{"paper":{"title":"Deep Interest Mining for Intent-Enriched Semantic IDs in Multimodal Generative Recommendation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A framework with deep interest mining, cross-modal alignment, and quality-aware reinforcement generates higher-quality Semantic IDs for generative recommendation.","cross_cats":["cs.AI"],"primary_cat":"cs.IR","authors_text":"Jinze Wang, Yangchen Zeng","submitted_at":"2026-03-03T13:36:22Z","abstract_excerpt":"Semantic IDs (SIDs) provide the discrete item vocabulary used by generative recommendation, but their quality depends on what item evidence is preserved before quantization. In product recommendation, surface metadata often misses latent usage intent, visual evidence may be only weakly reflected in text, and downstream policy learning provides sparse feedback about whether a generated SID corresponds to a semantically useful item. We introduce \\textbf{DeepInterestGR}, an intent-enriched SID framework for generative recommendation. Before SID quantization, \\textbf{CMSA} enriches item representa"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Extensive experiments demonstrate that our approach consistently outperforms state-of-the-art SID generation methods, achieving superior performance on multiple benchmarks. Ablation studies further validate the effectiveness of each proposed component.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The quality-aware reinforcement mechanism combined with deep interest mining and cross-modal alignment can jointly optimize Semantic ID generation to preserve semantics and distinguish high-quality from low-quality IDs without introducing training instability or new biases.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A new framework integrating deep interest mining, cross-modal semantic alignment, and quality-aware reinforcement learning generates higher-quality Semantic IDs and outperforms prior methods on recommendation benchmarks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A framework with deep interest mining, cross-modal alignment, and quality-aware reinforcement generates higher-quality Semantic IDs for generative recommendation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"342359216830f915d49c947e829b74f237bba8e9b9b3918e19db8f4bd52eede6"},"source":{"id":"2604.20861","kind":"arxiv","version":3},"verdict":{"id":"affc31ff-c50a-4aef-a6d0-89cd973e971f","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T16:57:28.161124Z","strongest_claim":"Extensive experiments demonstrate that our approach consistently outperforms state-of-the-art SID generation methods, achieving superior performance on multiple benchmarks. Ablation studies further validate the effectiveness of each proposed component.","one_line_summary":"A new framework integrating deep interest mining, cross-modal semantic alignment, and quality-aware reinforcement learning generates higher-quality Semantic IDs and outperforms prior methods on recommendation benchmarks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The quality-aware reinforcement mechanism combined with deep interest mining and cross-modal alignment can jointly optimize Semantic ID generation to preserve semantics and distinguish high-quality from low-quality IDs without introducing training instability or new biases.","pith_extraction_headline":"A framework with deep interest mining, cross-modal alignment, and quality-aware reinforcement generates higher-quality Semantic IDs for generative recommendation."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.20861/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}