{"id":"W4405971183","doi":"10.1038/s44159-024-00392-z","title":"Using natural language processing to analyse text data in behavioural science","year":2025,"lang":"en","type":"article","venue":"Nature Reviews Psychology","topic":"Mental Health via Writing","field":"Psychology","cited_by":106,"is_retracted":false,"has_abstract":false,"ca_institutions":"National Research Council Canada","funders":"","keywords":"Rigour; Computer science; Popularity; Interpretability; Artificial intelligence; Natural language processing; Social media; Computational sociology; Data science; Pipeline (software); Psychology; 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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003398763,0.0002765642,0.0006470281,0.001068412,0.0002125136,0.00005392326,0.001820878,0.0003686594,0.0002101579],"category_scores_gemma":[0.0005799993,0.0002451108,0.00006212135,0.004179028,0.0002169472,0.0002990529,0.0004719958,0.001675209,0.000173708],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002786808,"about_ca_system_score_gemma":0.0001812618,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002368319,"about_ca_topic_score_gemma":0.0004755823,"domain_scores_codex":[0.9961399,0.0004259974,0.0009137388,0.001359333,0.0002654626,0.0008955336],"domain_scores_gemma":[0.997635,0.00007650456,0.000253876,0.001788392,0.00008348128,0.0001627565],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.00009572213,0.0002322075,0.0552092,0.0002613022,0.0000113389,0.0001119042,0.001329883,4.930357e-7,0.006763731,0.0005570138,0.005160144,0.930267],"study_design_scores_gemma":[0.004010462,0.0002139935,0.8046465,0.004921011,0.000258648,0.0007687302,0.003706638,0.001656901,0.000314182,0.0001974411,0.177735,0.001570509],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6087409,0.357457,0.000435079,0.002735361,0.004542089,0.001989197,0.00002985206,0.0000968752,0.02397371],"genre_scores_gemma":[0.9726622,0.0001689458,0.01174309,0.01476337,0.0001484087,0.00005496912,0.00004068467,0.00002267414,0.0003956635],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9286966,"threshold_uncertainty_score":0.999533,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1702667086601565,"score_gpt":0.5825904396932856,"score_spread":0.4123237310331291,"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."}}