Paper Number

ICIS2025-2157

Paper Type

Short

Abstract

Current recommender systems (RSs) on short video platforms prioritize behavioral signals, often neglecting users’ deeper value-driven motivations. We propose a novel value-aligned recommendation framework grounded in Schwartz value theory, leveraging large language models (LLMs) to simulate users and infer perceived values from video content. Our approach integrates these values into recommendation algorithms, enhancing models like SASRec and FDSA. Computational experiments demonstrate improved performance across key metrics, validating the framework’s effectiveness. This research pioneers LLM-based value extraction from multimedia, offering a scalable, theory-driven solution for value-aware RSs. The findings suggest potential for more user-centric systems, with future work planned for large-scale field experiments to assess real-world impact.

Comments

12-GenAI

Share

COinS
 
Dec 14th, 12:00 AM

Individualized Value Discovery using LLMs for Short Video Recommendation

Current recommender systems (RSs) on short video platforms prioritize behavioral signals, often neglecting users’ deeper value-driven motivations. We propose a novel value-aligned recommendation framework grounded in Schwartz value theory, leveraging large language models (LLMs) to simulate users and infer perceived values from video content. Our approach integrates these values into recommendation algorithms, enhancing models like SASRec and FDSA. Computational experiments demonstrate improved performance across key metrics, validating the framework’s effectiveness. This research pioneers LLM-based value extraction from multimedia, offering a scalable, theory-driven solution for value-aware RSs. The findings suggest potential for more user-centric systems, with future work planned for large-scale field experiments to assess real-world impact.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.