Leveraging Knowledge Graph Embedding for Effective Conversational Recommendation

TL;DR

This work addresses the challenge of effectively integrating user, item, and attribute relations in conversational recommender systems. It proposes KG-CRS, a dynamic knowledge-graph framework with graph embeddings, a recommendation module, and a conversation policy trained via reinforcement learning, where negative entities are removed during dialogue to keep learning focused. The approach yields improved recommendation accuracy and more efficient, targeted conversations across three real datasets, with ablations confirming the value of graph-based representations and dynamic updates. The results suggest that explicit knowledge-graph modeling and dynamic interaction provide significant practical benefits for multi-round conversational recommendations.

Abstract

Conversational recommender system (CRS), which combines the techniques of dialogue system and recommender system, has obtained increasing interest recently. In contrast to traditional recommender system, it learns the user preference better through interactions (i.e. conversations), and then further boosts the recommendation performance. However, existing studies on CRS ignore to address the relationship among attributes, users, and items effectively, which might lead to inappropriate questions and inaccurate recommendations. In this view, we propose a knowledge graph based conversational recommender system (referred as KG-CRS). Specifically, we first integrate the user-item graph and item-attribute graph into a dynamic graph, i.e., dynamically changing during the dialogue process by removing negative items or attributes. We then learn informative embedding of users, items, and attributes by also considering propagation through neighbors on the graph. Extensive experiments on three real datasets validate the superiority of our method over the state-of-the-art approaches in terms of both the recommendation and conversation tasks.

Figures (15)

• Figure 1: A session of Conversational Recommender System.
• Figure 2: The KG-CRS framework.
• Figure 3: An example of knowledge graph in our study.
• Figure 4: Examples depicting the two pre-trained strategies.
• Figure 5: Comparative results with different $T$ (the maximum number of rounds).