{"id":"W4409366480","doi":"10.1609/aaai.v39i7.32768","title":"Leveraging Large Vision-Language Model as User Intent-Aware Encoder for Composed Image Retrieval","year":2025,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Image Retrieval and Classification Techniques","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Institute of Aging","funders":"National Natural Science Foundation of China","keywords":"Computer science; Encoder; Computer vision; Image (mathematics); Artificial intelligence; Image retrieval; Information retrieval; Language model; Natural language processing","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":[0.0009965892,0.000901169,0.0008172982,0.0005933636,0.0002232911,0.0008505888,0.001946835,0.0009673389,0.002380752],"category_scores_gemma":[0.003907625,0.0003361284,0.0007488328,0.0005012988,0.0005725545,0.002548513,0.001225722,0.002068658,0.001712709],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008164949,"about_ca_system_score_gemma":0.001323303,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003974327,"about_ca_topic_score_gemma":0.006497655,"domain_scores_codex":[0.9994647,0.0001338088,0.00003141073,0.0001786281,0.000132279,0.00005918695],"domain_scores_gemma":[0.9989547,0.0005001894,0.0000759495,0.0002084988,0.0001988411,0.00006180601],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004699819,0.0004735103,0.001871487,0.0004087454,0.0001042314,0.000265403,0.0003254279,0.1180331,0.05379134,0.01669554,0.01721349,0.7903478],"study_design_scores_gemma":[0.00002588678,0.0001406821,0.0002577242,0.00001741628,0.00003081361,0.00008284884,0.00003600868,0.9721327,0.01496594,0.009750679,0.002535376,0.00002390789],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02253155,0.001000718,0.9647029,0.000471395,0.00008378574,0.0001232077,0.0003658701,0.008673193,0.002047476],"genre_scores_gemma":[0.4796561,0.0008059886,0.5067603,0.001173818,0.0001652202,0.0003108013,0.0026115,0.0004930235,0.008023209],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003974327,"threshold_uncertainty_score":0.007964432,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04652793084147333,"score_gpt":0.3424371734148765,"score_spread":0.2959092425734031,"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."}}