Interactive Path Reasoning on Graph for Conversational Recommendation

TL;DR

This paper introduces a graph-based conversational path reasoning framework for recommendation systems, explicitly utilizing user feedback on attributes to improve multi-round conversational recommendations.

Contribution

It proposes a novel graph-based framework, CPR, and its instantiation SCPR, for explicit attribute-based reasoning in conversational recommendation systems, outperforming existing methods.

Findings

SCPR significantly outperforms state-of-the-art CRS methods.

The effectiveness of SCPR increases with more attributes.

Empirical validation on Yelp and LastFM datasets confirms improvements.

Abstract

Traditional recommendation systems estimate user preference on items from past interaction history, thus suffering from the limitations of obtaining fine-grained and dynamic user preference. Conversational recommendation system (CRS) brings revolutions to those limitations by enabling the system to directly ask users about their preferred attributes on items. However, existing CRS methods do not make full use of such advantage -- they only use the attribute feedback in rather implicit ways such as updating the latent user representation. In this paper, we propose Conversational Path Reasoning (CPR), a generic framework that models conversational recommendation as an interactive path reasoning problem on a graph. It walks through the attribute vertices by following user feedback, utilizing the user preferred attributes in an explicit way. By leveraging on the graph structure, CPR is able to prune off many irrelevant candidate attributes, leading to better chance of hitting user preferred attributes. To demonstrate how CPR works, we propose a simple yet effective instantiation named SCPR (Simple CPR). We perform empirical studies on the multi-round conversational recommendation scenario, the most realistic CRS setting so far that considers multiple rounds of asking attributes and recommending items. Through extensive experiments on two datasets Yelp and LastFM, we validate the effectiveness of our SCPR, which significantly outperforms the state-of-the-art CRS methods EAR (arXiv:2002.09102) and CRM (arXiv:1806.03277). In particular, we find that the more attributes there are, the more advantages our method can achieve.