Adaptive Personalized Conversational Information Retrieval

2025-08-12

AI Summary

Personalized conversational information retrieval (CIR) faces the challenge that user personalization needs dynamically evolve across dialogue turns. Existing approaches implicitly and uniformly fuse user profiles with dialogue context across all turns, leading to either redundant or insufficient personalization. To address this, we propose an adaptive personalized CIR framework. First, we introduce a personalized need identification mechanism that explicitly estimates the degree to which each query depends on the user profile. Second, we design a need-aware re-ranking fusion strategy that dynamically weights diverse LLM-generated enhanced queries—each integrating user profile and dialogue context—according to the inferred need strength. Third, the framework employs learnable weights to enable on-demand personalization enhancement. Experiments on the TREC iKAT dataset demonstrate significant improvements over state-of-the-art methods, validating that dynamically adapting personalization intensity is critical for enhancing multi-turn retrieval performance.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Machine Learning: Learning Preferences or Rankings
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search
• Semantics and Knowledge: Applications of semantic technologies for improving search, browsing, recommendation, personalization

Abstract

Personalized conversational information retrieval (CIR) systems aim to satisfy users' complex information needs through multi-turn interactions by considering user profiles. However, not all search queries require personalization. The challenge lies in appropriately incorporating personalization elements into search when needed. Most existing studies implicitly incorporate users' personal information and conversational context using large language models without distinguishing the specific requirements for each query turn. Such a ``one-size-fits-all'' personalization strategy might lead to sub-optimal results. In this paper, we propose an adaptive personalization method, in which we first identify the required personalization level for a query and integrate personalized queries with other query reformulations to produce various enhanced queries. Then, we design a personalization-aware ranking fusion approach to assign fusion weights dynamically to different reformulated queries, depending on the required personalization level. The proposed adaptive personalized conversational information retrieval framework APCIR is evaluated on two TREC iKAT datasets. The results confirm the effectiveness of adaptive personalization of APCIR by outperforming state-of-the-art methods.

Problem

• Determining when to apply personalization in conversational search
• Adaptively integrating personalized queries with reformulated queries
• Dynamically ranking queries based on personalization needs

Innovation

• Adaptive personalization level identification for queries
• Integration of personalized and reformulated queries
• Dynamic ranking fusion based on personalization level