{"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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0005003148,0.0003178634,0.0004723261,0.0001801768,0.0004668006,0.0002796528,0.0006464111,0.000248184,0.0004814222],"category_scores_gemma":[0.0001249914,0.0002771787,0.0002398093,0.0003961224,0.0002922697,0.0003355709,0.0003020794,0.0005465125,0.00001192415],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001470191,"about_ca_system_score_gemma":0.0007133228,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001629169,"about_ca_topic_score_gemma":0.001084161,"domain_scores_codex":[0.9976378,0.00028969,0.0003655021,0.0007580692,0.0005488393,0.0004000786],"domain_scores_gemma":[0.9986923,0.0001145497,0.0002257608,0.0003982248,0.0003893099,0.0001798342],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002684609,0.0004050228,0.7475275,0.0006175057,0.001501022,0.0001086157,0.1206964,0.01220022,0.0009942203,0.02494142,0.00164748,0.0890921],"study_design_scores_gemma":[0.001193996,0.0002010669,0.3381127,0.0003581365,0.0008048765,0.000002630364,0.004567609,0.5524861,0.0007162297,0.09618866,0.003325297,0.002042667],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9463547,0.00005673605,0.03944914,0.002087524,0.0002673691,0.0003431571,0.00004104687,0.0001552497,0.01124507],"genre_scores_gemma":[0.9205205,0.00002404485,0.0639513,0.0005754277,0.0007040751,0.00003688596,0.00008252986,0.0000175647,0.0140877],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.5402859,"threshold_uncertainty_score":0.9999681,"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."}}