{"id":"W4385573948","doi":"10.18653/v1/2022.findings-emnlp.440","title":"WordTies: Measuring Word Associations in Language Models via Constrained Sampling","year":2022,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Word (group theory); Natural language processing; Associative property; Artificial intelligence; Similarity (geometry); Proxy (statistics); Word Association; Language model; Property (philosophy); Cosine similarity; Machine learning; Linguistics; Mathematics; Pattern recognition (psychology)","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006600353,0.001412856,0.001364796,0.003352854,0.0009968394,0.002758773,0.001957794,0.00183261,0.00263996],"category_scores_gemma":[0.04531308,0.0007118505,0.001334886,0.003188516,0.001771875,0.006361578,0.003650138,0.002495578,0.0008532148],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009286565,"about_ca_system_score_gemma":0.001316422,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003332983,"about_ca_topic_score_gemma":0.005141667,"domain_scores_codex":[0.9949552,0.002525347,0.0003541838,0.00116093,0.0008272799,0.0001769843],"domain_scores_gemma":[0.9729536,0.02144342,0.001799805,0.002411443,0.0009570807,0.0004345776],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001457944,0.000640176,0.06616439,0.001035136,0.001468593,0.000441878,0.003875933,0.2326994,0.01771799,0.09118748,0.008233514,0.5750777],"study_design_scores_gemma":[0.00008487708,0.0002072408,0.005297199,0.00006165518,0.00007772138,0.000203909,0.0003638223,0.8606651,0.003842871,0.1267809,0.002348956,0.00006580934],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06677966,0.0003595331,0.9301524,0.0001722632,0.00003491545,0.0001827545,0.0004760851,0.001113023,0.0007293485],"genre_scores_gemma":[0.5819527,0.000365469,0.4119402,0.0002016142,0.0001227111,0.0007212397,0.003229207,0.0004479391,0.001018905],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006600353,"threshold_uncertainty_score":0.03490639,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07116683802841292,"score_gpt":0.269134695829356,"score_spread":0.197967857800943,"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."}}