{"meta":{"query_hash":"47001aa2d831","filters":{"venue":"Advances in Data Analysis and Classification"},"cohort_total":40,"direct_labels_cover":0,"predictions_cover":40,"exported":40,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/47001aa2d831","api":"https://metacan.xera.ac/api/v1/cohort?venue=Advances+in+Data+Analysis+and+Classification"},"results":[{"id":"W1875624538","doi":"10.1007/s11634-015-0219-5","title":"Factor probabilistic distance clustering (FPDC): a new clustering method","year":2015,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Università degli Studi di Napoli Federico II","keywords":"Cluster analysis; Correlation clustering; CURE data clustering algorithm; Single-linkage clustering; Probabilistic logic; Canopy clustering algorithm; Computer science; Data stream clustering; Fuzzy clustering; Determining the number of clusters in a data set; Clustering high-dimensional data; k-medians clustering; Transformation (genetics); Mathematics; Pattern recognition (psychology); Data mining; Artificial intelligence","score_opus":0.12235253318764872,"score_gpt":0.4083881885969034,"score_spread":0.2860356554092547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1875624538","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011951176,0.0007192656,0.9967601,0.00012680679,0.0001219096,0.000054091328,0.00011744802,0.00035847185,0.0005468645],"genre_scores_gemma":[0.03915475,0.0012054665,0.9558839,0.00018155322,0.000273096,0.0002045026,0.00063115614,0.0002979931,0.0021675325],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9926645,0.0017187134,0.00042034817,0.001765367,0.0031921363,0.00023891119],"domain_scores_gemma":[0.9955818,0.0010924345,0.00035465005,0.0010324306,0.0017795999,0.0001590322],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0037781529,0.0016513546,0.002357854,0.0045523457,0.0022730345,0.0026633928,0.0037815142,0.0026331637,0.0022695842],"category_scores_gemma":[0.010060121,0.0008283834,0.0022515785,0.007889587,0.0019187598,0.0033432723,0.0034210489,0.003131452,0.0020724712],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027776673,0.00011612515,0.002149991,0.00069898576,0.0003766307,0.00009979348,0.00040757566,0.08116129,0.008370257,0.06646979,0.019165006,0.8207068],"study_design_scores_gemma":[0.0000643426,0.00017484362,0.0019198905,0.00012762296,0.00020361092,0.0009411274,0.0001695635,0.82802284,0.009874055,0.09945981,0.058808766,0.00023353526],"about_ca_topic_score_codex":0.00695526,"about_ca_topic_score_gemma":0.0063181594,"teacher_disagreement_score":0.00695526,"about_ca_system_score_codex":0.0015881434,"about_ca_system_score_gemma":0.0033170464,"threshold_uncertainty_score":0.019981086},"labels":[],"label_agreement":null},{"id":"W1964231120","doi":"10.1007/s11634-014-0165-7","title":"Variational Bayes approximations for clustering via mixtures of normal inverse Gaussian distributions","year":2014,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":52,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Bayes' theorem; Gaussian; Univariate; Applied mathematics; Cluster analysis; Mixture model; Inverse Gaussian distribution; Multivariate normal distribution; Mathematics; Inverse; Inverse problem; Multivariate statistics; Mathematical optimization; Computer science; Statistics; Distribution (mathematics); Mathematical analysis; Bayesian probability; Chemistry; Computational chemistry","score_opus":0.028086007090192076,"score_gpt":0.3141116705601102,"score_spread":0.2860256634699181,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1964231120","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011976219,0.00035360505,0.99787617,0.00010365468,0.000021618236,0.000019511235,0.000041550433,0.000098963465,0.00028730292],"genre_scores_gemma":[0.115462646,0.0018860265,0.8745106,0.0002697157,0.00032522323,0.00046962203,0.0011677817,0.00058250065,0.0053258557],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99436414,0.003066254,0.0002788938,0.00091921765,0.0010896353,0.00028187037],"domain_scores_gemma":[0.9847596,0.011428095,0.00064974633,0.0013468765,0.0014890207,0.00032656113],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010721231,0.0019015643,0.00403691,0.0036788688,0.0015677185,0.004480257,0.007006572,0.003645211,0.003283729],"category_scores_gemma":[0.03816178,0.0028174103,0.0038338895,0.003978901,0.0039866567,0.0067672026,0.004178036,0.0071677957,0.0016083001],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001424107,0.00008258574,0.00094684266,0.00028250398,0.00021169613,0.00004816417,0.0004719657,0.58512616,0.0008111953,0.3268229,0.0044702822,0.080583304],"study_design_scores_gemma":[0.00001064568,0.000007232966,0.0001074029,0.00002609196,0.000015113071,0.000019389418,0.00001886896,0.8813138,0.00013884036,0.1174018,0.00092404755,0.00001681808],"about_ca_topic_score_codex":0.018911542,"about_ca_topic_score_gemma":0.020584801,"teacher_disagreement_score":0.018911542,"about_ca_system_score_codex":0.0044408934,"about_ca_system_score_gemma":0.0031705392,"threshold_uncertainty_score":0.05669999},"labels":[],"label_agreement":null},{"id":"W1972256887","doi":"10.1007/s11634-015-0204-z","title":"A mixture of generalized hyperbolic factor analyzers","year":2015,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Bayesian information criterion; Mixture model; Hyperbolic function; Cluster analysis; Gaussian; Generalized inverse Gaussian distribution; Maximization; Expectation–maximization algorithm","score_opus":0.07739344212886445,"score_gpt":0.3564501671026479,"score_spread":0.2790567249737834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1972256887","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0057210117,0.00016275432,0.9924744,0.00012191515,0.000029667961,0.000038451166,0.00007171569,0.00030379792,0.0010762381],"genre_scores_gemma":[0.25194412,0.0008674809,0.7311154,0.00045959663,0.0002257763,0.00025467615,0.00063065,0.0006352412,0.013867133],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9947247,0.0022000265,0.0002725328,0.0013858202,0.0010555174,0.0003614211],"domain_scores_gemma":[0.99291575,0.0033445547,0.00032854438,0.0014350886,0.0016164528,0.00035949756],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0060699484,0.0014296876,0.0018992516,0.002423189,0.0013402629,0.0043251235,0.0034022455,0.0023462577,0.008299369],"category_scores_gemma":[0.01843472,0.0013834586,0.0024897275,0.0030263339,0.00256162,0.006548292,0.004272107,0.0029023532,0.003184758],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008261484,0.00015944021,0.006130991,0.00024287001,0.00044411243,0.00036229123,0.0007172791,0.085107446,0.013998339,0.52737707,0.006837149,0.35779688],"study_design_scores_gemma":[0.00003594207,0.00008120538,0.0009212317,0.000045437544,0.000101757374,0.00038667096,0.000116987496,0.78804606,0.0032405725,0.20099092,0.005937116,0.00009612975],"about_ca_topic_score_codex":0.003893228,"about_ca_topic_score_gemma":0.003312906,"teacher_disagreement_score":0.008299369,"about_ca_system_score_codex":0.0012850227,"about_ca_system_score_gemma":0.0018032737,"threshold_uncertainty_score":0.032101333},"labels":[],"label_agreement":null},{"id":"W1973247608","doi":"10.1007/s11634-013-0137-3","title":"Dimension reduction for model-based clustering via mixtures of multivariate $$t$$ t -distributions","year":2013,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Dimensionality reduction; Cluster analysis; Subspace topology; Dimension (graph theory); Mixture model; Mathematics; Gaussian; Reduction (mathematics); Clustering high-dimensional data; Eigenvalues and eigenvectors; Data set; Curse of dimensionality; Algorithm; Multivariate normal distribution; Pattern recognition (psychology); Computer science; Multivariate statistics; Artificial intelligence; Statistics; Combinatorics; Physics","score_opus":0.03849028777574721,"score_gpt":0.3380154403337114,"score_spread":0.2995251525579642,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1973247608","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002219324,0.00017263347,0.99708813,0.00011790703,0.000019303272,0.000022085474,0.00005632671,0.00019321957,0.00011102303],"genre_scores_gemma":[0.09811257,0.0006339349,0.89744914,0.00023584075,0.00014292367,0.0005050857,0.0012092803,0.00040887293,0.0013023418],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9947396,0.0027204263,0.00035838888,0.00086859404,0.0010917754,0.00022125272],"domain_scores_gemma":[0.9916093,0.005275462,0.00046690734,0.0014605758,0.0009457963,0.0002419402],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005381884,0.0015882878,0.002795683,0.0022469885,0.00162221,0.002821217,0.0032989837,0.0022151985,0.0019434504],"category_scores_gemma":[0.021389509,0.0014862204,0.004931434,0.0027176477,0.0017331723,0.0033838928,0.0047333804,0.005299333,0.001367863],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003872689,0.0002602938,0.002158423,0.00047503487,0.0005553988,0.0001195116,0.00062402117,0.53623056,0.0055293734,0.13384323,0.010341247,0.3094756],"study_design_scores_gemma":[0.000011015948,0.000016072774,0.0002205252,0.000015409849,0.000026389469,0.00004011266,0.000018751058,0.9416967,0.00059906533,0.056536127,0.0007966928,0.00002313319],"about_ca_topic_score_codex":0.0060431524,"about_ca_topic_score_gemma":0.006206737,"teacher_disagreement_score":0.0060431524,"about_ca_system_score_codex":0.0017711484,"about_ca_system_score_gemma":0.0022337958,"threshold_uncertainty_score":0.02846247},"labels":[],"label_agreement":null},{"id":"W1990989764","doi":"10.1007/s11634-012-0108-0","title":"Adaptation of interval PCA to symbolic histogram variables","year":2012,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hôpital Saint-François d'Assise","funders":"","keywords":"Principal component analysis; Histogram; Symbolic data analysis; Mathematics; Projection (relational algebra); Representation (politics); Pattern recognition (psychology); Interval (graph theory); Artificial intelligence; Algorithm; Computer science; Statistics; Combinatorics","score_opus":0.05461420121741699,"score_gpt":0.32804995560749833,"score_spread":0.27343575439008133,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1990989764","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013534733,0.00019596035,0.98303425,0.00005738418,0.00011581271,0.000034080615,0.000108799395,0.0009170456,0.002002024],"genre_scores_gemma":[0.42144266,0.00080925366,0.5707063,0.00010829057,0.00035971153,0.00019736965,0.00089929684,0.0006372371,0.004839857],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9994085,0.0001629049,0.000027716856,0.0001427126,0.00020585668,0.000052230112],"domain_scores_gemma":[0.9989171,0.0002937554,0.0000479238,0.00033337137,0.0003584502,0.00004949838],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006543301,0.00049110566,0.0005849643,0.0008448911,0.00027057383,0.00089582347,0.0010159827,0.00030658575,0.0038784165],"category_scores_gemma":[0.0043693944,0.00022610494,0.0006408197,0.0017222671,0.00044854856,0.0010445461,0.0011893638,0.0013098664,0.0013008825],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021361512,0.00014992882,0.0014141718,0.00011492261,0.0000789679,0.00007266406,0.0000942049,0.07943669,0.019132,0.018688194,0.0039664265,0.8766381],"study_design_scores_gemma":[0.0000147521005,0.00006766618,0.0024634246,0.0000119639,0.000027217518,0.00009473828,0.000026180775,0.97251046,0.004984131,0.015806723,0.003967138,0.000025638401],"about_ca_topic_score_codex":0.0018155728,"about_ca_topic_score_gemma":0.0016554925,"teacher_disagreement_score":0.0038784165,"about_ca_system_score_codex":0.00022858163,"about_ca_system_score_gemma":0.00057505164,"threshold_uncertainty_score":0.01297456},"labels":[],"label_agreement":null},{"id":"W1994281788","doi":"10.1007/s11634-014-0182-6","title":"Mixture model averaging for clustering","year":2014,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; University of Guelph","funders":"","keywords":"Cluster analysis; Mixture model; Model selection; Bayesian information criterion; Pattern recognition (psychology); Bayesian probability; Rand index; Closeness; Single-linkage clustering","score_opus":0.04141965676927712,"score_gpt":0.3381464725389695,"score_spread":0.2967268157696924,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1994281788","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00048498283,0.0010024091,0.997735,0.00009090996,0.000053191412,0.0000134686625,0.0000464115,0.00022642373,0.00034721533],"genre_scores_gemma":[0.053423293,0.003053226,0.9353981,0.0002791935,0.0005472989,0.00037987265,0.0012913581,0.0008382494,0.0047893655],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9942139,0.0026268277,0.00031738228,0.0012286025,0.0013967352,0.00021653004],"domain_scores_gemma":[0.9925,0.0040411525,0.00037573758,0.0016307106,0.001289783,0.00016262557],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006213329,0.0019281199,0.004913501,0.004395217,0.001754555,0.0031062576,0.0047639436,0.0032992645,0.0039079348],"category_scores_gemma":[0.020984577,0.001872896,0.003574226,0.00663276,0.0022663644,0.0048794397,0.0034234328,0.00568718,0.0027514333],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014399852,0.000088942026,0.00072869327,0.0006135478,0.0006140641,0.00007531377,0.0002954455,0.28511235,0.0031762396,0.3712723,0.016052097,0.32182705],"study_design_scores_gemma":[0.000008676136,0.000014948232,0.00027855224,0.000034173896,0.000051755298,0.00004563636,0.000013988912,0.7879755,0.0008179138,0.20485948,0.00585945,0.00004001728],"about_ca_topic_score_codex":0.009253786,"about_ca_topic_score_gemma":0.008729414,"teacher_disagreement_score":0.009253786,"about_ca_system_score_codex":0.002263075,"about_ca_system_score_gemma":0.00196348,"threshold_uncertainty_score":0.032859564},"labels":[],"label_agreement":null},{"id":"W1997354990","doi":"10.1007/s11634-013-0139-1","title":"Estimating common principal components in high dimensions","year":2013,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":79,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Convergence (economics); Principal component analysis; Majorization; Orthonormal basis; Mathematical optimization; Simple (philosophy); Cluster analysis; Minification; Computer science; Matrix (chemical analysis); Function (biology); Mathematics; Algorithm; Artificial intelligence","score_opus":0.05393879753402075,"score_gpt":0.3370129092875392,"score_spread":0.28307411175351843,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1997354990","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008128011,0.0005022664,0.9908482,0.00012542143,0.000021406408,0.00002321256,0.00005501074,0.00012921578,0.00016720727],"genre_scores_gemma":[0.18663928,0.0018875212,0.8075024,0.00018128369,0.00025727646,0.00037561922,0.0011015611,0.0002253854,0.0018297256],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.994212,0.0028599487,0.0004110298,0.0010819056,0.0011143977,0.00032074607],"domain_scores_gemma":[0.9782326,0.0163379,0.0011147462,0.0023413596,0.0016378905,0.0003355247],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0075319195,0.002091794,0.0033576041,0.0038526964,0.0014948986,0.0036259347,0.0029571955,0.0026806337,0.0015617462],"category_scores_gemma":[0.033547197,0.0018702479,0.0028672977,0.004644762,0.002506052,0.005637416,0.004340088,0.0042842943,0.000998782],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00048392458,0.00039445114,0.008977471,0.00083834416,0.00071092776,0.00027753742,0.00095302233,0.41168633,0.006055087,0.119855575,0.0077225696,0.44204485],"study_design_scores_gemma":[0.0000293445,0.000034947298,0.0015506085,0.000048546182,0.000059335838,0.00009357566,0.00010968424,0.8646457,0.0008817219,0.13094224,0.0015524174,0.000051703413],"about_ca_topic_score_codex":0.005776784,"about_ca_topic_score_gemma":0.006505601,"teacher_disagreement_score":0.0075319195,"about_ca_system_score_codex":0.0009701941,"about_ca_system_score_gemma":0.002106128,"threshold_uncertainty_score":0.03983313},"labels":[],"label_agreement":null},{"id":"W2004773941","doi":"10.1007/s11634-009-0046-7","title":"Parsimonious cluster systems","year":2009,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Paul University","funders":"","keywords":"Bijection; Set (abstract data type); Cluster (spacecraft); Cluster analysis; Phylogenetic tree; Computer science; Combinatorics; Tree (set theory); Mathematics; Theoretical computer science; Biology; Artificial intelligence; Genetics","score_opus":0.029910501458001896,"score_gpt":0.30810436773482447,"score_spread":0.2781938662768226,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2004773941","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06363475,0.0007909833,0.911894,0.0011046387,0.00027802904,0.0003211832,0.0008778848,0.0016367078,0.019461732],"genre_scores_gemma":[0.4594582,0.00052540906,0.50524706,0.00040919715,0.00019917625,0.0004650775,0.0024841216,0.00068116677,0.03053056],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99776626,0.00070141145,0.00012729739,0.00058911525,0.0005636182,0.0002522292],"domain_scores_gemma":[0.9936168,0.0015095872,0.0002974354,0.0026367821,0.0013938722,0.00054554245],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00206463,0.0007526736,0.0014439876,0.0026241953,0.0044756,0.0030837536,0.00315229,0.0018454451,0.017795635],"category_scores_gemma":[0.01265275,0.0008135227,0.0009691936,0.0034807422,0.002141717,0.0042084,0.005456557,0.0023998225,0.003996974],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007157352,0.00011368116,0.00324291,0.0002923218,0.00014099306,0.00027600167,0.001112958,0.061152674,0.005251992,0.7358726,0.018845996,0.17298214],"study_design_scores_gemma":[0.000080467566,0.000110823916,0.0013702114,0.000042820124,0.00005878577,0.00033226953,0.00029024074,0.26435083,0.0020812934,0.709029,0.02220149,0.000051804447],"about_ca_topic_score_codex":0.0018478772,"about_ca_topic_score_gemma":0.0036627306,"teacher_disagreement_score":0.017795635,"about_ca_system_score_codex":0.0012208789,"about_ca_system_score_gemma":0.0016363302,"threshold_uncertainty_score":0.059532285},"labels":[],"label_agreement":null},{"id":"W2016714679","doi":"10.1007/s11634-010-0083-2","title":"Generalized GIPSCAL re-revisited: a fast convergent algorithm with acceleration by the minimal polynomial extrapolation","year":2011,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Analytical Chemistry and Chromatography","field":"Chemistry","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Korea Institute of Energy Research","keywords":"Extrapolation; Acceleration; Diagonal; Positive definiteness; Algorithm; Representation (politics); Polynomial; Mathematics; Time complexity; Matrix (chemical analysis); Computational complexity theory; Convergence (economics); Applied mathematics; Computer science; Mathematical analysis; Geometry; Positive-definite matrix","score_opus":0.03893491462284368,"score_gpt":0.2847833179878061,"score_spread":0.24584840336496241,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2016714679","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009341317,0.00032421382,0.9851313,0.00023703731,0.00016333713,0.000073243646,0.000104688996,0.0025473866,0.0020775713],"genre_scores_gemma":[0.068862304,0.00015070772,0.9255678,0.00016457375,0.00008862489,0.00014364457,0.00031194088,0.00085802516,0.0038523322],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.999244,0.00021624398,0.000038703693,0.00011844188,0.00029819194,0.00008448344],"domain_scores_gemma":[0.99871695,0.0003982656,0.000059643153,0.0003714567,0.0003724073,0.00008126652],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012887727,0.0012614825,0.0016830628,0.0010939165,0.00078905234,0.0012018627,0.002471613,0.0013182284,0.008340525],"category_scores_gemma":[0.0054739136,0.0004964425,0.0009891825,0.0015192833,0.00075405167,0.0012810994,0.0024614353,0.0026404483,0.003989259],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008863381,0.00019256494,0.001104145,0.00039664507,0.00015694673,0.0003292676,0.0002457584,0.1470604,0.016006883,0.043749984,0.023291893,0.7665792],"study_design_scores_gemma":[0.000076120894,0.000072548326,0.00020973124,0.000021408645,0.00002141324,0.00012227484,0.000032042335,0.9752192,0.0055182492,0.01214049,0.006542898,0.00002366927],"about_ca_topic_score_codex":0.0046415664,"about_ca_topic_score_gemma":0.0073810173,"teacher_disagreement_score":0.008340525,"about_ca_system_score_codex":0.00057325535,"about_ca_system_score_gemma":0.0022822958,"threshold_uncertainty_score":0.027901888},"labels":[],"label_agreement":null},{"id":"W2027846664","doi":"10.1007/s11634-012-0110-6","title":"Time series classification by class-specific Mahalanobis distance measures","year":2012,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Mahalanobis distance; Covariance matrix; Dynamic time warping; Covariance; Pattern recognition (psychology); Series (stratigraphy); k-nearest neighbors algorithm; Margin (machine learning)","score_opus":0.03447962751202253,"score_gpt":0.27522920034167747,"score_spread":0.24074957282965495,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2027846664","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10141822,0.00090425636,0.8921347,0.00023530608,0.00022546426,0.00013421089,0.0006766318,0.0009954716,0.0032757616],"genre_scores_gemma":[0.6742879,0.00081038784,0.31736374,0.000086775755,0.00019852148,0.00031100737,0.0028429092,0.00021398066,0.00388477],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99781114,0.0004662678,0.00027455337,0.00046080747,0.00086195575,0.00012539845],"domain_scores_gemma":[0.9972531,0.0008680813,0.0002942668,0.0004646489,0.0010296969,0.00009019436],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019780067,0.0006995988,0.00092372566,0.0035977573,0.00064626825,0.0021113337,0.0009113082,0.0010474424,0.0015238523],"category_scores_gemma":[0.007447287,0.00017868655,0.001185421,0.003312698,0.0005216214,0.0020186354,0.0009542584,0.0012199433,0.0014247756],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043240565,0.00033264793,0.016628314,0.00031151768,0.0003833321,0.00008988371,0.0002806281,0.049165804,0.01723975,0.017800659,0.0063019227,0.8910331],"study_design_scores_gemma":[0.000017115748,0.00019752243,0.018723456,0.000056185498,0.00010788484,0.00030213656,0.00028192994,0.93550175,0.014684666,0.024798937,0.005242471,0.000085885265],"about_ca_topic_score_codex":0.0013548823,"about_ca_topic_score_gemma":0.0015135485,"teacher_disagreement_score":0.0035977573,"about_ca_system_score_codex":0.0006291059,"about_ca_system_score_gemma":0.00077513396,"threshold_uncertainty_score":0.010460854},"labels":[],"label_agreement":null},{"id":"W2057846446","doi":"10.1007/s11634-009-0054-7","title":"Tests of ignoring and eliminating in nonsymmetric correspondence analysis","year":2009,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Sensory Analysis and Statistical Methods","field":"Agricultural and Biological Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Contingency table; Residual; Contingency; Row; Table (database); Correspondence analysis; Test (biology); Mathematics; Econometrics; Statistics; Computer science; Algorithm; Data mining; Epistemology","score_opus":0.06777325830079646,"score_gpt":0.36711674864542976,"score_spread":0.2993434903446333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2057846446","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5268863,0.0008944632,0.44835395,0.002245605,0.0007813001,0.0009609953,0.0012905598,0.0005511821,0.018035663],"genre_scores_gemma":[0.929972,0.00012927373,0.06652335,0.0005969481,0.0002741492,0.0007638178,0.00082641776,0.00017768647,0.0007363779],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.7809508,0.13976172,0.014353136,0.03147719,0.029582728,0.0038743312],"domain_scores_gemma":[0.20498222,0.74042416,0.013526962,0.029779831,0.008444496,0.0028422915],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.10081848,0.0015965264,0.0026962995,0.004592804,0.0044816867,0.004700281,0.0044045583,0.0031322944,0.007654856],"category_scores_gemma":[0.45910677,0.0008446871,0.003992441,0.005633133,0.015191981,0.009585312,0.00672,0.0067907744,0.0006789374],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.009920369,0.0015792424,0.2723689,0.0020674595,0.007796431,0.0052701924,0.010245661,0.023294145,0.012945127,0.35008425,0.009477292,0.294951],"study_design_scores_gemma":[0.00079779257,0.0039179833,0.11510503,0.00040108166,0.0013508922,0.0027897542,0.0056200554,0.14488989,0.01111158,0.7034617,0.010036682,0.000517469],"about_ca_topic_score_codex":0.00146547,"about_ca_topic_score_gemma":0.0009296139,"teacher_disagreement_score":0.10081848,"about_ca_system_score_codex":0.001687341,"about_ca_system_score_gemma":0.003761027,"threshold_uncertainty_score":0.5331855},"labels":[],"label_agreement":null},{"id":"W2061107170","doi":"10.1007/s11634-014-0168-4","title":"Feature selection for fault level diagnosis of planetary gearboxes","year":2014,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Gear and Bearing Dynamics Analysis","field":"Engineering","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Feature selection; Pattern recognition (psychology); Feature (linguistics); Fault (geology); Computer science; Artificial intelligence; Kernel (algebra); Minimum redundancy feature selection; Similarity (geometry); Selection (genetic algorithm); Feature vector; Feature extraction; Ranking (information retrieval); Algorithm; Data mining; Mathematics","score_opus":0.025546073147799032,"score_gpt":0.27256118428713894,"score_spread":0.2470151111393399,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2061107170","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.363656,0.0011721754,0.63054526,0.00024664638,0.00013092053,0.00006747674,0.0010403633,0.0021514553,0.0009896485],"genre_scores_gemma":[0.95313597,0.00015696816,0.044752616,0.0000304551,0.000043588116,0.000039482256,0.0011018979,0.0000373037,0.0007016862],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99967337,0.000059769773,0.0000391338,0.00006645783,0.00010622627,0.000055117238],"domain_scores_gemma":[0.9990742,0.0005439704,0.00006799604,0.0000804157,0.00019291295,0.000040445502],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000599621,0.0005375102,0.00081451255,0.0012598562,0.00026778338,0.0005612024,0.00044365806,0.0003551446,0.001099729],"category_scores_gemma":[0.0020122218,0.00013455107,0.0004549178,0.00069805275,0.00015787523,0.0004139912,0.0003600461,0.00040014563,0.00032756868],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00087718805,0.00026331356,0.013749922,0.00017409657,0.00009955946,0.0002634012,0.00008133008,0.041699022,0.05032932,0.0006690455,0.004670557,0.8871232],"study_design_scores_gemma":[0.000041978714,0.00024886074,0.02491604,0.000021115431,0.00008003488,0.00027757048,0.0000828025,0.9458085,0.024761358,0.0019996294,0.0017398258,0.000022331478],"about_ca_topic_score_codex":0.0014933278,"about_ca_topic_score_gemma":0.0011876967,"teacher_disagreement_score":0.0014933278,"about_ca_system_score_codex":0.00018024321,"about_ca_system_score_gemma":0.00032660607,"threshold_uncertainty_score":0.0036789775},"labels":[],"label_agreement":null},{"id":"W2065708566","doi":"10.1007/s11634-013-0155-1","title":"A LASSO-penalized BIC for mixture model selection","year":2013,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":58,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Lasso (programming language); Model selection; Selection (genetic algorithm); Mathematics; Computer science; Artificial intelligence; Statistics","score_opus":0.04016452122136844,"score_gpt":0.34145722688980523,"score_spread":0.3012927056684368,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2065708566","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009504786,0.00020678545,0.9976071,0.000205962,0.00006079603,0.00003718826,0.000080798985,0.00037995476,0.00047105283],"genre_scores_gemma":[0.05332834,0.0004324702,0.9379418,0.0006264013,0.00029920848,0.0005547467,0.0013017298,0.0010173568,0.004497998],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98833734,0.007835603,0.00042841144,0.0009649674,0.0020861102,0.00034757232],"domain_scores_gemma":[0.9885582,0.0066299313,0.000385454,0.0015044649,0.0024570012,0.00046497202],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011733128,0.002041687,0.0038900098,0.0024066337,0.0017232028,0.0028576807,0.005500396,0.003192741,0.006356998],"category_scores_gemma":[0.03979413,0.0016759043,0.0021021774,0.0031092388,0.0019073025,0.0026785848,0.0055313865,0.007030465,0.0040317443],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00081402523,0.00039167676,0.0020315475,0.00063533866,0.00067906245,0.0002607791,0.00036936588,0.37958297,0.005785726,0.15552627,0.042921938,0.4110012],"study_design_scores_gemma":[0.00004142516,0.000043775228,0.0002382526,0.00005064406,0.000036613295,0.00008976458,0.000023726501,0.9455036,0.00066468766,0.050205126,0.0030568137,0.000045663688],"about_ca_topic_score_codex":0.0053931167,"about_ca_topic_score_gemma":0.008165307,"teacher_disagreement_score":0.011733128,"about_ca_system_score_codex":0.0011323333,"about_ca_system_score_gemma":0.0036434433,"threshold_uncertainty_score":0.062051475},"labels":[],"label_agreement":null},{"id":"W2074660585","doi":"10.1007/s11634-013-0126-6","title":"Functional fuzzy clusterwise regression analysis","year":2013,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Fuzzy Systems and Optimization","field":"Mathematics","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University; McGill University","funders":"","keywords":"Fuzzy logic; Regression analysis; Regression; Function (biology); Computer science; Data mining; Mathematics; Mathematical optimization; Statistics; Artificial intelligence","score_opus":0.06690308779884142,"score_gpt":0.33977078719510917,"score_spread":0.27286769939626776,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2074660585","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0056682886,0.00043656633,0.9922655,0.00017529771,0.00006743444,0.000025474514,0.000078475634,0.00021943229,0.0010635523],"genre_scores_gemma":[0.34495074,0.0014218348,0.6393348,0.00030422164,0.00047225662,0.00031040152,0.0011864742,0.0006490854,0.011370091],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99536717,0.0027950907,0.00016190257,0.00066823646,0.0008314513,0.00017625246],"domain_scores_gemma":[0.99174374,0.003956578,0.00045841074,0.0014813134,0.0022090268,0.00015093136],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008975799,0.0012563218,0.0019479721,0.0027046916,0.0009737163,0.0016279636,0.0020646143,0.001082458,0.0036234476],"category_scores_gemma":[0.016916102,0.0005025953,0.001810965,0.002514661,0.0015168324,0.0017220319,0.0016460419,0.0021962174,0.0012196446],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043229345,0.00013064168,0.0035213162,0.00055195944,0.00075146847,0.00014794638,0.0002708862,0.21840526,0.0064867027,0.39712444,0.012559269,0.35961786],"study_design_scores_gemma":[0.000014890313,0.00007956688,0.0016479386,0.000040683615,0.00010408074,0.0000953355,0.000050023995,0.8898934,0.003038222,0.09959567,0.005404671,0.000035474557],"about_ca_topic_score_codex":0.0024374654,"about_ca_topic_score_gemma":0.001958241,"teacher_disagreement_score":0.008975799,"about_ca_system_score_codex":0.0012377612,"about_ca_system_score_gemma":0.0016559646,"threshold_uncertainty_score":0.04746914},"labels":[],"label_agreement":null},{"id":"W2082905687","doi":"10.1007/s11634-013-0152-4","title":"Infinite Dirichlet mixture models learning via expectation propagation","year":2013,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Dirichlet process; Mixture model; Model selection; Dirichlet distribution; Computer science; Hierarchical Dirichlet process; Inference; Automatic summarization; Cluster analysis; Bayesian inference; Minimum description length; Artificial intelligence; Latent Dirichlet allocation; Data mining; Selection (genetic algorithm); Synthetic data; Machine learning; Bayesian probability; Topic model; Mathematics","score_opus":0.030519176843317914,"score_gpt":0.3035554330616236,"score_spread":0.2730362562183057,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2082905687","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001668642,0.00037447092,0.9970862,0.00022221143,0.00002806746,0.00001701212,0.00005878383,0.00017805825,0.0003664835],"genre_scores_gemma":[0.20490438,0.00255463,0.7782052,0.00083126646,0.00074456073,0.0007768663,0.0019018718,0.0005728711,0.009508403],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99262494,0.0041366722,0.0003628814,0.0013500191,0.0011666195,0.00035883742],"domain_scores_gemma":[0.9733279,0.022639193,0.00075084827,0.0015834379,0.0013959152,0.00030268464],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009573735,0.0018702745,0.0048259897,0.0032708845,0.0016111288,0.0045298547,0.0066458364,0.003988808,0.004819538],"category_scores_gemma":[0.034792066,0.0025937054,0.003377579,0.004297863,0.0034814964,0.008995174,0.00505566,0.0071493997,0.002241221],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034726583,0.00021694586,0.0015580913,0.00036856215,0.00033283638,0.00013747872,0.0004615,0.45535564,0.00091762113,0.35222903,0.007576832,0.18049817],"study_design_scores_gemma":[0.000019709465,0.00001044046,0.000103873615,0.000028102837,0.000024292929,0.000030781153,0.0000134292195,0.83741,0.00024790442,0.16106822,0.001018398,0.00002484266],"about_ca_topic_score_codex":0.006404571,"about_ca_topic_score_gemma":0.0091215335,"teacher_disagreement_score":0.009573735,"about_ca_system_score_codex":0.002411637,"about_ca_system_score_gemma":0.0022773675,"threshold_uncertainty_score":0.050631404},"labels":[],"label_agreement":null},{"id":"W2102723201","doi":"10.1007/s11634-013-0124-8","title":"Clustering and classification via cluster-weighted factor analyzers","year":2013,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":53,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Cluster analysis; Mixture model; Expectation–maximization algorithm; Pattern recognition (psychology); Covariance matrix; Covariance; Latent variable; Ranking (information retrieval); Fuzzy clustering","score_opus":0.03672495443431135,"score_gpt":0.317044713978554,"score_spread":0.28031975954424265,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102723201","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025057113,0.00008065372,0.99679786,0.000040973227,0.000010710117,0.000030799005,0.000039065995,0.0002762012,0.00021807263],"genre_scores_gemma":[0.084636934,0.00024922343,0.9127181,0.00007391344,0.00006172367,0.00028194563,0.00047019043,0.00030589648,0.0012020654],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99273455,0.0032504457,0.00047102539,0.0015950137,0.0016161294,0.00033290085],"domain_scores_gemma":[0.99038976,0.005241272,0.0005653328,0.0012457645,0.0023534787,0.000204407],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007909577,0.0016502191,0.001816836,0.004424331,0.0014170536,0.0034090898,0.0027483485,0.0014544507,0.0034420975],"category_scores_gemma":[0.029911974,0.00091743807,0.0024734617,0.0049933516,0.0016900493,0.0042838855,0.003277497,0.0024394544,0.0019199654],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007525697,0.00021472629,0.0031057624,0.00026075038,0.00056408095,0.00013452755,0.0006881983,0.10037284,0.013273594,0.12447276,0.004745431,0.7514148],"study_design_scores_gemma":[0.000039361603,0.00004843883,0.0010972084,0.000031627787,0.00010832203,0.00010784221,0.000114767114,0.8420078,0.003513041,0.15017337,0.0026994348,0.00005880538],"about_ca_topic_score_codex":0.0040169153,"about_ca_topic_score_gemma":0.0045314487,"teacher_disagreement_score":0.007909577,"about_ca_system_score_codex":0.0009859847,"about_ca_system_score_gemma":0.0019527607,"threshold_uncertainty_score":0.04183036},"labels":[],"label_agreement":null},{"id":"W2112120651","doi":"10.1007/s11634-009-0056-5","title":"Regularized fuzzy clusterwise ridge regression","year":2009,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Fuzzy Systems and Optimization","field":"Mathematics","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Multicollinearity; Ordinary least squares; Fuzzy logic; Statistics; Regression analysis; Mathematics; Linear regression; Regression; Cluster analysis; Regression diagnostic; Local regression; Fuzzy clustering; Data mining; Polynomial regression; Computer science; Artificial intelligence","score_opus":0.06134931625472312,"score_gpt":0.36859024644580013,"score_spread":0.307240930191077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112120651","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018038673,0.00018404324,0.98079395,0.000099092606,0.000039486862,0.00001437835,0.00006137837,0.00023291862,0.0005359954],"genre_scores_gemma":[0.45316464,0.0003469692,0.54069275,0.0001278939,0.00018200254,0.00009640268,0.0005058295,0.0002546202,0.00462899],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9983321,0.0006639203,0.00008013029,0.00033737172,0.00046554036,0.000120918034],"domain_scores_gemma":[0.99643767,0.0010778452,0.00028538573,0.0009794986,0.0011279223,0.00009161613],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004149397,0.00058557314,0.0016470483,0.00086679123,0.00046545945,0.0012054445,0.0016086203,0.0010489434,0.0014442613],"category_scores_gemma":[0.00744277,0.00044638387,0.0011428484,0.0010731544,0.0011727398,0.001074536,0.0014367679,0.0014870712,0.0005571151],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00060132495,0.00014083929,0.00301511,0.00025282035,0.00035372263,0.00014577637,0.00017473657,0.5920129,0.019267695,0.114356786,0.0058276304,0.26385078],"study_design_scores_gemma":[0.0000068923528,0.0000241181,0.00033603146,0.000005189861,0.000013339672,0.000024974677,0.0000067558467,0.9863944,0.0015560254,0.011083023,0.0005394538,0.00000969173],"about_ca_topic_score_codex":0.0013869512,"about_ca_topic_score_gemma":0.0016874663,"teacher_disagreement_score":0.004149397,"about_ca_system_score_codex":0.0005887095,"about_ca_system_score_gemma":0.0009929484,"threshold_uncertainty_score":0.021944344},"labels":[],"label_agreement":null},{"id":"W2353061623","doi":"10.1007/s11634-016-0250-1","title":"Latent class model with conditional dependency per modes to cluster categorical data","year":2016,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Categorical variable; Conditional independence; Multinomial distribution; Independence (probability theory); Model selection; Latent class model; Mathematics; Local independence; Conditional probability distribution; Conditional probability; Mixture model; Latent variable; Class (philosophy); Expectation–maximization algorithm; Information Criteria; Dependency (UML); Statistics; Posterior probability; Algorithm; Latent variable model; Computer science; Artificial intelligence; Maximum likelihood; Bayesian probability","score_opus":0.06299037588866112,"score_gpt":0.3427291082985342,"score_spread":0.27973873240987307,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2353061623","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023426819,0.00015817655,0.9966697,0.00016633503,0.000035711222,0.000047137466,0.00014153632,0.00020552192,0.00023325472],"genre_scores_gemma":[0.1754778,0.0007566862,0.8133237,0.0006807277,0.00034252912,0.0015054876,0.0024565246,0.00067305943,0.0047835084],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9912641,0.0047223875,0.00044034183,0.0017076561,0.0013443064,0.00052127626],"domain_scores_gemma":[0.98460335,0.0094834315,0.0008073425,0.0033264905,0.0014271762,0.0003522961],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012196678,0.0014028737,0.002796453,0.0033636878,0.001755256,0.00297979,0.005208896,0.0027580974,0.0055218707],"category_scores_gemma":[0.03276096,0.0011973418,0.0046300762,0.0051127616,0.0025929457,0.0051333974,0.0041482467,0.008260289,0.0030273986],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009822195,0.00076715887,0.010208617,0.0004316874,0.0011052272,0.00016673868,0.0018028604,0.20938325,0.0044214297,0.3980517,0.018686207,0.35399288],"study_design_scores_gemma":[0.000041477528,0.000040522424,0.0010312746,0.000075645665,0.00011271503,0.00009619234,0.00008906439,0.78445405,0.0007011048,0.21049891,0.002789744,0.00006932936],"about_ca_topic_score_codex":0.007979418,"about_ca_topic_score_gemma":0.009922139,"teacher_disagreement_score":0.012196678,"about_ca_system_score_codex":0.0018854907,"about_ca_system_score_gemma":0.0031625708,"threshold_uncertainty_score":0.064502954},"labels":[],"label_agreement":null},{"id":"W2609430174","doi":"10.1007/s11634-017-0286-x","title":"Local generalized quadratic distance metrics: application to the k-nearest neighbors classifier","year":2017,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Face and Expression Recognition","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; University of Alberta","funders":"","keywords":"k-nearest neighbors algorithm; Metric (unit); Curse of dimensionality; Computer science; Mathematics; Artificial intelligence; Algorithm; Pattern recognition (psychology)","score_opus":0.041089338011552366,"score_gpt":0.3371891365524899,"score_spread":0.29609979854093754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2609430174","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020610504,0.0012125961,0.97573256,0.00021441591,0.00007803053,0.000058022877,0.00012368667,0.00038832778,0.0015819187],"genre_scores_gemma":[0.3620388,0.0008857284,0.63253397,0.000093335184,0.00014553058,0.00014823739,0.0004802608,0.00025354858,0.0034205841],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9971367,0.00088959874,0.00018458518,0.00034106668,0.0013623815,0.00008564015],"domain_scores_gemma":[0.9954803,0.0016909585,0.00031844457,0.00042318052,0.0019371179,0.00014996194],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028962395,0.00056358724,0.0017831712,0.0019323354,0.0008106453,0.0016402502,0.0016555893,0.0010578195,0.0015112004],"category_scores_gemma":[0.011252664,0.00026116392,0.00064171356,0.0027304695,0.000857917,0.001989581,0.0016607273,0.0010845425,0.0007494597],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023572728,0.00018975082,0.0036253112,0.00033011285,0.00010912197,0.00010449317,0.0003008085,0.17984177,0.00680084,0.042275626,0.007809927,0.7583765],"study_design_scores_gemma":[0.000009836979,0.00006229608,0.0010517448,0.000015714038,0.000013929435,0.0001032698,0.00004887582,0.97965425,0.0012411542,0.015347973,0.0024270175,0.000023847131],"about_ca_topic_score_codex":0.007449297,"about_ca_topic_score_gemma":0.0075800205,"teacher_disagreement_score":0.007449297,"about_ca_system_score_codex":0.0010950367,"about_ca_system_score_gemma":0.0015963317,"threshold_uncertainty_score":0.015316963},"labels":[],"label_agreement":null},{"id":"W2790767741","doi":"10.1007/s11634-020-00395-7","title":"Clustering discrete-valued time series","year":2020,"lang":"en","type":"preprint","venue":"Advances in Data Analysis and Classification","topic":"Time Series Analysis and Forecasting","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Cluster analysis; Series (stratigraphy); Autoregressive model; Computer science; Model selection; Expectation–maximization algorithm; STAR model; Time series; Data mining; Focus (optics); Algorithm; Mathematics; Artificial intelligence; Autoregressive integrated moving average; Machine learning; Econometrics; Statistics; Maximum likelihood","score_opus":0.04408920708021554,"score_gpt":0.30893820675356276,"score_spread":0.26484899967334724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2790767741","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.060629383,0.0018613755,0.93479496,0.00044257255,0.00024819336,0.00006474152,0.00026157429,0.00036385638,0.0013333797],"genre_scores_gemma":[0.6272942,0.0025521207,0.36129504,0.00012945915,0.00053770124,0.00014329205,0.002674824,0.00018498859,0.0051883557],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9991755,0.00021497511,0.000080126585,0.00023182743,0.00024755663,0.000049964696],"domain_scores_gemma":[0.9977748,0.0012000491,0.00016862556,0.0004045977,0.00038942523,0.000062468505],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015864205,0.00059580134,0.0009992094,0.0024975617,0.0004577492,0.0018280273,0.00085961045,0.0010905707,0.0015566696],"category_scores_gemma":[0.00756284,0.0003095076,0.00080950017,0.0026177207,0.0006603351,0.0015061335,0.00072383287,0.0009912367,0.00062743796],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026078094,0.00023447903,0.008395463,0.0005612091,0.0004130388,0.0001354455,0.00037785998,0.24548756,0.016111016,0.05886011,0.00765068,0.6615123],"study_design_scores_gemma":[0.000008764542,0.00002616322,0.0023470449,0.00002356886,0.000034465236,0.00004101699,0.000067289126,0.95162416,0.0017751831,0.042328853,0.0017121653,0.0000112679645],"about_ca_topic_score_codex":0.0018932832,"about_ca_topic_score_gemma":0.001298426,"teacher_disagreement_score":0.0024975617,"about_ca_system_score_codex":0.0006944213,"about_ca_system_score_gemma":0.00054711,"threshold_uncertainty_score":0.00838989},"labels":[],"label_agreement":null},{"id":"W2888010039","doi":"10.1007/s11634-018-0333-2","title":"Subspace clustering for the finite mixture of generalized hyperbolic distributions","year":2018,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Cluster analysis; Mixture model; Mathematics; Subspace topology; Applied mathematics; Hyperbolic function; Gaussian; Finite element method; Extension (predicate logic); Mathematical analysis; Computer science; Statistics; Physics","score_opus":0.05135122907555946,"score_gpt":0.3540267272002225,"score_spread":0.30267549812466304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2888010039","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003349117,0.00023680099,0.99579465,0.00010110641,0.00001448692,0.000027777998,0.00007061401,0.00014425523,0.000261136],"genre_scores_gemma":[0.1931338,0.0014565224,0.7947109,0.00027971444,0.00023502819,0.00059273734,0.0018808864,0.00063900754,0.0070713917],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99559855,0.0023790041,0.00022208988,0.00085806835,0.00072698237,0.0002152848],"domain_scores_gemma":[0.99043745,0.005746431,0.00064859225,0.0015986882,0.0011828128,0.0003860082],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0070269145,0.0012131736,0.0025593792,0.0030901532,0.001534034,0.0030986804,0.004380107,0.0023828277,0.003864953],"category_scores_gemma":[0.021378065,0.0012643859,0.0030610075,0.0028298197,0.0031500808,0.004783451,0.0041317916,0.0039145057,0.0017600685],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020805823,0.00009597844,0.0015816286,0.00031434424,0.00025640472,0.000078294535,0.0005372898,0.42401788,0.002229314,0.44746217,0.0044062966,0.11881228],"study_design_scores_gemma":[0.000008261008,0.000011735807,0.00020298622,0.00001697647,0.000012508331,0.000026605054,0.000029576298,0.86674637,0.00023192167,0.13171835,0.000967546,0.000027171764],"about_ca_topic_score_codex":0.008353349,"about_ca_topic_score_gemma":0.008369314,"teacher_disagreement_score":0.008353349,"about_ca_system_score_codex":0.002338896,"about_ca_system_score_gemma":0.0023768514,"threshold_uncertainty_score":0.037162304},"labels":[],"label_agreement":null},{"id":"W2891365885","doi":"10.1007/s11634-019-00377-4","title":"Mixtures of skewed matrix variate bilinear factor analyzers","year":2019,"lang":"en","type":"preprint","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Random variate; Kurtosis; Skewness; Bilinear interpolation; Statistics; Mathematics; Skew; Matrix (chemical analysis); Gaussian; Inverse Gaussian distribution; Applied mathematics; Computer science; Distribution (mathematics); Mathematical analysis; Physics; Random variable","score_opus":0.04467400490465936,"score_gpt":0.3678903485041041,"score_spread":0.3232163435994448,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2891365885","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009858417,0.00011545849,0.98882246,0.00006347895,0.000026253709,0.000039525192,0.000110216875,0.00028799663,0.00067615096],"genre_scores_gemma":[0.35014662,0.00074317656,0.6372366,0.00025089688,0.00017160615,0.00038552424,0.0011225287,0.0006511174,0.009292029],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99368864,0.002782838,0.0002878919,0.0012623131,0.0014682802,0.00051009894],"domain_scores_gemma":[0.98995847,0.0047891666,0.0007184,0.0021025874,0.0020039545,0.00042742077],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005543455,0.0013575715,0.0013914217,0.0023516314,0.00082191825,0.0036350817,0.0016968505,0.0013090563,0.007207683],"category_scores_gemma":[0.026387712,0.0011940274,0.0017137732,0.0035044078,0.0016583588,0.005170983,0.003699074,0.0021105856,0.0026881157],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016126528,0.00026338716,0.008328746,0.00029493263,0.00035831108,0.00041061765,0.00095122296,0.13099599,0.031371552,0.3584635,0.0054964926,0.4614527],"study_design_scores_gemma":[0.000038526567,0.00009655882,0.0019686972,0.00004833831,0.00006450767,0.00042383553,0.00016457663,0.8031267,0.007931267,0.18046191,0.0055767894,0.00009825015],"about_ca_topic_score_codex":0.0015191603,"about_ca_topic_score_gemma":0.0018770698,"teacher_disagreement_score":0.007207683,"about_ca_system_score_codex":0.0008807502,"about_ca_system_score_gemma":0.0011056866,"threshold_uncertainty_score":0.029316962},"labels":[],"label_agreement":null},{"id":"W2990751717","doi":"10.1007/s11634-019-00379-2","title":"Is-ClusterMPP: clustering algorithm through point processes and influence space towards high-dimensional data","year":2019,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Data Management and Algorithms","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université TÉLUQ; Université du Québec à Montréal","funders":"","keywords":"Cluster analysis; Curse of dimensionality; Algorithm; Data point; Computer science; Subspace topology; Mathematics; Statistical physics; Artificial intelligence; Physics","score_opus":0.0354850183539955,"score_gpt":0.31527472121398364,"score_spread":0.27978970285998817,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2990751717","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025267424,0.00011101668,0.99309945,0.000071649054,0.0000528801,0.0000919236,0.00020842126,0.0031333617,0.00070442946],"genre_scores_gemma":[0.04310918,0.00016818185,0.95166695,0.00007781869,0.0000639659,0.0003566966,0.0011959798,0.0008934099,0.002467859],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99855477,0.0003041665,0.0000734127,0.00030085488,0.00066154473,0.00010533108],"domain_scores_gemma":[0.99874395,0.0002818114,0.00006124617,0.000295683,0.00055319624,0.000064191256],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001543198,0.001473635,0.0017921159,0.002692093,0.0022498874,0.0021220532,0.0038439678,0.0017889062,0.0056795697],"category_scores_gemma":[0.005146992,0.0007726229,0.0019873003,0.0046054386,0.0009175259,0.0018611707,0.0036128156,0.0019359668,0.0037339067],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00052293803,0.00021667221,0.0032702463,0.00053824845,0.0004199035,0.00021090366,0.00056732586,0.15987642,0.0112265805,0.03657522,0.031346813,0.75522876],"study_design_scores_gemma":[0.000056533947,0.000064284795,0.00078649103,0.000025499421,0.0000550326,0.00017974009,0.00006965774,0.9575888,0.0087392265,0.019328969,0.0130661735,0.000039516694],"about_ca_topic_score_codex":0.007608857,"about_ca_topic_score_gemma":0.00921559,"teacher_disagreement_score":0.007608857,"about_ca_system_score_codex":0.00092062715,"about_ca_system_score_gemma":0.00239973,"threshold_uncertainty_score":0.019000053},"labels":[],"label_agreement":null},{"id":"W3041023245","doi":"10.1007/s11634-020-00408-5","title":"Active learning of constraints for weighted feature selection","year":2020,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Face and Expression Recognition","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Conseil National de la Recherche Scientifique; Agence Universitaire de la Francophonie","keywords":"Pairwise comparison; Feature selection; Laplacian matrix; Computer science; Cluster analysis; Graph; Feature (linguistics); Constraint (computer-aided design); Selection (genetic algorithm); Laplace operator; Data mining; Set (abstract data type); Machine learning; Theoretical computer science; Artificial intelligence; Mathematical optimization; Mathematics","score_opus":0.03703733615695371,"score_gpt":0.31498423960111605,"score_spread":0.27794690344416234,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3041023245","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040808693,0.0002223109,0.99499965,0.00011031195,0.000030189663,0.000028094597,0.000051800973,0.000117211974,0.0003595995],"genre_scores_gemma":[0.43314147,0.0008006272,0.55558234,0.00050603587,0.0003645394,0.0009560666,0.0012160523,0.0004926582,0.0069402466],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99763346,0.0011713335,0.00013991894,0.00038908276,0.0004986331,0.00016749138],"domain_scores_gemma":[0.98822695,0.009264238,0.0004079156,0.0006761056,0.0011424837,0.00028222118],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004850321,0.0014298748,0.00312221,0.0013811048,0.0007099782,0.002044236,0.004445631,0.0022306098,0.004404713],"category_scores_gemma":[0.015616124,0.0014302772,0.0012757854,0.001976438,0.0018350903,0.0034325977,0.002896661,0.0029434059,0.0007639159],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005331479,0.0003062326,0.0009757907,0.0004486448,0.00029534937,0.00012612427,0.00015959454,0.58874047,0.0057930863,0.102790594,0.008514685,0.29131627],"study_design_scores_gemma":[0.000020481459,0.000027246764,0.000051921754,0.000010991714,0.0000092366045,0.000010320316,0.0000045016473,0.9843411,0.00054357847,0.014497549,0.00047702112,0.0000060007856],"about_ca_topic_score_codex":0.0033490686,"about_ca_topic_score_gemma":0.003897906,"teacher_disagreement_score":0.004850321,"about_ca_system_score_codex":0.0009934272,"about_ca_system_score_gemma":0.0014301798,"threshold_uncertainty_score":0.025651276},"labels":[],"label_agreement":null},{"id":"W3119925796","doi":"10.1007/s11634-020-00432-5","title":"Functional data clustering by projection into latent generalized hyperbolic subspaces","year":2021,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Cluster analysis; Linear subspace; Projection (relational algebra); Dimension (graph theory); Basis (linear algebra); Mathematics; Clustering high-dimensional data; Computer science; Applied mathematics; Pattern recognition (psychology); Algorithm; Artificial intelligence; Combinatorics; Pure mathematics","score_opus":0.07372087970399593,"score_gpt":0.3428838780099273,"score_spread":0.2691629983059314,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3119925796","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034900904,0.0000609708,0.9960813,0.00005302835,0.0000055555242,0.000021181653,0.000035154782,0.000095882206,0.00015692045],"genre_scores_gemma":[0.20771678,0.00049428194,0.787617,0.00011016884,0.00007710472,0.00036925197,0.0008116728,0.0003716488,0.002432064],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99617285,0.0022701626,0.00017398315,0.00066158175,0.00057345734,0.00014800552],"domain_scores_gemma":[0.99519557,0.0023439275,0.00037831703,0.0010979379,0.0007891793,0.00019502269],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004697367,0.0010811223,0.0017413202,0.002402109,0.0011021055,0.0023333114,0.002328334,0.0011538398,0.0019961693],"category_scores_gemma":[0.010546065,0.0011459499,0.0023730744,0.002392168,0.002530766,0.003309672,0.0048579685,0.0027139063,0.00089026825],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033607802,0.0001809043,0.0030550028,0.00040138472,0.00043459784,0.00010443122,0.0010798678,0.34268883,0.0064670774,0.3244637,0.0038818358,0.31690633],"study_design_scores_gemma":[0.0000108920485,0.000027568805,0.000400703,0.000020799787,0.000019176872,0.00004291565,0.00006567218,0.87508285,0.00075516565,0.12249433,0.0010506582,0.000029274526],"about_ca_topic_score_codex":0.0033922167,"about_ca_topic_score_gemma":0.0029252851,"teacher_disagreement_score":0.004697367,"about_ca_system_score_codex":0.0011226637,"about_ca_system_score_gemma":0.0017061854,"threshold_uncertainty_score":0.024842322},"labels":[],"label_agreement":null},{"id":"W3126183644","doi":"10.1007/s11634-022-00514-6","title":"Determinantal consensus clustering","year":2022,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Institut de Valorisation des Données","keywords":"Cluster analysis; Point process; Single-linkage clustering; Mathematics; Computer science; Kernel (algebra); Similarity (geometry); Consensus clustering; Data mining; Artificial intelligence; Correlation clustering; CURE data clustering algorithm; Combinatorics; Statistics","score_opus":0.053565366313241526,"score_gpt":0.36025569701971255,"score_spread":0.306690330706471,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3126183644","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.011461968,0.00042988203,0.98090255,0.00033036558,0.0001447537,0.00004847888,0.0000787851,0.00049599924,0.00610717],"genre_scores_gemma":[0.6241547,0.0007937188,0.3486748,0.000512854,0.00040312958,0.00025641953,0.0010361517,0.0007658947,0.023402372],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99615127,0.001358851,0.0001644325,0.0010721693,0.0010209696,0.00023214368],"domain_scores_gemma":[0.99267346,0.0020045144,0.00041274878,0.002033559,0.0024334418,0.00044221996],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0033155147,0.00080613245,0.0019959745,0.0026508651,0.0020482887,0.0027067184,0.0025358049,0.002153284,0.005756711],"category_scores_gemma":[0.012573305,0.0008660825,0.0012368024,0.0022266484,0.0025710575,0.002565794,0.003932743,0.002266012,0.0026528311],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026408737,0.00010062675,0.0018017656,0.00027558728,0.00023035471,0.00017283585,0.00048682268,0.22310592,0.008048375,0.5463808,0.017672999,0.20145981],"study_design_scores_gemma":[0.000018818955,0.000039779363,0.0006082726,0.000019429484,0.000020690333,0.00011064238,0.00008450648,0.81945497,0.00255164,0.17005306,0.006997537,0.0000407363],"about_ca_topic_score_codex":0.0021903692,"about_ca_topic_score_gemma":0.0028838783,"teacher_disagreement_score":0.005756711,"about_ca_system_score_codex":0.0016606364,"about_ca_system_score_gemma":0.0018443248,"threshold_uncertainty_score":0.019258082},"labels":[],"label_agreement":null},{"id":"W4200498417","doi":"10.1007/s11634-021-00487-y","title":"Correction to: Multivariate cluster weighted models using skewed distributions","year":2021,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Forecasting Techniques and Applications","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Multivariate statistics; Cluster (spacecraft); Statistics; Line (geometry); Econometrics; Computer science; Mathematics; Programming language; Geometry","score_opus":0.2059572849996396,"score_gpt":0.45221873836132187,"score_spread":0.24626145336168226,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200498417","genre_codex":"editorial","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023397985,0.0012105842,0.06668356,0.03608512,0.86844623,0.0001548264,0.011160825,0.010362994,0.0035559866],"genre_scores_gemma":[0.20277208,0.003815111,0.21977001,0.04843771,0.20120972,0.0015898123,0.025862124,0.027128449,0.2694149],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9868482,0.0031919098,0.002132882,0.0027757338,0.0039853295,0.0010660437],"domain_scores_gemma":[0.8623195,0.032219253,0.007996543,0.02804672,0.06577625,0.003641692],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.010632709,0.004390727,0.005316165,0.0064492263,0.0040091965,0.0057052146,0.006473968,0.008941461,0.21089847],"category_scores_gemma":[0.189497,0.0029133058,0.0034715747,0.008055618,0.0021755788,0.0049694125,0.0050411886,0.010565981,0.09058345],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017135979,0.000026564181,0.00034076493,0.00032156752,0.00011451254,0.00038442135,0.0000656535,0.0009477968,0.0002510303,0.0031393305,0.9798086,0.0144284135],"study_design_scores_gemma":[0.00046752187,0.00010303898,0.0033093463,0.00081481796,0.0003241507,0.002485679,0.00034751825,0.040993255,0.0034214759,0.05089583,0.8964042,0.00043316025],"about_ca_topic_score_codex":0.008127699,"about_ca_topic_score_gemma":0.013537987,"teacher_disagreement_score":0.21089847,"about_ca_system_score_codex":0.0033416427,"about_ca_system_score_gemma":0.006244827,"threshold_uncertainty_score":0.70552564},"labels":[],"label_agreement":null},{"id":"W4304806684","doi":"10.1007/s11634-022-00522-6","title":"On smoothing and scaling language model for sentiment based information retrieval","year":2022,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Sentiment Analysis and Opinion Mining","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Sentiment analysis; Latent Dirichlet allocation; Relevance (law); Social media; Smoothing; Probabilistic logic; Field (mathematics); Information retrieval; Language model; Artificial intelligence; Dirichlet distribution; Topic model; Data mining; World Wide Web; Mathematics","score_opus":0.034786673591868544,"score_gpt":0.32444836377708386,"score_spread":0.2896616901852153,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4304806684","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01557853,0.0007476521,0.9815026,0.00032932305,0.00015099566,0.00005953692,0.00013562183,0.0007500749,0.0007457561],"genre_scores_gemma":[0.4817678,0.001965324,0.50051624,0.0006039741,0.00069547794,0.0003693293,0.0015714302,0.00047366385,0.012036732],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987638,0.0004388864,0.0001227122,0.0002308648,0.0003317134,0.00011203685],"domain_scores_gemma":[0.9968335,0.0018153935,0.00015371996,0.00035743418,0.0007620864,0.00007781758],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027822529,0.00078084465,0.0014312268,0.001347361,0.0008017309,0.0012688483,0.0012898798,0.00097157055,0.0023698502],"category_scores_gemma":[0.0085250195,0.00041006773,0.0013585706,0.00180163,0.0006150722,0.0032501207,0.00093989016,0.0017338771,0.0013973202],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00056616333,0.0003990508,0.0021938377,0.00036642782,0.00028729724,0.00024299009,0.00047416388,0.27143213,0.026503608,0.09194465,0.014917541,0.5906722],"study_design_scores_gemma":[0.000008687254,0.000035127257,0.00020654454,0.0000056427134,0.000021882097,0.000022006134,0.0000146693155,0.9823437,0.0010539254,0.015276932,0.000997135,0.00001383472],"about_ca_topic_score_codex":0.0066877245,"about_ca_topic_score_gemma":0.0049644797,"teacher_disagreement_score":0.0066877245,"about_ca_system_score_codex":0.00076439796,"about_ca_system_score_gemma":0.0009994721,"threshold_uncertainty_score":0.014714122},"labels":[],"label_agreement":null},{"id":"W4309346662","doi":"10.1007/s11634-022-00526-2","title":"A dual subspace parsimonious mixture of matrix normal distributions","year":2022,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Identifiability; Subspace topology; Dual (grammatical number); Set (abstract data type); Covariance matrix; Cluster analysis; Algorithm; Principal component analysis; Computer science; Matrix (chemical analysis); Covariance; Mathematics; Column (typography); Pattern recognition (psychology); Mathematical optimization; Data mining; Applied mathematics; Artificial intelligence; Statistics","score_opus":0.023172619704976628,"score_gpt":0.3273198306380522,"score_spread":0.30414721093307556,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309346662","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004785498,0.00015103046,0.993915,0.000115973235,0.00002943961,0.000016188273,0.00007115108,0.0000954129,0.00082025863],"genre_scores_gemma":[0.20282051,0.00075890374,0.7807552,0.0003630753,0.00023353184,0.00024183262,0.0010750613,0.00030452036,0.013447298],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9972711,0.0013135052,0.00010492465,0.0004815582,0.00065093953,0.00017797141],"domain_scores_gemma":[0.99686855,0.0014623622,0.00019673965,0.0005159239,0.00068925886,0.0002671373],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0038338343,0.0008178892,0.0016127932,0.0017504469,0.00083874754,0.0022255268,0.0023987645,0.0019311926,0.0040277313],"category_scores_gemma":[0.00833473,0.0010081346,0.0016584401,0.0021262749,0.0015797602,0.0031900713,0.003495932,0.0024172463,0.0021447407],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00060505443,0.0002134747,0.0019784595,0.000262347,0.00018952905,0.00017624385,0.00036734142,0.33454603,0.0088737765,0.41035005,0.0067369663,0.23570065],"study_design_scores_gemma":[0.000016617965,0.000028722747,0.00016313861,0.000015587153,0.000018942636,0.00010515556,0.000019358236,0.93171126,0.0007364209,0.06473294,0.0024256695,0.000026153954],"about_ca_topic_score_codex":0.0025594546,"about_ca_topic_score_gemma":0.0024012101,"teacher_disagreement_score":0.0040277313,"about_ca_system_score_codex":0.00087465596,"about_ca_system_score_gemma":0.0012569746,"threshold_uncertainty_score":0.020275474},"labels":[],"label_agreement":null},{"id":"W4309636459","doi":"10.1007/s11634-022-00528-0","title":"A power-controlled reliability assessment for multi-class probabilistic classifiers","year":2022,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Advanced Statistical Methods and Models","field":"Mathematics","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Probabilistic logic; Reliability (semiconductor); Computer science; Probabilistic classification; Classifier (UML); Statistic; Bayesian probability; Artificial intelligence; Machine learning; Class (philosophy); Naive Bayes classifier; Sample size determination; Sampling distribution; Data mining; Mathematics; Statistics; Power (physics); Support vector machine","score_opus":0.16832838724178958,"score_gpt":0.48285519578488406,"score_spread":0.31452680854309445,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309636459","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009315494,0.00022014831,0.98884404,0.00008139711,0.00003336987,0.00007397034,0.000049115635,0.00026220814,0.0011202686],"genre_scores_gemma":[0.6957506,0.00042833478,0.29972947,0.0001732647,0.0002568743,0.00048799784,0.00041902738,0.00037340506,0.002381019],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9891439,0.0040480625,0.00046434547,0.0017617836,0.004250381,0.00033157057],"domain_scores_gemma":[0.9593536,0.022751052,0.002324846,0.0067043514,0.008526643,0.00033949147],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013306472,0.0014087444,0.0015913956,0.0031216461,0.0009901348,0.0023898487,0.0028155767,0.0019904692,0.0026869182],"category_scores_gemma":[0.0635392,0.0005953461,0.0014208055,0.0018271715,0.0023601267,0.0037947078,0.0024206787,0.0020566199,0.0008435481],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009073204,0.00024164942,0.0070443614,0.0006892686,0.0003633817,0.0003274956,0.0007265987,0.39475912,0.03790169,0.097262,0.0045688143,0.45520833],"study_design_scores_gemma":[0.000020268355,0.0002661149,0.0020731622,0.000054434793,0.00008198764,0.000221744,0.000044065448,0.95027655,0.008158584,0.037250284,0.0015055888,0.000047265272],"about_ca_topic_score_codex":0.0014224227,"about_ca_topic_score_gemma":0.0010430247,"teacher_disagreement_score":0.013306472,"about_ca_system_score_codex":0.0010830289,"about_ca_system_score_gemma":0.0011749986,"threshold_uncertainty_score":0.070372164},"labels":[],"label_agreement":null},{"id":"W4313595020","doi":"10.1007/s11634-022-00532-4","title":"Flexible mixture regression with the generalized hyperbolic distribution","year":2023,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Interpretability; Covariate; Component (thermodynamics); Flexibility (engineering); Generalized additive model; Mathematics; Computer science; Generalized linear model; Regression; Regression analysis; Variable (mathematics); Mixture distribution; Set (abstract data type); Mathematical optimization; Applied mathematics; Algorithm; Statistics; Artificial intelligence; Random variable","score_opus":0.039099777819178984,"score_gpt":0.3375096525388514,"score_spread":0.2984098747196724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313595020","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027757385,0.00037519512,0.99620503,0.00016031714,0.000026842654,0.000021060341,0.000045665987,0.00014063511,0.00024945303],"genre_scores_gemma":[0.23991507,0.002195682,0.74364805,0.00052267354,0.000402973,0.00051432353,0.0009715068,0.0005683806,0.011261388],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9932648,0.0039660535,0.00026676318,0.0012683753,0.000941089,0.00029286722],"domain_scores_gemma":[0.9871035,0.008858199,0.0007972166,0.0020450363,0.00092024164,0.0002758659],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014011325,0.001672509,0.0032845892,0.0020846801,0.0010574854,0.0030859036,0.005615221,0.0031620658,0.0038944338],"category_scores_gemma":[0.025535597,0.0024432614,0.0035387033,0.0034957775,0.0031035023,0.0061173863,0.005627927,0.0057748123,0.0019722052],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003952297,0.0001208348,0.002673112,0.0003367208,0.00046754768,0.00019374941,0.00039532792,0.47226006,0.0029929113,0.37977886,0.003948001,0.13643761],"study_design_scores_gemma":[0.000026894628,0.000019809533,0.00034146328,0.000028943923,0.000038481667,0.000048882277,0.0000143613615,0.913575,0.00038421564,0.084116966,0.0013663826,0.000038559654],"about_ca_topic_score_codex":0.007310715,"about_ca_topic_score_gemma":0.0066274465,"teacher_disagreement_score":0.014011325,"about_ca_system_score_codex":0.0018433812,"about_ca_system_score_gemma":0.0020023964,"threshold_uncertainty_score":0.07409984},"labels":[],"label_agreement":null},{"id":"W4376255225","doi":"10.1007/s11634-023-00542-w","title":"Model-based clustering of functional data via mixtures of t distributions","year":2023,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"MacEwan University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Outlier; Cluster analysis; Multivariate statistics; Mixture model; Computer science; Functional data analysis; Data mining; Expectation–maximization algorithm; Multivariate normal distribution; Mathematics; Pattern recognition (psychology); Artificial intelligence; Statistics; Machine learning; Maximum likelihood","score_opus":0.09709102288779793,"score_gpt":0.3592540555855143,"score_spread":0.2621630326977164,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4376255225","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002929911,0.0001342965,0.99653,0.000066754066,0.000013118232,0.000023157163,0.00003962765,0.00014908283,0.00011405109],"genre_scores_gemma":[0.21272853,0.00093885,0.7816223,0.00023806743,0.00016777281,0.00060585304,0.0012434102,0.0005149559,0.0019401844],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99565965,0.0023559774,0.00024226769,0.00095533003,0.00057609676,0.00021067863],"domain_scores_gemma":[0.990516,0.0063923365,0.0008671282,0.0011164844,0.0008677205,0.00024024965],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007277831,0.0014463051,0.003024496,0.003750999,0.0014070985,0.0034360131,0.003818878,0.0030806416,0.0017890783],"category_scores_gemma":[0.025074653,0.0014648589,0.0056141904,0.0039210855,0.0024033126,0.0040537333,0.003232171,0.0037591245,0.0013696626],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046250937,0.00018843655,0.0027851814,0.00038321217,0.0005255898,0.00012635779,0.0006939837,0.6992131,0.004411909,0.09104504,0.003627635,0.19653705],"study_design_scores_gemma":[0.000016981263,0.00002276144,0.0004020793,0.000024550667,0.000032042713,0.000054112395,0.000027473634,0.9407816,0.0005241168,0.057530418,0.000550959,0.000032991386],"about_ca_topic_score_codex":0.007539074,"about_ca_topic_score_gemma":0.0068915565,"teacher_disagreement_score":0.007539074,"about_ca_system_score_codex":0.0018350177,"about_ca_system_score_gemma":0.0016056013,"threshold_uncertainty_score":0.038489282},"labels":[],"label_agreement":null},{"id":"W4387095339","doi":"10.1007/s11634-023-00558-2","title":"Parsimony and parameter estimation for mixtures of multivariate leptokurtic-normal distributions","year":2023,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"NextGenerationEU; Ministero dell'Università e della Ricerca; Natural Sciences and Engineering Research Council of Canada; Università Cattolica del Sacro Cuore","keywords":"Kurtosis; Multivariate statistics; Statistics; Mathematics; Estimation; Multivariate normal distribution; Multivariate analysis; Estimation theory; Econometrics; Economics","score_opus":0.04852271423534097,"score_gpt":0.3659672425237264,"score_spread":0.31744452828838543,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387095339","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00693874,0.00020170482,0.9922312,0.00012110642,0.000011726049,0.00003588732,0.0000671468,0.00009918639,0.00029329158],"genre_scores_gemma":[0.25484055,0.00085639587,0.73953784,0.00019673009,0.00013567743,0.00046407236,0.0013131798,0.0003767977,0.0022787272],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.991886,0.0055574737,0.00032961424,0.0010618162,0.0008858854,0.0002791992],"domain_scores_gemma":[0.9711168,0.023686357,0.001624582,0.0020907586,0.0011125797,0.00036890322],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013682592,0.0014608762,0.0020219071,0.0042370744,0.0014202164,0.0031400556,0.0029647101,0.0024377634,0.00400633],"category_scores_gemma":[0.057829536,0.0014367877,0.0026496463,0.0033107493,0.0027995463,0.0041981493,0.0040756906,0.0038313405,0.0015058748],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043238152,0.00014113415,0.007818972,0.00043887028,0.00036211306,0.00044184036,0.0012482106,0.35890108,0.004920646,0.44375214,0.005029914,0.17651281],"study_design_scores_gemma":[0.000021123886,0.00002045699,0.000987681,0.00006138628,0.000030325735,0.00015063978,0.000047512698,0.8246418,0.0005807036,0.17177834,0.0016275662,0.000052407064],"about_ca_topic_score_codex":0.002303893,"about_ca_topic_score_gemma":0.0033964412,"teacher_disagreement_score":0.013682592,"about_ca_system_score_codex":0.0013943445,"about_ca_system_score_gemma":0.0011357378,"threshold_uncertainty_score":0.07236129},"labels":[],"label_agreement":null},{"id":"W4389888669","doi":"10.1007/s11634-023-00576-0","title":"QDA classification of high-dimensional data with rare and weak signals","year":2023,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Blind Source Separation Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Natural Sciences and Engineering Research Council of Canada; Natural Science Foundation of Shandong Province","keywords":"Quadratic classifier; Linear discriminant analysis; Mathematics; Artificial intelligence; Covariance; Pattern recognition (psychology); Covariance matrix; Gaussian; Classifier (UML); Machine learning; Statistics; Computer science","score_opus":0.05659637533914804,"score_gpt":0.33905421800217767,"score_spread":0.28245784266302965,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389888669","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.038566597,0.00059958606,0.95853853,0.00048628068,0.00019875112,0.00004037986,0.00029992583,0.00034428472,0.00092567765],"genre_scores_gemma":[0.7182433,0.000830434,0.2720972,0.00047410064,0.0006094776,0.00019472218,0.0018341236,0.00016478453,0.0055518206],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9982028,0.00052954996,0.00016398607,0.00039387355,0.00052378,0.00018593793],"domain_scores_gemma":[0.9941141,0.0031329757,0.00043018992,0.0010343652,0.0010930547,0.00019537898],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039524403,0.0006807324,0.0014449216,0.0017997023,0.0010146033,0.002220787,0.0014084687,0.001309479,0.0020483267],"category_scores_gemma":[0.011644694,0.00037624134,0.0010062136,0.0015836541,0.0012542163,0.0016318611,0.0017098531,0.0015911078,0.0010013336],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010992433,0.0004793627,0.01397172,0.00078247354,0.00034248576,0.00045096167,0.00053796626,0.14324075,0.04703628,0.06824257,0.012185648,0.71163046],"study_design_scores_gemma":[0.000019606387,0.000065688175,0.0030581304,0.00002758933,0.000036378864,0.00017759876,0.000093776485,0.95633376,0.0055754418,0.032200515,0.0023880687,0.000023473092],"about_ca_topic_score_codex":0.0014252681,"about_ca_topic_score_gemma":0.001529152,"teacher_disagreement_score":0.0039524403,"about_ca_system_score_codex":0.0006797023,"about_ca_system_score_gemma":0.0010835681,"threshold_uncertainty_score":0.020902753},"labels":[],"label_agreement":null},{"id":"W4396529244","doi":"10.1007/s11634-024-00590-w","title":"Clustering functional data via variational inference","year":2024,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Fundação de Amparo à Pesquisa do Estado de São Paulo","keywords":"Cluster analysis; Inference; Computer science; Artificial intelligence; Data mining; Machine learning","score_opus":0.08627126853721646,"score_gpt":0.3698520144562673,"score_spread":0.28358074591905086,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396529244","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013263159,0.0002049063,0.99807966,0.00013028464,0.000011441439,0.000014303585,0.00003048072,0.000073756564,0.00012887121],"genre_scores_gemma":[0.14260094,0.0015366802,0.85087585,0.00033120747,0.00026416086,0.00042051158,0.0011362587,0.000440556,0.0023938345],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99475217,0.0029806367,0.00028720128,0.00096832507,0.00083542056,0.00017618002],"domain_scores_gemma":[0.9835639,0.012803236,0.00081752206,0.0015898995,0.001011725,0.00021373166],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009425928,0.0015904943,0.0035318935,0.0052210945,0.0013891979,0.0037921725,0.0047407453,0.0030423196,0.0021550488],"category_scores_gemma":[0.027622152,0.0025105164,0.003935583,0.004050656,0.004524178,0.004793063,0.004004447,0.0054858318,0.00085760193],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013526992,0.00011369954,0.001992902,0.00046937377,0.00046211664,0.00007526291,0.00040630813,0.49304625,0.002019351,0.36153707,0.0039770156,0.13576543],"study_design_scores_gemma":[0.000010048608,0.000009144213,0.00017824734,0.00003006576,0.0000188073,0.0000252314,0.000014967983,0.85216767,0.00025643638,0.14644676,0.0008232494,0.000019402363],"about_ca_topic_score_codex":0.007764962,"about_ca_topic_score_gemma":0.00799351,"teacher_disagreement_score":0.009425928,"about_ca_system_score_codex":0.002524189,"about_ca_system_score_gemma":0.0023453527,"threshold_uncertainty_score":0.04984963},"labels":[],"label_agreement":null},{"id":"W4400011049","doi":"10.1007/s11634-024-00598-2","title":"Dirichlet compound negative multinomial mixture models and applications","year":2024,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Bayesian Methods and Mixture Models","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Multinomial distribution; Dirichlet distribution; Mathematics; Econometrics; Mathematical analysis","score_opus":0.04164186050322628,"score_gpt":0.3451560601491029,"score_spread":0.3035141996458766,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400011049","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0043916726,0.007492473,0.98377824,0.0011746769,0.0001503993,0.00002926499,0.00014978167,0.0002209858,0.0026125712],"genre_scores_gemma":[0.2586828,0.024342591,0.6881556,0.0013507755,0.0025336838,0.00076391123,0.0015688867,0.0006915852,0.021910144],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9958514,0.0022911176,0.00020847446,0.0006844956,0.0007902162,0.00017436745],"domain_scores_gemma":[0.9820047,0.014757262,0.0007018113,0.0011064991,0.0011366389,0.00029303823],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008056026,0.0016413017,0.0039004635,0.0034689629,0.0015199495,0.0039658756,0.0038691247,0.0036135232,0.0046678535],"category_scores_gemma":[0.02936632,0.0019301753,0.00290566,0.005582453,0.0035448691,0.0055570956,0.0038852643,0.0061094277,0.0017121979],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000099775236,0.00010061208,0.0013415327,0.00032737732,0.00014641149,0.00014994902,0.00041441398,0.079523265,0.0006883218,0.8297479,0.006145345,0.0813151],"study_design_scores_gemma":[0.000013073577,0.000010701323,0.00028256775,0.00006457775,0.000032724973,0.00015660765,0.00003861,0.39701307,0.00021067413,0.5969927,0.005146412,0.000038332702],"about_ca_topic_score_codex":0.0053360583,"about_ca_topic_score_gemma":0.0052120597,"teacher_disagreement_score":0.008056026,"about_ca_system_score_codex":0.0024278944,"about_ca_system_score_gemma":0.001374974,"threshold_uncertainty_score":0.042604864},"labels":[],"label_agreement":null},{"id":"W4403793086","doi":"10.1007/s11634-024-00608-3","title":"A sparse exponential family latent block model for co-clustering","year":2024,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Amirkabir University of Technology","keywords":"Exponential family; Cluster analysis; Exponential function; Block (permutation group theory); Computer science; Mathematics; Statistics; Combinatorics","score_opus":0.1106521907296518,"score_gpt":0.39391825692417415,"score_spread":0.2832660661945223,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403793086","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003486987,0.0002607537,0.9950067,0.00018511293,0.00004285424,0.000038037353,0.00031447987,0.00018685966,0.00047825623],"genre_scores_gemma":[0.40751758,0.0020517693,0.5575198,0.0006490221,0.00044479262,0.0012419606,0.0056081284,0.0006342183,0.024332674],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9969138,0.0015458305,0.0001262636,0.00063965836,0.0004936605,0.00028076835],"domain_scores_gemma":[0.9931178,0.003804082,0.00047755643,0.0013037693,0.0010007496,0.00029604993],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0047492348,0.0010308773,0.002345404,0.0015780224,0.0012815697,0.0019984818,0.005507947,0.0032469048,0.0057915724],"category_scores_gemma":[0.013773707,0.0012311883,0.0021023152,0.0037997242,0.0018565005,0.004344972,0.0029428154,0.004161143,0.004105047],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036498514,0.00018687542,0.0019324301,0.00021772872,0.00019895591,0.00014403477,0.00037618013,0.6011391,0.0021991143,0.29669362,0.0110188965,0.08552806],"study_design_scores_gemma":[0.000011023415,0.000013577731,0.00014235821,0.000010258249,0.000014354157,0.000027311668,0.000014410435,0.9650204,0.00013621249,0.033500988,0.0010952312,0.000013830325],"about_ca_topic_score_codex":0.013713874,"about_ca_topic_score_gemma":0.01758404,"teacher_disagreement_score":0.013713874,"about_ca_system_score_codex":0.0018239425,"about_ca_system_score_gemma":0.0024128053,"threshold_uncertainty_score":0.027268112},"labels":[],"label_agreement":null},{"id":"W4407498245","doi":"10.1007/s11634-025-00625-w","title":"Random models for adjusting fuzzy rand index extensions","year":2025,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Rough Sets and Fuzzy Logic","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Okanagan University College; University of British Columbia, Okanagan Campus; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Index (typography); Fuzzy logic; Computer science; Econometrics; Mathematics; Statistics; Artificial intelligence; World Wide Web","score_opus":0.052200240792401084,"score_gpt":0.3365206615945156,"score_spread":0.2843204208021145,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407498245","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007361723,0.00034483828,0.99024993,0.00020080776,0.00008122703,0.00006050643,0.000082591985,0.00028516457,0.0013331756],"genre_scores_gemma":[0.42491975,0.00087636017,0.5561008,0.00050822383,0.00036769648,0.0007363635,0.00090568344,0.0007098223,0.0148753],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98688865,0.008756097,0.0003474732,0.002172385,0.0013993345,0.00043599322],"domain_scores_gemma":[0.9557358,0.031211276,0.0026045751,0.0076897275,0.0022892964,0.00046925867],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.027095186,0.0015157573,0.002426815,0.0029862272,0.00093983463,0.0033286312,0.005067119,0.002484437,0.0067518423],"category_scores_gemma":[0.089416735,0.0012756396,0.0026691915,0.0027689082,0.0021047716,0.0061658644,0.002755288,0.0042348597,0.0015555078],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033649392,0.00014726879,0.0020428977,0.00016432749,0.00036809806,0.0001045462,0.0002288476,0.5566649,0.00058518234,0.33804426,0.004451964,0.0968613],"study_design_scores_gemma":[0.00002639652,0.00004741302,0.00039166363,0.00003245072,0.00004571984,0.00003590464,0.00001725321,0.87434584,0.00021733393,0.123176634,0.0016304026,0.000032858454],"about_ca_topic_score_codex":0.0029508772,"about_ca_topic_score_gemma":0.0035237935,"teacher_disagreement_score":0.027095186,"about_ca_system_score_codex":0.0020923156,"about_ca_system_score_gemma":0.0012958192,"threshold_uncertainty_score":0.14329475},"labels":[],"label_agreement":null},{"id":"W4410920770","doi":"10.1007/s11634-025-00637-6","title":"Mixed-type kernel-based quantification of similarity for clustering","year":2025,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Advanced Clustering Algorithms Research","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Okanagan University College; University of British Columbia, Okanagan Campus","funders":"University of British Columbia","keywords":"Kernel (algebra); Similarity (geometry); Cluster analysis; Artificial intelligence; Computer science; Pattern recognition (psychology); Type (biology); Mathematics; Data mining; Combinatorics; Geology","score_opus":0.09468252969729223,"score_gpt":0.3986018150346892,"score_spread":0.303919285337397,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410920770","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003950176,0.00019701543,0.9949944,0.000045258566,0.0000267404,0.000023508986,0.000048018603,0.00018237483,0.0005325302],"genre_scores_gemma":[0.24529389,0.00040764015,0.7498999,0.00012217174,0.00012909004,0.00021325203,0.0003725477,0.00031526873,0.0032461635],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99498993,0.0017941638,0.00041683327,0.00083682843,0.0017287189,0.00023346931],"domain_scores_gemma":[0.99213105,0.0024299582,0.00072956784,0.002188446,0.0022030212,0.0003178754],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0041791056,0.0008677308,0.0013969677,0.0032772245,0.0008860683,0.0034340057,0.0027566063,0.0017247489,0.0024336723],"category_scores_gemma":[0.016226893,0.00055075885,0.0014414595,0.0037125049,0.0017433023,0.006097429,0.0034153748,0.002599036,0.0013961683],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00061741634,0.00022472035,0.0031179953,0.000748225,0.00037476904,0.00013292013,0.00057818706,0.09166138,0.050273433,0.51747686,0.00507403,0.32971996],"study_design_scores_gemma":[0.000016884942,0.000118852135,0.0013261051,0.000041737378,0.000053755655,0.00033311202,0.00009355048,0.87022537,0.0135590695,0.109232105,0.004924844,0.00007460771],"about_ca_topic_score_codex":0.0010051233,"about_ca_topic_score_gemma":0.0009494374,"teacher_disagreement_score":0.0041791056,"about_ca_system_score_codex":0.0015045925,"about_ca_system_score_gemma":0.0011524975,"threshold_uncertainty_score":0.022101462},"labels":[],"label_agreement":null},{"id":"W4416664067","doi":"10.1007/s11634-025-00659-0","title":"Data-driven logistic regression ensembles with applications in genomics","year":2025,"lang":"en","type":"article","venue":"Advances in Data Analysis and Classification","topic":"Gene expression and cancer classification","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Mitacs; KU Leuven","keywords":"Genomics; Regularization (linguistics); Ranking (information retrieval); Big data; Logistic regression; Computational genomics; Predictive modelling; Binary number","score_opus":0.04371606132484556,"score_gpt":0.35820006926448344,"score_spread":0.3144840079396379,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416664067","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0157878,0.00092576497,0.9807741,0.00042874104,0.00012708822,0.000033139007,0.0002626169,0.0012078385,0.00045292653],"genre_scores_gemma":[0.45264876,0.0012294432,0.5394014,0.0004300185,0.0005364518,0.00033038933,0.0017336349,0.0004433919,0.003246481],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99779403,0.0012760084,0.000110329696,0.0003004904,0.00043209808,0.0000869552],"domain_scores_gemma":[0.9883874,0.008677856,0.00033782437,0.0008785953,0.001488781,0.00022947337],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006074179,0.0009649617,0.00194217,0.0009789836,0.0005485788,0.0009754499,0.0014436981,0.0011395328,0.0013712745],"category_scores_gemma":[0.01949812,0.0007420417,0.0012223696,0.0016338737,0.0004483767,0.0012095666,0.0020018076,0.002961843,0.00086834247],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024625545,0.00017866617,0.0039511104,0.00011441845,0.00031926626,0.00009720036,0.00009593565,0.69403076,0.002481186,0.009210921,0.0055811433,0.2836932],"study_design_scores_gemma":[0.0000064318283,0.0000116861665,0.000182009,0.0000058422675,0.000011094513,0.000009505055,0.0000039324514,0.99347407,0.0003873272,0.0054689157,0.0004329537,0.000006238212],"about_ca_topic_score_codex":0.0033175144,"about_ca_topic_score_gemma":0.0043266616,"teacher_disagreement_score":0.006074179,"about_ca_system_score_codex":0.00050737365,"about_ca_system_score_gemma":0.00090168376,"threshold_uncertainty_score":0.032123685},"labels":[],"label_agreement":null}]}