Improving Conversational Recommendation with Contextual Adaptation of External Recommenders and LLM-based Reranking

TL;DR

This paper presents CARE, a framework that combines external recommender systems with LLMs to improve conversational recommendation accuracy by contextual adaptation and reranking.

Contribution

The paper introduces CARE, a novel framework that integrates external recommenders with LLMs for domain-specific and context-aware conversational recommendations.

Findings

Incorporating CARE improves recommendation accuracy by up to 54%.

LLMs effectively rerank candidate items from external recommenders based on context.

CARE significantly outperforms zero/few-shot LLM approaches in CRS tasks.

Abstract

We tackle the challenge of integrating large language models (LLMs) with external recommender systems to enhance domain expertise in conversational recommendation (CRS). Current LLM-based CRS approaches primarily rely on zero/few-shot methods for generating item recommendations based on user queries, but this method faces two significant challenges: (1) without domain-specific adaptation, LLMs frequently recommend items not in the target item space, resulting in low recommendation accuracy; and (2) LLMs largely rely on dialogue context for content-based recommendations, neglecting the collaborative relationships among item sequences. To address these limitations, we introduce the CARE (Contextual Adaptation of Recommenders) framework. CARE (a) integrates external recommender systems as domain experts, producing candidate items through entity-level insights, and (b) customizes LLMs as rerankers to enhance the accuracy by leveraging contextual information. Our results demonstrate that incorporating CARE framework significantly enhances recommendation accuracy of LLMs by an average of 54% and 25% for ReDial and INSPIRED datasets. The most effective CARE strategy involves LLMs selecting and reranking candidate items that external recommenders provide based on contextual insights.