{"id":"W4415159347","doi":"10.48550/arxiv.2504.10004","title":"An Image is Worth $K$ Topics: A Visual Structural Topic Model with Pretrained Image Embeddings","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Freie Universität Berlin; Forskningsrådet om Hälsa, Arbetsliv och Välfärd; Vetenskapsrådet; York University","keywords":"Representation (politics); Toolbox; Image (mathematics); Topic model; Politics; Computational model; Sketch","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":[0.0010801,0.0006840494,0.0005467753,0.001022818,0.0003181697,0.001382522,0.001216986,0.001403129,0.003137755],"category_scores_gemma":[0.004642949,0.0004879932,0.001086278,0.001014204,0.0009039009,0.002924294,0.001186798,0.002095135,0.001083098],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008353091,"about_ca_system_score_gemma":0.0004127743,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004339851,"about_ca_topic_score_gemma":0.005658532,"domain_scores_codex":[0.9995864,0.0001299917,0.00001464281,0.0001744815,0.00004054129,0.00005393598],"domain_scores_gemma":[0.9986477,0.0008585387,0.000142729,0.000165994,0.000118274,0.0000667463],"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":[0.001326056,0.0005167686,0.01528991,0.0005985561,0.0003819763,0.0004198735,0.001944271,0.2814648,0.01767103,0.1058251,0.03034162,0.54422],"study_design_scores_gemma":[0.00002497181,0.00005661137,0.002108063,0.00003381683,0.00004133182,0.000112233,0.0001011928,0.9528918,0.001051166,0.04093023,0.002628514,0.00002001602],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.2221649,0.002227454,0.7614032,0.00323389,0.000342561,0.0001686047,0.002246808,0.0014455,0.006767127],"genre_scores_gemma":[0.9250248,0.0008194905,0.06390324,0.0004342672,0.0003602098,0.000211969,0.002243567,0.0002285231,0.006773892],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004339851,"threshold_uncertainty_score":0.01049685,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03992358868207933,"score_gpt":0.4058565986922922,"score_spread":0.3659330100102128,"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."}}