Generate, Not Recommend: Personalized Multimodal Content Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Jiongnan, Dou, Zhicheng, Hu, Ning, Xiong, Chenyan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915319822417920
author Liu, Jiongnan
Dou, Zhicheng
Hu, Ning
Xiong, Chenyan
author_facet Liu, Jiongnan
Dou, Zhicheng
Hu, Ning
Xiong, Chenyan
contents To address the challenge of information overload from massive web contents, recommender systems are widely applied to retrieve and present personalized results for users. However, recommendation tasks are inherently constrained to filtering existing items and lack the ability to generate novel concepts, limiting their capacity to fully satisfy user demands and preferences. In this paper, we propose a new paradigm that goes beyond content filtering and selecting: directly generating personalized items in a multimodal form, such as images, tailored to individual users. To accomplish this, we leverage any-to-any Large Multimodal Models (LMMs) and train them in both supervised fine-tuning and online reinforcement learning strategy to equip them with the ability to yield tailored next items for users. Experiments on two benchmark datasets and user study confirm the efficacy of the proposed method. Notably, the generated images not only align well with users' historical preferences but also exhibit relevance to their potential future interests.
format Preprint
id arxiv_https___arxiv_org_abs_2506_01704
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Generate, Not Recommend: Personalized Multimodal Content Generation
Liu, Jiongnan
Dou, Zhicheng
Hu, Ning
Xiong, Chenyan
Artificial Intelligence
Computation and Language
To address the challenge of information overload from massive web contents, recommender systems are widely applied to retrieve and present personalized results for users. However, recommendation tasks are inherently constrained to filtering existing items and lack the ability to generate novel concepts, limiting their capacity to fully satisfy user demands and preferences. In this paper, we propose a new paradigm that goes beyond content filtering and selecting: directly generating personalized items in a multimodal form, such as images, tailored to individual users. To accomplish this, we leverage any-to-any Large Multimodal Models (LMMs) and train them in both supervised fine-tuning and online reinforcement learning strategy to equip them with the ability to yield tailored next items for users. Experiments on two benchmark datasets and user study confirm the efficacy of the proposed method. Notably, the generated images not only align well with users' historical preferences but also exhibit relevance to their potential future interests.
title Generate, Not Recommend: Personalized Multimodal Content Generation
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2506.01704