{"id":"W4306317280","doi":"10.1145/3511808.3557673","title":"Probing the Robustness of Pre-trained Language Models for Entity Matching","year":2022,"lang":"en","type":"article","venue":"Proceedings of the 31st ACM International Conference on Information & Knowledge Management","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":15,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta","funders":"","keywords":"Robustness (evolution); Computer science; Spurious relationship; Machine learning; Software deployment; Artificial intelligence; Training set; Data modeling; Data mining; Database; Software engineering","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":[0.004558297,0.0001616544,0.0002152227,0.0004092506,0.0004157288,0.0004125845,0.00448108,0.00002724145,0.0003302906],"category_scores_gemma":[0.0008270525,0.0001056414,0.0001783285,0.0005222901,0.00009992437,0.001518978,0.003248688,0.0001778895,0.00002691776],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001404291,"about_ca_system_score_gemma":0.000038242,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002124748,"about_ca_topic_score_gemma":0.00002976576,"domain_scores_codex":[0.9968696,0.00004670431,0.001006731,0.0002427252,0.001643277,0.0001909966],"domain_scores_gemma":[0.9970588,0.0002315043,0.001132341,0.0005146904,0.001030529,0.00003209771],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0002042132,0.0001653268,0.00002831661,0.0002484307,0.0001226124,2.42615e-8,0.01424532,0.01817952,0.0001120268,0.9337533,0.01763782,0.01530316],"study_design_scores_gemma":[0.002411914,0.0001618744,0.001303548,0.0003902251,0.0001566022,0.000005444537,0.1205307,0.1764683,0.0008562396,0.4066288,0.2904783,0.0006079737],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.2262121,0.00003495954,0.1989497,0.01452577,0.00384444,0.005412468,0.0008386213,0.0001397303,0.5500422],"genre_scores_gemma":[0.9870851,0.0000106365,0.003072685,0.0002488031,0.00005109036,0.0004601219,0.0000761438,0.000007960501,0.008987467],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.760873,"threshold_uncertainty_score":0.8327037,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1525473395701562,"score_gpt":0.3846648573911663,"score_spread":0.2321175178210101,"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."}}