PersonaX: A Recommendation Agent Oriented User Modeling Framework for Long Behavior Sequence

2025-03-04

AI Summary

Existing large language model (LLM)-based user modeling faces three key bottlenecks: (1) long behavioral sequences exceeding context-length limits; (2) incomplete interest coverage due to relevance- or recency-biased sampling; and (3) high online inference latency from real-time retrieval. To address these, we propose an offline sub-behavioral sequence (SBS) selection and multi-textual user profiling framework. Our method introduces the first offline SBS sampling mechanism that jointly optimizes representativeness and diversity; enables fine-grained, multi-perspective textual profile pre-generation and caching for zero-latency online inference; and establishes the first agent-agnostic, scalable LLM-based user modeling paradigm. Experiments show that, using only 30–50% of raw behavioral data (sequence length 480), our approach improves recommendation accuracy by 3–11 percentage points on AgentCF and by 10–50 percentage points on Agent4Rec, significantly enhancing both user profile completeness and recommendation precision.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Recommendation agents leverage large language models for user modeling LLM UM to construct textual personas guiding alignment with real users. However existing LLM UM methods struggle with long user generated content UGC due to context limitations and performance degradation. To address this sampling strategies prioritize relevance or recency are often applied yet they inevitably neglect the diverse user interests embedded within the discarded behaviors resulting in incomplete modeling and degraded profiling quality. Furthermore relevance based sampling requires real time retrieval forcing the user modeling process to operate online which introduces significant latency overhead. In this paper we propose PersonaX an agent agnostic LLM UM framework that tackles these challenges through sub behavior sequence SBS selection and offline multi persona construction. PersonaX extracts compact SBS segments offline to capture diverse user interests generating fine grained textual personas that are cached for efficient online retrieval. This approach ensures that the user persona used for prompting remains highly relevant to the current context while eliminating the need for online user modeling. For SBS selection we ensure both efficiency length less than five and high representational quality by balancing prototypicality and diversity within the sampled data. Extensive experiments validate the effectiveness and versatility of PersonaX in high quality user profiling. Utilizing only 30 to 50 percent of the behavioral data with a sequence length of 480 integrating PersonaX with AgentCF yields an absolute performance improvement of 3 to 11 percent while integration with Agent4Rec results in a gain of 10 to 50 percent. PersonaX as an agent agnostic framework sets a new benchmark for scalable user modeling paving the way for more accurate and efficient LLM driven recommendation agents.

Problem

• Addresses limitations of LLM user modeling with long behavior sequences.
• Proposes offline SBS selection for diverse user interest capture.
• Eliminates online modeling latency, enhancing recommendation agent efficiency.

Innovation

• Offline sub-behavior sequence selection for diverse interests
• Cached fine-grained textual personas for efficient retrieval
• Balanced prototypicality and diversity in sampled data