{"id":"W4225634029","doi":"10.2139/ssrn.4066511","title":"Scholar Metrics Scraper (Sms): Automated Retrieval of Citation and Author Data","year":2022,"lang":"en","type":"article","venue":"SSRN Electronic Journal","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Scraper site; Citation; Information retrieval; Computer science; Data mining; World Wide Web","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.003147647,0.00153333,0.001564106,0.0154498,0.0007831018,0.002473044,0.001368673,0.001008515,0.02357242],"category_scores_gemma":[0.01986961,0.0007400872,0.000793593,0.01121077,0.0002394,0.002584949,0.002877913,0.0007905486,0.02796457],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005762678,"about_ca_system_score_gemma":0.002684707,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005429093,"about_ca_topic_score_gemma":0.01137548,"domain_scores_codex":[0.9964589,0.0007671154,0.0004798879,0.000439524,0.001601168,0.0002534242],"domain_scores_gemma":[0.9870832,0.004337128,0.002129499,0.002014864,0.003397149,0.00103825],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001326005,0.0005040807,0.01115889,0.001558077,0.0004215722,0.0003521044,0.0008215777,0.0008362715,0.01922597,0.001863196,0.4286516,0.5332806],"study_design_scores_gemma":[0.00230471,0.001866207,0.1111887,0.0006822699,0.0008345545,0.00158678,0.002054087,0.1159077,0.1287497,0.01249796,0.6216175,0.0007097769],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"software","genre_gemma":"software","genre_scores_codex":[0.1454316,0.003325795,0.1127811,0.002381123,0.001321676,0.002482311,0.2321313,0.4678042,0.03234099],"genre_scores_gemma":[0.2717367,0.002018081,0.3333549,0.001362404,0.001479672,0.00229397,0.3041259,0.01784777,0.06578054],"genre_candidate":"software","genre_consensus":"software","teacher_disagreement_score":0.9968523,"threshold_uncertainty_score":0.0788576,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.444686786635901,"score_gpt":0.5439794446972804,"score_spread":0.09929265806137944,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}