A Framework for Generating Conversational Recommendation Datasets from Behavioral Interactions

TL;DR

This paper introduces ConvRecStudio, a comprehensive framework leveraging large language models to generate realistic, multi-turn conversational recommendation datasets grounded in real user interactions, enabling improved personalized recommendation models.

Contribution

The paper presents a novel three-stage pipeline for synthesizing multi-turn conversational datasets from behavioral data using LLMs, bridging collaborative filtering and conversational recommendation paradigms.

Findings

Generated over 12,000 dialogs per dataset across three domains.

Human and automatic evaluations confirm the naturalness and coherence of the conversations.

The proposed model improves Hit@1 by 10.9% on Yelp compared to baselines.

Abstract

Modern recommendation systems typically follow two complementary paradigms: collaborative filtering, which models long-term user preferences from historical interactions, and conversational recommendation systems (CRS), which interact with users in natural language to uncover immediate needs. Each captures a different dimension of user intent. While CRS models lack collaborative signals, leading to generic or poorly personalized suggestions, traditional recommenders lack mechanisms to interactively elicit immediate needs. Unifying these paradigms promises richer personalization but remains challenging due to the lack of large-scale conversational datasets grounded in real user behavior. We present ConvRecStudio, a framework that uses large language models (LLMs) to simulate realistic, multi-turn dialogs grounded in timestamped user-item interactions and reviews. ConvRecStudio follows a three-stage pipeline: (1) Temporal Profiling, which constructs user profiles and community-level item sentiment trajectories over fine-grained aspects; (2) Semantic Dialog Planning, which generates a structured plan using a DAG of flexible super-nodes; and (3) Multi-Turn Simulation, which instantiates the plan using paired LLM agents for the user and system, constrained by executional and behavioral fidelity checks. We apply ConvRecStudio to three domains -- MobileRec, Yelp, and Amazon Electronics -- producing over 12K multi-turn dialogs per dataset. Human and automatic evaluations confirm the naturalness, coherence, and behavioral grounding of the generated conversations. To demonstrate utility, we build a cross-attention transformer model that jointly encodes user history and dialog context, achieving gains in Hit@K and NDCG@K over baselines using either signal alone or naive fusion. Notably, our model achieves a 10.9% improvement in Hit@1 on Yelp over the strongest baseline.