Behavior-Guided Candidate Calibration for Multimodal Recommendation CCS: Information systems Recommender systems CCS: Information systems Multimedia and multimodal retrieval
Zesheng Li, Chengchang Pan, Honggang Qi · 2026
Read Behavior-Guided Candidate Calibration for Multimodal Recommendation CCS: Information systems Recommender systems CCS: Information systems Multimedia and multimodal retrieval by Zesheng Li, Chengchang Pan, Honggang Qi (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-22073.txt.
Description
Zesheng Li, Chengchang Pan, and Honggang Qi present BRIDGE, a behavior-guided method for multimodal recommendation. More cross-view alignment does not always lift ranking: tests on Amazon Baby show moderate alignment helps, while stronger alignment suppresses task-specific variation, and low-frequency parts capture shared structure. BRIDGE keeps the backbone and puts behavior evidence into candidate calibration, keeping the base top-K candidate set in training and inference. On Amazon Baby, Sports, and Electronics it beats baselines by up to 7.3 percent in Recall at 20 and 14.9 percent in NDCG at 20.
Catalog record
| CREATOR | Zesheng Li, Chengchang Pan, Honggang Qi |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56828 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.22073