{"id":"W2532655604","doi":"10.2196/publichealth.6327","title":"“When ‘Bad’ is ‘Good’”: Identifying Personal Communication and Sentiment in Drug-Related Tweets","year":2016,"lang":"en","type":"article","venue":"JMIR Public Health and Surveillance","topic":"Mental Health via Writing","field":"Psychology","cited_by":89,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"National Institute on Drug Abuse; National Institute of Mental Health","keywords":"Computer science; Sentiment analysis; Artificial intelligence; Naive Bayes classifier; Support vector machine; Machine learning; Classifier (UML); Social media; Binary classification; Coding (social sciences); Natural language processing; World Wide Web","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002694544,0.0001686491,0.0003206891,0.000181172,0.0002796804,0.0000620493,0.0001501911,0.0001131866,0.0005253445],"category_scores_gemma":[0.00005414612,0.0001441715,0.00002653151,0.0002055382,0.0001200609,0.0002100078,0.0001197298,0.000245916,0.00007717674],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002048807,"about_ca_system_score_gemma":0.0001547204,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001034405,"about_ca_topic_score_gemma":0.0006216635,"domain_scores_codex":[0.9969718,0.0009135642,0.0006552615,0.0004715946,0.0002268821,0.000760869],"domain_scores_gemma":[0.9984126,0.0003917907,0.000256794,0.0003446633,0.00004905476,0.0005450919],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00003047412,0.0001052586,0.8574719,0.0001816129,0.00001565344,0.000004441417,0.02037855,5.520809e-9,0.00001982107,0.001637359,0.00707749,0.1130775],"study_design_scores_gemma":[0.002787646,0.00008843995,0.9286256,0.0001699129,3.88728e-7,0.00005407356,0.003458053,0.00008354921,0.000001680724,0.0003865464,0.06409623,0.000247844],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.932512,0.005724424,0.00003641379,0.0541972,0.0002117636,0.0006998453,0.00003086987,0.00007162802,0.006515826],"genre_scores_gemma":[0.9928366,0.0006562939,0.0001337503,0.003508948,0.00003831504,0.0001435291,0.00002968416,0.00002221718,0.002630675],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1128296,"threshold_uncertainty_score":0.5879142,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05157809221172287,"score_gpt":0.37456530973284,"score_spread":0.3229872175211171,"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."}}