Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Yunxiao Shi, Wujiang Xu, Zeqi Zhang, Xing Zi, Qiang Wu, Min Xu

TL;DR

PersonaX tackles the challenge of modeling users from long behavioral histories for LLM-based recommendation agents by performing offline core-set construction of Sub-Behavior Sequences (SBS) through hierarchical clustering, adaptive budget allocation, and a prototypicality-diversity objective to generate multiple textual personas cached for online retrieval. It decouples profile generation from online inference, enabling fast, cached retrieval that improves downstream ranking while using only about $30 ext{-}50 ext{ } ext{%}$ of the historical data. Empirical results with AgentCF and Agent4Rec across CDs50, CDs200, and Books480 show consistent improvements (roughly $3$–$11 ext{%}$ for AgentCF and $10 ext{–}50 ext{%}$ for Agent4Rec) and significant reductions in online latency, especially on long sequences. The work offers a scalable, model-agnostic solution with practical guidance on hyper-parameter tuning and data efficiency for long-horizon user modeling in production recommendation systems.

Abstract

User profile embedded in the prompt template of personalized recommendation agents play a crucial role in shaping their decision-making process. High-quality user profiles are essential for aligning agent behavior with real user interests. Typically, these profiles are constructed by leveraging LLMs for user profile modeling (LLM-UM). However, this process faces several challenges: (1) LLMs struggle with long user behaviors due to context length limitations and performance degradation. (2) Existing methods often extract only partial segments from full historical behavior sequence, inevitably discarding diverse user interests embedded in the omitted content, leading to incomplete modeling and suboptimal profiling. (3) User profiling is often tightly coupled with the inference context, requiring online processing, which introduces significant latency overhead. In this paper, we propose PersonaX, an agent-agnostic LLM-UM framework to address these challenges. It augments downstream recommendation agents to achieve better recommendation performance and inference efficiency. PersonaX (a) segments complete historical behaviors into clustered groups, (b) selects multiple sub behavior sequences (SBS) with a balance of prototypicality and diversity to form a high quality core set, (c) performs offline multi-persona profiling to capture diverse user interests and generate fine grained, cached textual personas, and (d) decouples user profiling from online inference, enabling profile retrieval instead of real time generation. Extensive experiments demonstrate its effectiveness: using only 30 to 50% of behavioral data (sequence length 480), PersonaX enhances AgentCF by 3 to 11% and Agent4Rec by 10 to 50%. As a scalable and model-agnostic LLM-UM solution, PersonaX sets a new benchmark in scalable user modeling.

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

TL;DR

PersonaX tackles the challenge of modeling users from long behavioral histories for LLM-based recommendation agents by performing offline core-set construction of Sub-Behavior Sequences (SBS) through hierarchical clustering, adaptive budget allocation, and a prototypicality-diversity objective to generate multiple textual personas cached for online retrieval. It decouples profile generation from online inference, enabling fast, cached retrieval that improves downstream ranking while using only about of the historical data. Empirical results with AgentCF and Agent4Rec across CDs50, CDs200, and Books480 show consistent improvements (roughly – for AgentCF and for Agent4Rec) and significant reductions in online latency, especially on long sequences. The work offers a scalable, model-agnostic solution with practical guidance on hyper-parameter tuning and data efficiency for long-horizon user modeling in production recommendation systems.

Abstract

User profile embedded in the prompt template of personalized recommendation agents play a crucial role in shaping their decision-making process. High-quality user profiles are essential for aligning agent behavior with real user interests. Typically, these profiles are constructed by leveraging LLMs for user profile modeling (LLM-UM). However, this process faces several challenges: (1) LLMs struggle with long user behaviors due to context length limitations and performance degradation. (2) Existing methods often extract only partial segments from full historical behavior sequence, inevitably discarding diverse user interests embedded in the omitted content, leading to incomplete modeling and suboptimal profiling. (3) User profiling is often tightly coupled with the inference context, requiring online processing, which introduces significant latency overhead. In this paper, we propose PersonaX, an agent-agnostic LLM-UM framework to address these challenges. It augments downstream recommendation agents to achieve better recommendation performance and inference efficiency. PersonaX (a) segments complete historical behaviors into clustered groups, (b) selects multiple sub behavior sequences (SBS) with a balance of prototypicality and diversity to form a high quality core set, (c) performs offline multi-persona profiling to capture diverse user interests and generate fine grained, cached textual personas, and (d) decouples user profiling from online inference, enabling profile retrieval instead of real time generation. Extensive experiments demonstrate its effectiveness: using only 30 to 50% of behavioral data (sequence length 480), PersonaX enhances AgentCF by 3 to 11% and Agent4Rec by 10 to 50%. As a scalable and model-agnostic LLM-UM solution, PersonaX sets a new benchmark in scalable user modeling.

Paper Structure

This paper contains 45 sections, 3 theorems, 10 equations, 10 figures, 7 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Preliminary
  3. User Modeling
  4. Sub-Behavior Sequence (SBS) Selection.
  5. Method
  6. Behavior Clustering
  7. Sampling Budget Allocation
  8. In-Cluster Selection
  9. Offline Profiling and Online Selection
  10. Efficiency Analysis
  11. Experiments
  12. Experimental Setup
  13. Datasets
  14. Evaluation
  15. Downstream Recommendation Agent
  16. ...and 30 more sections

Key Result

Lemma 1

The prototypicality component $f_p$ in eq:objective is modular and hence submodular.

Figures (10)

  • Figure 1: PersonaX as a tool for user modeling given historical behaviors. PersonaX deliver retrieved persona snippet to downstream agent for decision making. In the left part, Persona.A illustrates the behavior distribution, clustering results, and selected/unselected samples. Persona.B presents distribution of the sampling budgets allocation results at a 50% selection ratio.
  • Figure 2: Online time cost analysis.
  • Figure 3: Analysis of the impact of sampling size on user modeling.
  • Figure 4: Impact of $\tau$ and $\alpha$ on PersonaX.
  • Figure 5: Sampling process for a user in $\texttt{Books}_{\texttt{480}}$ with a 50% selection ratio. Points are color-coded and outlined. Non-transparent points signify data selected, whereas transparent points delineate behaviors not sampled. A offers a holistic perspective on the user's comprehensive behavior distribution, capturing the full extent of engagement patterns. BẼ presents parts of behaviors distributions and sampling process under varying configurations of hyper-parameters. Triangles denote the centroids of the clusters.
  • ...and 5 more figures

Theorems & Definitions (7)

  • Definition 1
  • Lemma 1
  • proof
  • Lemma 2
  • proof
  • Proposition 1
  • proof