Balancing User-Item Structure and Interaction with Large Language Models and Optimal Transport for Multimedia Recommendation

Li, Haodong; Qi, Lianyong; Liu, Weiming; Xu, Xiaolong; Dou, Wanchun; Cao, Yang; Zhang, Xuyun; Beheshti, Amin; Zhou, Xiaokang

doi:10.24963/ijcai.2025/337

Balancing User-Item Structure and Interaction with Large Language Models and Optimal Transport for Multimedia Recommendation

Haodong Li, Lianyong Qi, Weiming Liu, Xiaolong Xu, Wanchun Dou, Yang Cao, Xuyun Zhang, Amin Beheshti, Xiaokang Zhou

Download BibTeX Cite

Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence. Main Track. Pages 3027-3035.

https://doi.org/10.24963/ijcai.2025/337

Keywords

Data Mining: DM: Collaborative filtering
Data Mining: DM: Recommender systems
Computer Vision: CV: Multimodal learning
Natural Language Processing: NLP: Language models

Abstract

The rapid growth of multimedia content has driven the development of recommender systems. Most previous work focuses on uncovering latent relationships among items to learn better representations. However, this approach does not sufficiently account for user affinities, potentially leading to an imbalance in the structure modeling of users and items. Moreover, the sparsity and imbalance of user-item interactions further hinder effective representation learning. To address these challenges, we propose a framework called BLAST, which balances structures and interactions via large language models and optimal transport for multimodal recommendation. Specifically, we utilize large language models to summarize side information and generate user profiles. Based on these profiles, we design an intra- and inter-entity structure balancing module to capture item-item and user-user relationships, integrating these affinities into the final representations. Furthermore, we impose constraints on negative sample selection, augment the training data with false negative items and the optimal transport algorithm, thereby leading to smoother interactions. We evaluate BLAST on three real-world datasets, and the results demonstrate that our method significantly outperforms state-of-the-art baselines, which validates the superiority and effectiveness of BLAST.

Haodong Li, Lianyong Qi, Weiming Liu, Xiaolong Xu, Wanchun Dou, Yang Cao, Xuyun Zhang, Amin Beheshti, and Xiaokang Zhou. 2025. Balancing User-Item Structure and Interaction with Large Language Models and Optimal Transport for Multimedia Recommendation. In Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence, IJCAI-25. IJCAI, 3027–3035. https://doi.org/10.24963/ijcai.2025/337

@inproceedings{ijcai2025p337,
  title     = {Balancing User-Item Structure and Interaction with Large Language Models and Optimal Transport for Multimedia Recommendation},
  author    = {Li, Haodong and Qi, Lianyong and Liu, Weiming and Xu, Xiaolong and Dou, Wanchun and Cao, Yang and Zhang, Xuyun and Beheshti, Amin and Zhou, Xiaokang},
  booktitle = {Proceedings of the Thirty-Fourth International Joint Conference on
               Artificial Intelligence, {IJCAI-25}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  editor    = {James Kwok},
  pages     = {3027--3035},
  year      = {2025},
  month     = {8},
  note      = {Main Track},
  doi       = {10.24963/ijcai.2025/337},
  url       = {https://doi.org/10.24963/ijcai.2025/337},
}

TY  - CPAPER
TI  - Balancing User-Item Structure and Interaction with Large Language Models and Optimal Transport for Multimedia Recommendation
AU  - Li, Haodong
AU  - Qi, Lianyong
AU  - Liu, Weiming
AU  - Xu, Xiaolong
AU  - Dou, Wanchun
AU  - Cao, Yang
AU  - Zhang, Xuyun
AU  - Beheshti, Amin
AU  - Zhou, Xiaokang
T2  - Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence, IJCAI-25
PB  - International Joint Conferences on Artificial Intelligence Organization
PY  - 2025
SP  - 3027
EP  - 3035
DO  - 10.24963/ijcai.2025/337
UR  - https://doi.org/10.24963/ijcai.2025/337
ER  -