072656,0.2438965,0.3212596],"genre_scores_gemma":[0.09507395,0.00198522,0.1956939,0.003194506,0.0008605554,0.001220808,0.3398724,0.04277418,0.3193245],"genre_candidate":"other","genre_consensus":"other","teacher_disagreement_score":0.7634414,"threshold_uncertainty_score":0,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2773782067","doi":"","title":"Chat Disentanglement: Identifying Semantic Reply Relationships with Random Forests and Recurrent Neural Networks","year":2017,"lang":"en","type":"article","venue":"International Joint Conference on Natural Language Processing","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Thread (computing); Random forest; Recurrent neural network; Classifier (UML); Artificial intelligence; Machine learning; Artificial neural network; Natural language processing; Data mining; Programming language","authors":[{"name":"Shikib Mehri","is_ca":false},{"name":"Giuseppe Carenini","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06351610469634573,"gpt":0.3261799200032158,"spread":0.2626638153068701,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003155158,0.001503502,0.0009468862,0.002752788,0.0007243079,0.001159694,0.001670173,0.001323856,0.001641202],"category_scores_gemma":[0.008988333,0.0004414877,0.001084341,0.001421497,0.0004702559,0.00229264,0.001247327,0.0022187,0.001566684],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005925386,"about_ca_system_score_gemma":0.0009297446,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005649163,"about_ca_topic_score_gemma":0.01186592,"domain_scores_codex":[0.9987155,0.0004504313,0.00008588241,0.000409263,0.0001814825,0.0001575541],"domain_scores_gemma":[0.995201,0.002746862,0.0006579058,0.0005203387,0.0006354248,0.0002384384],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0009850193,0.001041862,0.03801064,0.0002991072,0.0003162561,0.00049001,0.001373655,0.1247446,0.02222825,0.008482952,0.01237323,0.7896544],"study_design_scores_gemma":[0.00001332752,0.00004492322,0.001656668,0.00001342148,0.00002442117,0.00003675342,0.00008850621,0.9881421,0.002580139,0.006482776,0.0009024745,0.00001453598],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1688306,0.0007042571,0.8211468,0.0004794054,0.0001562678,0.0002240039,0.00114413,0.005163367,0.002151193],"genre_scores_gemma":[0.7638399,0.000176022,0.2265305,0.0001431346,0.000175316,0.0002484096,0.003817233,0.0002879011,0.004781576],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005649163,"threshold_uncertainty_score":0.01668626,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2148429219","doi":"10.1214/10-aoas378","title":"Detecting multiple authorship of United States Supreme Court legal decisions using function words","year":2011,"lang":"en","type":"article","venue":"The Annals of Applied Statistics","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Supreme court; Style (visual arts); Function (biology); Legal writing; Writing style; Statistical analysis","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.2448864315332253,"gpt":0.347889792177648,"spread":0.1030033606444227,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00575741,0.0004954014,0.0004931457,0.01025941,0.001046472,0.002671517,0.0004690576,0.0007387932,0.001435474],"category_scores_gemma":[0.08043879,0.0002283329,0.0003710158,0.005525404,0.00118372,0.002737869,0.001476462,0.0007525908,0.0005552983],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005420168,"about_ca_system_score_gemma":0.0006207642,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008307722,"about_ca_topic_score_gemma":0.001241353,"domain_scores_codex":[0.992353,0.003857824,0.001169582,0.0009074754,0.00130895,0.0004031732],"domain_scores_gemma":[0.8277453,0.1250159,0.02660729,0.009003758,0.009429147,0.002198548],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0005703649,0.0001551373,0.8283071,0.0001797888,0.0001352689,0.0005070437,0.008539611,0.001753425,0.008934806,0.003688864,0.001065399,0.1461632],"study_design_scores_gemma":[0.00005359523,0.0004381701,0.8638601,0.0001906005,0.0002016015,0.002242994,0.01326714,0.06342433,0.02217809,0.02719643,0.006750641,0.0001963639],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9905393,0.0001814239,0.007059358,0.00007097789,0.00002763042,0.00002321938,0.000244178,0.00006341421,0.001790432],"genre_scores_gemma":[0.9962519,0.00003719314,0.003156318,0.00001005826,0.00001995221,0.0000147895,0.0002245507,0.00001441085,0.0002708788],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01025941,"threshold_uncertainty_score":0.03044844,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4392238204","doi":"10.3390/info15030131","title":"Authorship Attribution Methods, Challenges, and Future Research Directions: A Comprehensive Survey","year":2024,"lang":"en","type":"article","venue":"Information","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":29,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto; York University","funders":"York University","keywords":"Attribution; Data science; Authorship attribution; Survey research; Computer science; Psychology; Applied psychology; Natural language processing; Social psychology","authors":[{"name":"Xie He","is_ca":true},{"name":"Arash Habibi Lashkari","is_ca":true},{"name":"Nikhill Vombatkere","is_ca":true},{"name":"Dilli Prasad Sharma","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.255335490359698,"gpt":0.4439869260173184,"spread":0.1886514356576204,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04743221,0.001919125,0.002627963,0.02416645,0.003397082,0.01235876,0.003784343,0.003111773,0.005698478],"category_scores_gemma":[0.2335697,0.0008036018,0.001312766,0.03037653,0.003285546,0.02341585,0.004006236,0.003027878,0.004195843],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002634594,"about_ca_system_score_gemma":0.003809854,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001695486,"about_ca_topic_score_gemma":0.001433064,"domain_scores_codex":[0.9492635,0.02290155,0.006420431,0.007009729,0.01325079,0.001154065],"domain_scores_gemma":[0.6294324,0.2923003,0.01796898,0.02915982,0.02804181,0.003096655],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001411529,0.0001414386,0.03843274,0.004617826,0.0001804028,0.00008211374,0.001862227,0.003075101,0.0003856675,0.03037818,0.03260465,0.8880986],"study_design_scores_gemma":[0.00009473289,0.0002378205,0.05137688,0.01348921,0.000416344,0.003142225,0.01098688,0.07617706,0.005732324,0.312833,0.5249836,0.0005299714],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.07190178,0.4302938,0.3960884,0.0363596,0.005824539,0.0009535691,0.01159113,0.004818629,0.04216864],"genre_scores_gemma":[0.5361188,0.2320991,0.1880758,0.003855171,0.00945791,0.001117383,0.0177746,0.001335536,0.01016578],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.04743221,"threshold_uncertainty_score":0.2508485,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2250418537","doi":"","title":"Author Verification Using Common N-Gram Profiles of Text Documents","year":2014,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"","keywords":"Computer science; n-gram; Natural language processing; Artificial intelligence; Classifier (UML); Set (abstract data type); Class (philosophy); Identification (biology); Information retrieval; Language model","authors":[{"name":"Magdalena Jankowska","is_ca":true},{"name":"Evangelos Milios","is_ca":true},{"name":"Vlado Kešelj","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04998160390730941,"gpt":0.3299500645796463,"spread":0.2799684606723369,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001976226,0.0009130938,0.001102358,0.007679213,0.001176205,0.002246869,0.0008547279,0.001323838,0.001325374],"category_scores_gemma":[0.01172572,0.0002368472,0.0006949042,0.004544138,0.0004899505,0.003823898,0.002331339,0.0008608416,0.002431691],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006437936,"about_ca_system_score_gemma":0.0008550158,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00171542,"about_ca_topic_score_gemma":0.003522143,"domain_scores_codex":[0.9954601,0.001027038,0.0005146459,0.001213201,0.001460686,0.000324334],"domain_scores_gemma":[0.9909207,0.003201607,0.001730626,0.00187256,0.001789706,0.0004847312],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001622425,0.0005018772,0.05265907,0.0007118862,0.0001975644,0.0008565885,0.001369335,0.009837159,0.07651111,0.004616748,0.008364639,0.8427517],"study_design_scores_gemma":[0.00009222982,0.0005978553,0.06350665,0.0002294227,0.0001651083,0.003876163,0.001831264,0.7375559,0.1338671,0.0277193,0.03037692,0.0001821765],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6984717,0.002632815,0.2740919,0.000480609,0.0002924617,0.0003941022,0.005947844,0.007097823,0.01059074],"genre_scores_gemma":[0.8467505,0.0004468881,0.1412669,0.0000537196,0.0001050575,0.0001090575,0.008194058,0.0001971183,0.002876703],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.007679213,"threshold_uncertainty_score":0.01045138,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2011672789","doi":"10.1075/ssol.4.1.02nic","title":"Toward a science of science fiction","year":2014,"lang":"en","type":"article","venue":"Scientific Study of Literature","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"University of British Columbia; John Templeton Foundation","keywords":"Operationalization; Fantasy; Test (biology); Fiction theory; Literary fiction; Literature; Techno-thriller; Humanism; Literary criticism; Computer science; Linguistics; Epistemology; Art; Philosophy","authors":[{"name":"Ryan Nichols","is_ca":false},{"name":"Justin Lynn","is_ca":true},{"name":"Benjamin Grant Purzycki","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03059280971897983,"gpt":0.2935062790963196,"spread":0.2629134693773398,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01376923,0.0005798189,0.0007375379,0.01258346,0.005441948,0.02067191,0.001281983,0.002776643,0.003389166],"category_scores_gemma":[0.04760298,0.0004229082,0.0006071508,0.007873432,0.03504195,0.02197196,0.005326427,0.005138295,0.00149987],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003720014,"about_ca_system_score_gemma":0.003660573,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001136824,"about_ca_topic_score_gemma":0.001088772,"domain_scores_codex":[0.9895647,0.006041975,0.0004611046,0.001164759,0.002564571,0.0002029539],"domain_scores_gemma":[0.955111,0.02840704,0.003608597,0.005050125,0.006535378,0.001287898],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00002446547,0.00005259829,0.005119504,0.0002610252,0.00001414182,0.00008419817,0.05873054,0.0002493398,0.000818775,0.8784984,0.006484739,0.04966236],"study_design_scores_gemma":[0.00001007815,0.00002006231,0.002164997,0.000378881,0.00000912287,0.0001422052,0.02096246,0.0009660331,0.0006524932,0.8433429,0.1313325,0.00001820347],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.1686011,0.01110098,0.2928053,0.1155487,0.003256032,0.0005967173,0.0008108895,0.0005442313,0.406736],"genre_scores_gemma":[0.7942697,0.005027997,0.1725004,0.006081705,0.001743162,0.0006773314,0.0005472203,0.0003348142,0.01881768],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02067191,"threshold_uncertainty_score":0.07281947,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2147608180","doi":"10.1109/hicss.2009.68","title":"An N-Gram Based Approach to Automatically Identifying Web Page Genre","year":2009,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"Killam Trusts","keywords":"Computer science; Web page; Feature (linguistics); Information retrieval; Set (abstract data type); Front page; Listing (finance); World Wide Web; Static web page; Artificial intelligence; Web navigation","authors":[{"name":"Jack Mason","is_ca":true},{"name":"Michael Shepherd","is_ca":true},{"name":"Jack Duffy","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04993646941629454,"gpt":0.3202699916950935,"spread":0.270333522278799,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001501497,0.001303614,0.00104254,0.008518155,0.001127335,0.001435326,0.000848992,0.001146792,0.002718869],"category_scores_gemma":[0.007037117,0.0003021383,0.0009176338,0.005598708,0.0004943883,0.002561122,0.001032006,0.001198802,0.003398503],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005581747,"about_ca_system_score_gemma":0.0008347665,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0044045,"about_ca_topic_score_gemma":0.009400314,"domain_scores_codex":[0.9982457,0.0005237036,0.0001679891,0.000425091,0.0005061958,0.0001312971],"domain_scores_gemma":[0.9955992,0.002024443,0.0004629182,0.0006380239,0.001027113,0.0002483632],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0007351995,0.0006288423,0.01875754,0.0003930122,0.0001660685,0.0002555299,0.0006047529,0.002510432,0.04827377,0.002189423,0.01205384,0.9134315],"study_design_scores_gemma":[0.0001910519,0.0008420915,0.0850508,0.0002124754,0.0003488096,0.002917798,0.001749737,0.7830234,0.05692299,0.0215787,0.04686559,0.0002966226],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2554587,0.003257402,0.6965175,0.001048669,0.0005485586,0.001320943,0.007385842,0.01866552,0.01579694],"genre_scores_gemma":[0.3553881,0.0007311791,0.6274088,0.0002309513,0.0003086697,0.0004866339,0.008215684,0.0003920235,0.006837876],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008518155,"threshold_uncertainty_score":0.00909549,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2066783897","doi":"10.1145/2699910","title":"A Visualizable Evidence-Driven Approach for Authorship Attribution","year":2015,"lang":"en","type":"article","venue":"ACM Transactions on Information and System Security","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University; Concordia University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Stylometry; Computer science; Authorship attribution; Context (archaeology); Identification (biology); Attribution; The Internet; Data science; Information retrieval; World Wide Web; Natural language processing","authors":[{"name":"Steven H. H. Ding","is_ca":true},{"name":"Benjamin C. M. Fung","is_ca":true},{"name":"Mourad Debbabi","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1074829618696274,"gpt":0.3201323593290201,"spread":0.2126493974593927,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0105476,0.001434388,0.001250962,0.008343361,0.00113735,0.006107213,0.004058761,0.003105052,0.003831064],"category_scores_gemma":[0.06413136,0.0008806087,0.001598055,0.003547027,0.002559663,0.007638124,0.006807819,0.003815936,0.001230122],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001500216,"about_ca_system_score_gemma":0.001651887,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001343078,"about_ca_topic_score_gemma":0.001681501,"domain_scores_codex":[0.9886108,0.006581163,0.0006597639,0.001822295,0.001978412,0.0003476354],"domain_scores_gemma":[0.9511378,0.03028277,0.004613786,0.007494764,0.005365197,0.001105494],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0009290891,0.0006148062,0.01483372,0.001082066,0.0003073118,0.001189895,0.003944532,0.1328925,0.01619728,0.1511982,0.007015775,0.6697949],"study_design_scores_gemma":[0.00007600648,0.000176107,0.002198214,0.0002174851,0.00007906627,0.0005183289,0.0007262999,0.7566369,0.01118518,0.2163486,0.0117354,0.0001024155],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.008464939,0.0004254099,0.9869468,0.0006039563,0.0000554706,0.0002426904,0.0003596918,0.001023764,0.001877267],"genre_scores_gemma":[0.3344666,0.0003800414,0.661497,0.0002220532,0.0001177579,0.0003749224,0.0008663606,0.0001632658,0.001912044],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.0105476,"threshold_uncertainty_score":0.05578172,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3047361733","doi":"10.1002/nem.2130","title":"De‐anonymizing Ethereum blockchain smart contracts through code attribution","year":2020,"lang":"en","type":"article","venue":"International Journal of Network Management","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Saskatchewan; University of New Brunswick","funders":"","keywords":"Computer science; Blockchain; Anonymity; Database transaction; Stylometry; Computer security; Code (set theory); Field (mathematics); Source code; Smart contract; World Wide Web; Artificial intelligence; Database; Programming language","authors":[{"name":"Shlomi Linoy","is_ca":true},{"name":"Natalia Stakhanova","is_ca":true},{"name":"Suprio Ray","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04856664035440313,"gpt":0.3014062167132896,"spread":0.2528395763588865,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003243129,0.0005160735,0.0007723207,0.002739075,0.001084644,0.002257723,0.0009116873,0.001055073,0.001612373],"category_scores_gemma":[0.02270402,0.0003233002,0.0005578211,0.002772504,0.001462873,0.003132315,0.003629806,0.001200933,0.001069499],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009978665,"about_ca_system_score_gemma":0.001469601,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001262026,"about_ca_topic_score_gemma":0.001551404,"domain_scores_codex":[0.9941131,0.001709796,0.0004226608,0.0007839576,0.002539034,0.0004313712],"domain_scores_gemma":[0.9829777,0.004779156,0.002833428,0.007000026,0.002104693,0.0003050295],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0007608674,0.0002189469,0.02420837,0.0003403317,0.0001257496,0.001191579,0.002175809,0.1985655,0.0347504,0.1089764,0.009647176,0.6190389],"study_design_scores_gemma":[0.00003172062,0.00005517664,0.003117546,0.00008631148,0.00002425882,0.0004352804,0.0005614556,0.8675929,0.03015527,0.08530542,0.01258159,0.0000530791],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2359297,0.0003617125,0.7507038,0.0009853269,0.000167131,0.0003086812,0.0009790214,0.003182249,0.007382483],"genre_scores_gemma":[0.8180522,0.0002212833,0.1745967,0.000149529,0.0000646388,0.0001962517,0.001446033,0.0002050789,0.005068161],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003243129,"threshold_uncertainty_score":0.01715153,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3179438688","doi":"10.1109/access.2021.3095967","title":"Authorship Classification in a Resource Constraint Language Using Convolutional Neural Networks","year":2021,"lang":"en","type":"article","venue":"IEEE Access","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Athabasca University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Word2vec; Artificial intelligence; Convolutional neural network; Embedding; Classifier (UML); Hyperparameter; Natural language processing; Pattern recognition (psychology); Document classification; Machine learning","authors":[{"name":"Md. Rajib Hossain","is_ca":false},{"name":"Mohammed Moshiul Hoque","is_ca":false},{"name":"M. Ali Akber Dewan","is_ca":true},{"name":"Nazmul Siddique","is_ca":false},{"name":"Nazmul Islam","is_ca":false},{"name":"Iqbal H. Sarker","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1227674961873143,"gpt":0.3669254085632319,"spread":0.2441579123759176,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005875365,0.0007675612,0.0003819389,0.001297009,0.000425336,0.001061189,0.0007357853,0.0005681217,0.001441185],"category_scores_gemma":[0.001898728,0.0001725637,0.0004960137,0.001113769,0.0003786301,0.002076214,0.000863201,0.0008172106,0.0009714915],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001121081,"about_ca_system_score_gemma":0.0005582445,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009220676,"about_ca_topic_score_gemma":0.01558007,"domain_scores_codex":[0.9994924,0.000118006,0.00004241086,0.0001522196,0.00009923764,0.00009568907],"domain_scores_gemma":[0.9992009,0.0003103054,0.0001137059,0.0001445462,0.0001848791,0.00004569965],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001023334,0.0005472793,0.03103863,0.0003669886,0.0001678587,0.0008824022,0.0006740295,0.07825371,0.04517948,0.003657735,0.01657393,0.8216347],"study_design_scores_gemma":[0.00002191554,0.0001202024,0.01161794,0.00004236067,0.00003987956,0.0001797442,0.0004030312,0.9547108,0.02302648,0.003361163,0.006445426,0.00003102123],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9087998,0.001175954,0.07497898,0.0006049458,0.0002149163,0.0001376368,0.002384119,0.003365523,0.008338202],"genre_scores_gemma":[0.9599831,0.0003050371,0.02738873,0.0001017836,0.00005275142,0.00006174388,0.004755624,0.00006867378,0.007282419],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.009220676,"threshold_uncertainty_score":0.01833403,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2344088191","doi":"10.5220/0005710103540361","title":"Authorship Attribution using Variable Length Part-of-Speech Patterns","year":2016,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":24,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Moncton","funders":"","keywords":"Variable (mathematics); Computer science; Speech recognition; Artificial intelligence; Natural language processing; Mathematics","authors":[{"name":"Yao Jean Marc Pokou","is_ca":true},{"name":"Philippe Fournier‐Viger","is_ca":true},{"name":"Chadia Moghrabi","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07725854955679153,"gpt":0.2956823975046892,"spread":0.2184238479478977,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003319516,0.0006352061,0.0007077968,0.009184351,0.0008797311,0.002256095,0.0008561144,0.001579015,0.001717862],"category_scores_gemma":[0.02319427,0.0002719738,0.0008160951,0.007049588,0.0005053426,0.003089651,0.001162892,0.001014382,0.001938003],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004647368,"about_ca_system_score_gemma":0.0005635357,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008046052,"about_ca_topic_score_gemma":0.001380566,"domain_scores_codex":[0.9961495,0.001307015,0.0004998753,0.001011864,0.0006848991,0.0003468833],"domain_scores_gemma":[0.9673299,0.02000481,0.005267708,0.003228388,0.003163902,0.001005388],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.002837919,0.0008580649,0.3699991,0.0006720756,0.0006477298,0.001931606,0.001883529,0.02460641,0.05436761,0.009978332,0.01311933,0.5190985],"study_design_scores_gemma":[0.00009331037,0.0003788351,0.1446278,0.0001711151,0.0003794904,0.002606156,0.001344906,0.756065,0.03747227,0.04464253,0.01200962,0.0002088757],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8350188,0.002033394,0.1455148,0.000980107,0.0005620297,0.0001422196,0.008358384,0.001816281,0.005574067],"genre_scores_gemma":[0.9638288,0.0002751398,0.03027778,0.00004497527,0.0002094314,0.00004425203,0.003986597,0.00008395541,0.001249064],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.009184351,"threshold_uncertainty_score":0.01755548,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2901907079","doi":"10.1145/3236391","title":"Arabic Authorship Attribution","year":2018,"lang":"en","type":"article","venue":"ACM Transactions on Asian and Low-Resource Language Information Processing","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Cybercrime; Computer science; Anonymity; Authorship attribution; The Internet; Identity (music); Cyberspace; Online identity; Stylometry; Artificial intelligence; Computer security; World Wide Web","authors":[{"name":"Malik H. Altakrori","is_ca":true},{"name":"Farkhund Iqbal","is_ca":false},{"name":"Benjamin C. M. Fung","is_ca":true},{"name":"Steven H. H. Ding","is_ca":true},{"name":"Abdallah Tubaishat","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01375791771308319,"gpt":0.2626968309772169,"spread":0.2489389132641337,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002904224,0.001307234,0.0005584363,0.009752351,0.002260045,0.004161596,0.0007413141,0.001057772,0.08837211],"category_scores_gemma":[0.03364207,0.0002647929,0.0003835658,0.006656724,0.0007192533,0.002567737,0.002469351,0.0007860347,0.05125925],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008603069,"about_ca_system_score_gemma":0.001218168,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007378662,"about_ca_topic_score_gemma":0.0008996131,"domain_scores_codex":[0.9953679,0.0009856735,0.0007085862,0.0007838957,0.001752644,0.0004013004],"domain_scores_gemma":[0.9744319,0.00630853,0.004470806,0.006139753,0.007569179,0.001079752],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0008544648,0.0001170049,0.0213996,0.001312159,0.00005910128,0.0006413808,0.002180952,0.00212647,0.006325555,0.0136142,0.1974024,0.7539667],"study_design_scores_gemma":[0.0000677564,0.0001457778,0.02698934,0.0006349151,0.00005826314,0.002201886,0.002282653,0.02256819,0.02344766,0.01936696,0.9020669,0.0001697632],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.2012637,0.005493093,0.08116202,0.005533899,0.0115134,0.001926396,0.1067138,0.03727564,0.549118],"genre_scores_gemma":[0.6812867,0.002170104,0.09899853,0.0007082081,0.001732058,0.0009151732,0.04878244,0.002649119,0.1627576],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.08837211,"threshold_uncertainty_score":0.2956342,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3095732274","doi":"10.1007/978-3-030-60290-1_17","title":"DeepStyle: User Style Embedding for Authorship Attribution of Short Texts","year":2020,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Computer science; Task (project management); Salient; Authorship attribution; Embedding; Word embedding; Style (visual arts); Word (group theory); Artificial intelligence; Natural language processing; Attribution; Writing style; Information retrieval; Linguistics","authors":[{"name":"Zhiqiang Hu","is_ca":false},{"name":"Roy Ka-Wei Lee","is_ca":true},{"name":"Lei Wang","is_ca":false},{"name":"Ee‐Peng Lim","is_ca":false},{"name":"Bo Dai","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04678901294171871,"gpt":0.3068086415893817,"spread":0.260019628647663,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001388799,0.001506889,0.0007217505,0.003074692,0.0005486397,0.001733101,0.000848335,0.00105466,0.009637575],"category_scores_gemma":[0.006243602,0.0005157617,0.0007666668,0.002416185,0.0003779026,0.003836175,0.002337802,0.001535762,0.007651487],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003994533,"about_ca_system_score_gemma":0.0004771504,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007881228,"about_ca_topic_score_gemma":0.002080721,"domain_scores_codex":[0.998642,0.0003497593,0.0001123103,0.0004154459,0.0003756489,0.0001048971],"domain_scores_gemma":[0.9968374,0.001227458,0.0002460013,0.001004461,0.0004789256,0.0002058315],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000466877,0.0002795387,0.004217875,0.0003169589,0.00009452069,0.0001425567,0.0004927425,0.007773196,0.01509986,0.004636804,0.05147928,0.9149998],"study_design_scores_gemma":[0.00007953386,0.0003161242,0.007822172,0.0001680738,0.000106475,0.0007900284,0.0003259207,0.8435799,0.03785366,0.05185326,0.05696937,0.0001355475],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06126724,0.002105491,0.8813132,0.0004843925,0.001370708,0.0002039577,0.00964102,0.03676166,0.006852424],"genre_scores_gemma":[0.467827,0.001669328,0.4632616,0.0003033133,0.001107377,0.000437387,0.02462968,0.004532073,0.0362323],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009637575,"threshold_uncertainty_score":0.03224087,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2774157693","doi":"10.1109/icdmw.2017.51","title":"Estimating Personality from Social Media Posts","year":2017,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Queen's University","funders":"","keywords":"Personality psychology; Personality; Lexicon; Computer science; Repertoire; Social media; Artificial intelligence; Negotiation; Arabic; Robot; Natural language processing; Social psychology; Linguistics; Psychology; World Wide Web; Sociology","authors":[{"name":"Nasser A Alsadhan","is_ca":true},{"name":"David B. Skillicorn","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09161803524495755,"gpt":0.3393627365054629,"spread":0.2477447012605054,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001070731,0.0006891934,0.0002894847,0.002986026,0.0004263927,0.0007496897,0.0002639284,0.0004662292,0.001637308],"category_scores_gemma":[0.007327182,0.000174201,0.0002622801,0.001214615,0.0001788012,0.0010269,0.0005228058,0.0004145055,0.001817387],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002566986,"about_ca_system_score_gemma":0.0001947513,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001786931,"about_ca_topic_score_gemma":0.003761136,"domain_scores_codex":[0.9992519,0.000272724,0.00007596702,0.0001493316,0.0001658763,0.00008411446],"domain_scores_gemma":[0.9941901,0.002948533,0.0008929527,0.0005256622,0.00112428,0.000318581],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.001028767,0.00043394,0.5834768,0.0003026744,0.0001914121,0.0006045181,0.001901571,0.006903175,0.019701,0.0009122143,0.01079121,0.3737527],"study_design_scores_gemma":[0.00002963578,0.000481381,0.6597559,0.00008574325,0.0001328751,0.000886836,0.003201389,0.3005736,0.02003396,0.003455998,0.01125679,0.0001058692],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9635307,0.0002524851,0.02511663,0.0002981414,0.0001113973,0.0001449535,0.004259819,0.0007957144,0.005490204],"genre_scores_gemma":[0.9793553,0.0001233529,0.01535929,0.00003105219,0.00009918297,0.00005708759,0.00245545,0.00002184003,0.00249739],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.002986026,"threshold_uncertainty_score":0.00566268,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4213077949","doi":"10.1037/cep0000273","title":"Bridging people and perspectives: General and language-specific social network structure predict mentalizing across diverse sociolinguistic contexts.","year":2022,"lang":"en","type":"article","venue":"Canadian Journal of Experimental Psychology/Revue canadienne de psychologie expérimentale","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Betweenness centrality; Mentalization; Psychology; Cognitive psychology; PsycINFO; Social cognition; Cognition; Bridging (networking); Context (archaeology); Developmental psychology; Computer science; Neuroscience; Biology; MEDLINE","authors":[{"name":"Mehrgol Tiv","is_ca":false},{"name":"Ethan Kutlu","is_ca":false},{"name":"Elisabeth O’Regan","is_ca":false},{"name":"Debra Titone","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03936839491776121,"gpt":0.320760304366242,"spread":0.2813919094484807,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00053278,0.0002354656,0.000161084,0.0006582785,0.0005345614,0.00113283,0.000221552,0.0004248953,0.003937795],"category_scores_gemma":[0.004670362,0.00015839,0.0001972544,0.0005965821,0.0008590443,0.001062705,0.001040399,0.00046846,0.0002560286],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005371775,"about_ca_system_score_gemma":0.0003002114,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01339983,"about_ca_topic_score_gemma":0.03007301,"domain_scores_codex":[0.9998016,0.00005424474,0.00001439096,0.00005418136,0.0000401468,0.00003544832],"domain_scores_gemma":[0.9974065,0.0008116686,0.001089322,0.0002084278,0.00009429233,0.0003897388],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0001071462,0.0001172002,0.9843376,0.00002122062,0.00003521414,0.00004498612,0.008295462,0.00008059275,0.000887474,0.0002212502,0.0002080854,0.005643797],"study_design_scores_gemma":[0.000003597717,0.0000380984,0.9954543,0.000008869078,0.000009928416,0.000026711,0.003670995,0.000276408,0.0001273571,0.0002500992,0.0001289843,0.000004675838],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9992717,0.00002261275,0.0000718074,0.00005123403,0.000001394749,0.000003108021,0.00005738913,0.000001473868,0.0005192965],"genre_scores_gemma":[0.9995787,0.0000333084,0.0001095852,0.00001318771,0.00000179116,0.000005855422,0.0001011609,0.00000115423,0.0001552381],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01339983,"threshold_uncertainty_score":0.02664369,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W112115809","doi":"","title":"Herbert west: deanonymizer","year":2011,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Anonymity; Computer science; Selection (genetic algorithm); Face (sociological concept); Quality (philosophy); Simple (philosophy); Internet privacy; Data science; World Wide Web; Computer security; Artificial intelligence; Sociology; Epistemology","authors":[{"name":"Mihir Nanavati","is_ca":true},{"name":"Nathan Taylor","is_ca":true},{"name":"William Aiello","is_ca":true},{"name":"Andrew Warfield","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07896113860058249,"gpt":0.254769367506567,"spread":0.1758082289059845,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006817743,0.0009491135,0.00102211,0.005880988,0.003271877,0.00464548,0.001532552,0.002317049,0.1626659],"category_scores_gemma":[0.04305355,0.0008501829,0.0003298302,0.006106521,0.001599831,0.0110581,0.003674931,0.002559603,0.1487633],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001638551,"about_ca_system_score_gemma":0.003533686,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002636239,"about_ca_topic_score_gemma":0.003258523,"domain_scores_codex":[0.9914266,0.002696888,0.0007899555,0.001687472,0.002909695,0.0004894728],"domain_scores_gemma":[0.9764506,0.0066248,0.002107057,0.004472354,0.007711858,0.002633294],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00007275955,0.00001494594,0.0008150142,0.0001237668,0.000005552446,0.00009051785,0.0005023334,0.0001403793,0.0005929487,0.02963784,0.807933,0.1600709],"study_design_scores_gemma":[0.00001283999,0.000008280396,0.0002968531,0.0000633534,0.000003575334,0.0001632349,0.0001851072,0.0006169321,0.0008433891,0.008000125,0.9897801,0.00002615068],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"software","genre_scores_codex":[0.008117647,0.01754248,0.2765057,0.1371031,0.03993601,0.0007770221,0.0085636,0.05279019,0.4586642],"genre_scores_gemma":[0.09404498,0.01228284,0.08957411,0.01141829,0.008988642,0.0006347202,0.005638757,0.01156182,0.7658558],"genre_candidate":"software","genre_consensus":null,"teacher_disagreement_score":0.1626659,"threshold_uncertainty_score":0.5441716,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2042368861","doi":"10.1109/vast.2012.6400484","title":"Relative N-gram signatures: Document visualization at the level of character N-grams","year":2012,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"Boeing","keywords":"Visualization; n-gram; Computer science; Classifier (UML); Character (mathematics); Artificial intelligence; Gram; Data visualization; Natural language processing; Pattern recognition (psychology); Information retrieval; Mathematics; Language model","authors":[{"name":"Magdalena Jankowska","is_ca":true},{"name":"Vlado Kešelj","is_ca":true},{"name":"Evangelos Milios","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09470629678166564,"gpt":0.341107670361583,"spread":0.2464013735799173,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001184376,0.00107917,0.0004794192,0.004954836,0.0006314787,0.002785339,0.0007233349,0.0009550264,0.008730805],"category_scores_gemma":[0.008001673,0.0002542543,0.0003441672,0.003471433,0.0004458349,0.003387381,0.001719779,0.001070086,0.002302919],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004168788,"about_ca_system_score_gemma":0.0005524513,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001372254,"about_ca_topic_score_gemma":0.001209905,"domain_scores_codex":[0.9994125,0.0001719331,0.00006031923,0.00008736588,0.000215019,0.00005281028],"domain_scores_gemma":[0.9962798,0.001749528,0.0004629134,0.0004458007,0.0008698548,0.0001921035],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00158274,0.0001843237,0.008642868,0.001135732,0.0001073524,0.0008830819,0.005632313,0.008884261,0.1091868,0.05044056,0.0678303,0.7454897],"study_design_scores_gemma":[0.000201089,0.000396651,0.01869401,0.0006563774,0.0001619393,0.002410149,0.004243915,0.4465709,0.1472481,0.1487254,0.2303027,0.0003888477],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06897633,0.001175898,0.8586186,0.001632552,0.0003855988,0.0002668676,0.00781028,0.04861436,0.01251949],"genre_scores_gemma":[0.3045543,0.000845985,0.6834388,0.0002389381,0.0001871072,0.0002785199,0.003341315,0.002779946,0.004335147],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008730805,"threshold_uncertainty_score":0.02920741,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W93917463","doi":"","title":"Authorship Attribution for Small Texts: Literary and Forensic Experiments.","year":2007,"lang":"en","type":"article","venue":"PAN","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Attribution; Authorship attribution; Psychology; History; Computer science; Linguistics; Social psychology; Natural language processing; Philosophy","authors":[{"name":"Olga Feiguina","is_ca":false},{"name":"Graeme Hirst","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06904991508512409,"gpt":0.319997208466426,"spread":0.2509472933813019,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0103413,0.0005769152,0.0005483904,0.002335891,0.002186236,0.00212587,0.001122117,0.001690658,0.004647957],"category_scores_gemma":[0.1031242,0.0003101622,0.000289264,0.002388262,0.001782792,0.003232916,0.002356799,0.001537149,0.001260132],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005317935,"about_ca_system_score_gemma":0.000496336,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003025742,"about_ca_topic_score_gemma":0.0003470231,"domain_scores_codex":[0.9912937,0.005172874,0.0005245213,0.001132817,0.001587919,0.0002881668],"domain_scores_gemma":[0.8066755,0.1530006,0.01006963,0.02185409,0.006282801,0.00211736],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.02223603,0.02439225,0.07759953,0.002680194,0.0006260879,0.002981199,0.02838924,0.02628573,0.1297161,0.05639623,0.03560634,0.5930911],"study_design_scores_gemma":[0.001978973,0.008831976,0.09181485,0.0006410611,0.0004977196,0.00696717,0.01517916,0.4837612,0.2376699,0.117805,0.03443926,0.0004137384],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9789444,0.0003354757,0.01306144,0.0004634887,0.0002165756,0.0002996818,0.0007889948,0.0003122915,0.005577595],"genre_scores_gemma":[0.9829966,0.0001363967,0.01269673,0.00008626075,0.0001190277,0.0001738866,0.0008666987,0.00007314964,0.002851297],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.0103413,"threshold_uncertainty_score":0.05469066,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2119140562","doi":"10.1007/s10032-007-0060-2","title":"Genre as noise: noise in genre","year":2007,"lang":"en","type":"article","venue":"International Journal on Document Analysis and Recognition (IJDAR)","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Noise (video); Perspective (graphical); Hierarchy; Feature (linguistics); Word error rate; Natural language processing; Artificial intelligence; Emphasis (telecommunications); Speech recognition; Selection (genetic algorithm); Pattern recognition (psychology); Linguistics","authors":[{"name":"Andrea Stubbe","is_ca":false},{"name":"Christoph Ringlstetter","is_ca":true},{"name":"Klaus U. Schulz","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02048356387265479,"gpt":0.317666999386885,"spread":0.2971834355142302,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003772506,0.0005028419,0.0008792196,0.003861069,0.001357845,0.007157531,0.0006359284,0.0009855335,0.0047258],"category_scores_gemma":[0.03588212,0.0003335887,0.0005015738,0.004091608,0.002161922,0.005945658,0.002629093,0.001862126,0.0008613692],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007977855,"about_ca_system_score_gemma":0.0007061757,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001149518,"about_ca_topic_score_gemma":0.001056893,"domain_scores_codex":[0.9961378,0.001363727,0.0003252073,0.0007651328,0.001131294,0.0002767358],"domain_scores_gemma":[0.9653045,0.02417602,0.002869202,0.003167474,0.002823853,0.001658897],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.002365535,0.0005775017,0.2201733,0.0006423327,0.0002363584,0.0007168999,0.008353383,0.008914478,0.02232106,0.0951684,0.01165759,0.6288732],"study_design_scores_gemma":[0.0001165848,0.0006824649,0.2776867,0.0006987177,0.0006184805,0.003620225,0.01255278,0.2893348,0.01724143,0.3528464,0.04426995,0.0003314269],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7148488,0.006456466,0.2371968,0.004281227,0.001351072,0.0001647325,0.001835523,0.0008718336,0.03299347],"genre_scores_gemma":[0.9830838,0.0005583148,0.01240457,0.0001783241,0.0005013599,0.00004399843,0.0004236398,0.0001356076,0.002670407],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.007157531,"threshold_uncertainty_score":0.01995116,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2477122767","doi":"10.1109/bracis.2016.071","title":"Authorship Attribution via Network Motifs Identification","year":2016,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Automatic summarization; Authorship attribution; Identification (biology); Attribution; Function (biology); Style (visual arts); Motif (music)","authors":[],"retraction":null,"screen_n_in":null,"score":{"opus":0.02819452153527435,"gpt":0.2619039526048588,"spread":0.2337094310695844,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001846663,0.0005661376,0.000630223,0.007977158,0.0006940058,0.00181604,0.0006608413,0.0008823797,0.002260241],"category_scores_gemma":[0.01841636,0.0002824889,0.0004584639,0.00398671,0.0005774312,0.002886261,0.001257942,0.000731305,0.0007203665],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007755033,"about_ca_system_score_gemma":0.0005347604,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001267865,"about_ca_topic_score_gemma":0.00167864,"domain_scores_codex":[0.998171,0.0006915581,0.0001406433,0.0004416746,0.0003956232,0.0001595412],"domain_scores_gemma":[0.9890191,0.006582763,0.002150041,0.0008162138,0.001162688,0.0002691018],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0006894122,0.0002389841,0.1316649,0.0005700047,0.000323743,0.0006373011,0.001617973,0.07704359,0.01909142,0.03884124,0.006124906,0.7231566],"study_design_scores_gemma":[0.00002638003,0.00009081372,0.02880382,0.0001151221,0.00009798229,0.0007446995,0.000580662,0.8642119,0.01180277,0.08654211,0.00691878,0.0000648827],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.4308226,0.001168266,0.5549173,0.0008363301,0.00011754,0.0002169833,0.002077565,0.001773908,0.008069433],"genre_scores_gemma":[0.9107173,0.0003180101,0.08585744,0.00003339425,0.00007096989,0.0001134286,0.001020014,0.00006152642,0.001807966],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.007977158,"threshold_uncertainty_score":0.009766161,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W140340549","doi":"10.63317/2id75yrzanby","title":"Using the Complexity of the Distribution of Lexical Elements as a Feature in Authorship Attribution","year":2008,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Punctuation; Authorship attribution; Computer science; Artificial intelligence; Natural language processing; Feature (linguistics); Support vector machine; Noun; Linguistics","authors":[{"name":"Leanne Spracklin","is_ca":true},{"name":"Diana Inkpen","is_ca":true},{"name":"Amiya Nayak","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.2062589589989079,"gpt":0.3485729465110212,"spread":0.1423139875121133,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003859967,0.0008225345,0.0008928261,0.005572944,0.0006123842,0.002213079,0.000797908,0.001250066,0.001196789],"category_scores_gemma":[0.03419756,0.0004218061,0.0007751461,0.003594026,0.001066591,0.00518641,0.00134692,0.001181695,0.0007398419],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006531607,"about_ca_system_score_gemma":0.0004236531,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005627148,"about_ca_topic_score_gemma":0.0006002139,"domain_scores_codex":[0.9970521,0.0009468131,0.0003719454,0.0005557194,0.0009085485,0.0001649826],"domain_scores_gemma":[0.9538644,0.03501554,0.004908878,0.00381051,0.001883395,0.000517314],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001162271,0.0006210837,0.2231395,0.0005398668,0.000393991,0.0007752111,0.001181705,0.07765642,0.06562121,0.01102933,0.002358966,0.6155204],"study_design_scores_gemma":[0.00003980326,0.0004823197,0.1086611,0.00006656859,0.0001338281,0.001627987,0.0003200019,0.8027042,0.05200565,0.03160568,0.002101222,0.0002515849],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.53397,0.0003722316,0.4618914,0.000258886,0.00008609093,0.00008823915,0.0005997791,0.001090372,0.001642988],"genre_scores_gemma":[0.9539549,0.0001300282,0.04479859,0.00001597687,0.00006542081,0.00004926448,0.0003902194,0.00006203394,0.0005335411],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.005572944,"threshold_uncertainty_score":0.0204137,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4385572818","doi":"10.18653/v1/2022.emnlp-main.128","title":"AfroLID: A Neural Language Identification Tool for African Languages","year":2022,"lang":"en","type":"article","venue":"","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"Alliance de recherche numérique du Canada; Social Sciences and Humanities Research Council of Canada; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; Advanced Micro Devices","keywords":"Computer science; Domain (mathematical analysis); Identification (biology); Natural language processing; Set (abstract data type); Cover (algebra); Artificial intelligence; Exploit; Language identification; Natural language; Programming language; Engineering","authors":[{"name":"Ife Adebara","is_ca":true},{"name":"AbdelRahim Elmadany","is_ca":true},{"name":"Muhammad Abdul-Mageed","is_ca":true},{"name":"Alcides Alcoba Inciarte","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02260801631974186,"gpt":0.3036981982625557,"spread":0.2810901819428138,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00180814,0.00121622,0.0005742416,0.003986715,0.0009952963,0.001554157,0.001426511,0.001060866,0.01433393],"category_scores_gemma":[0.007219915,0.0004062756,0.0008179242,0.001541238,0.0004746787,0.003102521,0.003583293,0.001461606,0.01044192],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000674888,"about_ca_system_score_gemma":0.001389164,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005347307,"about_ca_topic_score_gemma":0.01222241,"domain_scores_codex":[0.9989303,0.0002411444,0.0001267943,0.0003188246,0.0002718379,0.0001110665],"domain_scores_gemma":[0.9978291,0.001052127,0.0001671977,0.0004151531,0.0004280297,0.0001083577],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004163014,0.0002421466,0.01384628,0.001058632,0.0001454574,0.0005007813,0.001199797,0.006982852,0.01833751,0.004469713,0.140123,0.8126776],"study_design_scores_gemma":[0.0002444244,0.0004095107,0.02064799,0.000577952,0.0001582605,0.002955642,0.003184123,0.5470566,0.0947416,0.02834777,0.3013426,0.0003335278],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1408201,0.002380594,0.503893,0.001467874,0.001040774,0.0008119686,0.0479851,0.2693668,0.03223384],"genre_scores_gemma":[0.3573421,0.0006389169,0.5509309,0.0007500783,0.0001168495,0.001005248,0.06057357,0.003678045,0.02496437],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01433393,"threshold_uncertainty_score":0.04795176,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1970306474","doi":"10.1353/cjp.2011.0028","title":"Description, Disagreement, and Fictional Names","year":2011,"lang":"en","type":"article","venue":"Canadian Journal of Philosophy","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Lethbridge","funders":"","keywords":"Fictional universe; Philosophy; Proper noun; Epistemology; Emptiness; Linguistics; Narrative","authors":[{"name":"Peter Alward","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08777744241595391,"gpt":0.2282456605309692,"spread":0.1404682181150153,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01090873,0.0004138109,0.0006518441,0.003768216,0.00745155,0.008155659,0.00211757,0.003916963,0.006253047],"category_scores_gemma":[0.03292422,0.0004406544,0.0004372675,0.00480941,0.03088658,0.02634582,0.007716376,0.003635401,0.0005664536],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.008160269,"about_ca_system_score_gemma":0.00265565,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00286268,"about_ca_topic_score_gemma":0.001656565,"domain_scores_codex":[0.985657,0.007801922,0.0007410701,0.001738257,0.003067757,0.0009940114],"domain_scores_gemma":[0.9767449,0.01438329,0.002335244,0.003130347,0.002637933,0.0007682601],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000008877507,0.000003066525,0.0003245466,0.00001812912,0.000001869912,0.00002582449,0.004645198,0.0001084482,0.00002616633,0.9904054,0.000673574,0.003758789],"study_design_scores_gemma":[0.000008883741,0.000005353258,0.0002902962,0.00005843625,0.000005899523,0.00009191473,0.004727292,0.0009179414,0.0002118445,0.9638602,0.02980708,0.00001488042],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.1661493,0.007612613,0.2400296,0.04706286,0.0006731388,0.0001396422,0.0005113138,0.0003377761,0.5374838],"genre_scores_gemma":[0.9826484,0.0005877924,0.007512926,0.0008932973,0.0001386311,0.00005592791,0.0001701268,0.00005053341,0.007942374],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01090873,"threshold_uncertainty_score":0.05920714,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3110983064","doi":"10.1007/978-3-030-61675-5","title":"Machine Learning for Authorship Attribution and Cyber Forensics","year":2020,"lang":"en","type":"book","venue":"International series on computer entertainment and media technology","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University; Concordia University","funders":"","keywords":"Authorship attribution; Attribution; Computer forensics; Computer science; Computer security; Data science; Psychology; Internet privacy; Natural language processing; Digital forensics; Social psychology","authors":[{"name":"Farkhund Iqbal","is_ca":false},{"name":"Mourad Debbabi","is_ca":true},{"name":"Benjamin C. M. Fung","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02230978867466325,"gpt":0.2477327119029573,"spread":0.225422923228294,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002116902,0.0007788144,0.000954896,0.002893984,0.000637728,0.002522431,0.001244037,0.001414524,0.007510176],"category_scores_gemma":[0.00967066,0.0004882682,0.0005501629,0.003299164,0.00126344,0.004948406,0.001111731,0.002511788,0.005664438],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007820235,"about_ca_system_score_gemma":0.0005704628,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005852744,"about_ca_topic_score_gemma":0.0007872712,"domain_scores_codex":[0.9985757,0.0003807114,0.00008548277,0.000253856,0.0006494196,0.00005496784],"domain_scores_gemma":[0.9918368,0.00611437,0.00028732,0.0008955641,0.0007954396,0.00007039207],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00003546977,0.00006050074,0.00042766,0.0002417753,0.0000280551,0.00005019013,0.0000637918,0.01180256,0.0009689723,0.07574468,0.07198077,0.8385954],"study_design_scores_gemma":[0.000009925047,0.00004824539,0.001138433,0.0002715853,0.00002699318,0.0004846668,0.00006925685,0.3839182,0.005163957,0.5005516,0.1082658,0.00005132473],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"other","genre_scores_codex":[0.006588633,0.05265705,0.8746688,0.005581482,0.003159641,0.00009141079,0.0008411642,0.003041457,0.05337045],"genre_scores_gemma":[0.2557347,0.04276486,0.4822898,0.0009811157,0.006520176,0.0002749895,0.002777819,0.001053526,0.2076031],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.007510176,"threshold_uncertainty_score":0.02512407,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2250996967","doi":"","title":"Cognate and Misspelling Features for Natural Language Identification","year":2013,"lang":"en","type":"article","venue":"Workshop on Innovative Use of NLP for Building Educational Applications","topic":"Authorship Attribution and Profiling","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta","funders":"","keywords":"Bigram; Computer science; Cognate; Natural language processing; Artificial intelligence; Classifier (UML); Natural language; Word (group theory); Identification (biology); Spelling; Syntax; Language identification; Linguistics","authors":[{"name":"Garrett Nicolai","is_ca":true},{"name":"Bradley Hauer","is_ca":true},{"name":"Mohammad Salameh","is_ca":true},{"name":"Lei Yao","is_ca":true},{"name":"Grzegorz Kondrak","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04016925118674335,"gpt":0.347697391551569,"spread":0.3075281403648256,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002527087,0.001327926,0.0008709689,0.003890466,0.001020371,0.001381296,0.000874899,0.0009793027,0.002749864],"category_scores_gemma":[0.0109757,0.00023774,0.0007656406,0.00193756,0.0004682339,0.002097928,0.001537054,0.00150325,0.002163836],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003309601,"about_ca_system_score_gemma":0.0008284198,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00158971,"about_ca_topic_score_gemma":0.002710665,"domain_scores_codex":[0.9971951,0.0009006651,0.0003001769,0.0005909849,0.0007678008,0.0002452282],"domain_scores_gemma":[0.9895887,0.005806613,0.0008951103,0.001398646,0.001700785,0.0006100758],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000848415,0.0008313481,0.04739359,0.0003870119,0.000159598,0.0004180543,0.0005387546,0.008026438,0.04497234,0.001149729,0.007068025,0.8882068],"study_design_scores_gemma":[0.0001519073,0.001221826,0.06688757,0.0001859972,0.0003158285,0.001696663,0.001313319,0.7375987,0.1559478,0.01834087,0.01605993,0.0002794624],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7547152,0.001557654,0.2256594,0.0006205891,0.0002668394,0.0002527125,0.003101665,0.007741486,0.006084489],"genre_scores_gemma":[0.8842067,0.0001697323,0.1094606,0.0000904362,0.00009078635,0.0001442146,0.003975436,0.0002351529,0.00162718],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003890466,"threshold_uncertainty_score":0.01336467,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}