English

Personalized Showcases: Generating Multi-Modal Explanations for Recommendations

Information Retrieval 2023-04-07 v2 Artificial Intelligence Computer Vision and Pattern Recognition

Abstract

Existing explanation models generate only text for recommendations but still struggle to produce diverse contents. In this paper, to further enrich explanations, we propose a new task named personalized showcases, in which we provide both textual and visual information to explain our recommendations. Specifically, we first select a personalized image set that is the most relevant to a user's interest toward a recommended item. Then, natural language explanations are generated accordingly given our selected images. For this new task, we collect a large-scale dataset from Google Local (i.e.,~maps) and construct a high-quality subset for generating multi-modal explanations. We propose a personalized multi-modal framework which can generate diverse and visually-aligned explanations via contrastive learning. Experiments show that our framework benefits from different modalities as inputs, and is able to produce more diverse and expressive explanations compared to previous methods on a variety of evaluation metrics.

Keywords

Cite

@article{arxiv.2207.00422,
  title  = {Personalized Showcases: Generating Multi-Modal Explanations for Recommendations},
  author = {An Yan and Zhankui He and Jiacheng Li and Tianyang Zhang and Julian McAuley},
  journal= {arXiv preprint arXiv:2207.00422},
  year   = {2023}
}

Comments

Accepted to SIGIR-23, with additional dataset details. Code and data: https://github.com/zzxslp/Gest

R2 v1 2026-06-24T12:11:07.737Z