Enhancing Sequential Recommendations through Multi-Perspective Reflections and Iteration

Weicong Qin,Yi Xu,Weijie Yu,Chenglei Shen,Xiao Zhang,Ming He,Jianping Fan,Jun Xu

2024-09-10

Abstract:Sequence recommendation (SeqRec) aims to predict the next item a user will interact with by understanding user intentions and leveraging collaborative filtering information. Large language models (LLMs) have shown great promise in recommendation tasks through prompt-based, fixed reflection libraries, and fine-tuning techniques. However, these methods face challenges, including lack of supervision, inability to optimize reflection sources, inflexibility to diverse user needs, and high computational costs. Despite promising results, current studies primarily focus on reflections of users' explicit preferences (e.g., item titles) while neglecting implicit preferences (e.g., brands) and collaborative filtering information. This oversight hinders the capture of preference shifts and dynamic user behaviors. Additionally, existing approaches lack mechanisms for reflection evaluation and iteration, often leading to suboptimal recommendations. To address these issues, we propose the Mixture of REflectors (MoRE) framework, designed to model and learn dynamic user preferences in SeqRec. Specifically, MoRE introduces three reflectors for generating LLM-based reflections on explicit preferences, implicit preferences, and collaborative signals. Each reflector incorporates a self-improving strategy, termed refining-and-iteration, to evaluate and iteratively update reflections. Furthermore, a meta-reflector employs a contextual bandit algorithm to select the most suitable expert and corresponding reflections for each user's recommendation, effectively capturing dynamic preferences. Extensive experiments on three real-world datasets demonstrate that MoRE consistently outperforms state-of-the-art methods, requiring less training time and GPU memory compared to other LLM-based approaches in SeqRec.

Information Retrieval,Computation and Language

What problem does this paper attempt to address?

The paper aims to address the issue of dynamic user preference modeling in sequential recommendation (SeqRec). Specifically, current methods mainly face the following challenges:

1. **Single Preference**: Existing methods primarily rely on users' explicit preferences (such as product titles) while ignoring implicit preferences (such as brands) and collaborative filtering information, which leads to an inability to capture changes in user preferences.
2. **Poor Adaptability**: User interests change over time, and current methods struggle to dynamically select explicit preferences, implicit preferences, and collaborative filtering information, thereby affecting the accuracy and relevance of the recommendation system.
3. **Insufficient Update Strategy**: An ideal reflective update strategy needs to balance effectiveness and efficiency, but existing methods often achieve only one of these, either lacking adaptability or consuming excessive resources.

To address the above issues, the authors propose a new framework called MoRE (Mixture of REflectors), which generates reflections based on users' historical interactions through multi-perspective reflectors and introduces a meta-reflector to evaluate and iterate these reflections, thereby achieving effective modeling of dynamic user preferences. Experimental results show that MoRE outperforms existing recommendation methods on 3 real-world datasets and also has advantages in GPU memory usage and training time.