{"id":"W7160410202","doi":"10.1145/3786995.3786997","title":"From Transformers Come Themes: Evaluating BERTopic for Qualitative Analysis of Social Media Data","year":2025,"lang":"","type":"article","venue":"","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Latent Dirichlet allocation; Qualitative analysis; Social media; Qualitative research; Qualitative property; Sentence; Relevance (law); Topic model","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":[],"consensus_categories":[],"category_scores_codex":[0.102068,0.0008577229,0.0006657055,0.002921671,0.002242224,0.004307404,0.001880902,0.001042587,0.004589905],"category_scores_gemma":[0.2776347,0.0005350707,0.001133031,0.002817991,0.004265042,0.005171942,0.005893752,0.001409754,0.0006801793],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004271768,"about_ca_system_score_gemma":0.003518286,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001769423,"about_ca_topic_score_gemma":0.005352801,"domain_scores_codex":[0.9181718,0.0718229,0.002553915,0.002722252,0.003877996,0.0008511194],"domain_scores_gemma":[0.642356,0.3112787,0.01150927,0.01243249,0.02017035,0.002253205],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.002148489,0.0006271568,0.02710652,0.006231008,0.0002195837,0.0006133349,0.5672062,0.007244334,0.01831276,0.03537362,0.004234952,0.330682],"study_design_scores_gemma":[0.0009729916,0.003656216,0.05066293,0.004837684,0.0005513826,0.0009545658,0.5538917,0.1441253,0.03907469,0.1320769,0.06856719,0.0006285063],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5569984,0.0004488364,0.4157264,0.001777486,0.0001481847,0.01188974,0.001917983,0.0007492219,0.01034384],"genre_scores_gemma":[0.5611892,0.0001882149,0.4126302,0.0003357425,0.00003159451,0.02373164,0.0007909199,0.0002355944,0.0008670263],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.102068,"threshold_uncertainty_score":0.5397938,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3151434078999855,"score_gpt":0.5883443346130658,"score_spread":0.2732009267130803,"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."}}