SMARTREC: A Conversational Recommendation System Using Semantic Knowledge Graphs

Sudha Vijayakumar, Jorjeta G. Jetcheva

Abstract: Conversational Recommendation Systems (CRSes) are emerging as the next-generation enabler of e-commerce services. Conversational interactions enable an interactive and engaging experience for the user and provide personalization capabilities unmatched in traditional recommendation system scenarios. In this paper, we propose a novel real-time CRS based on an intelligent domain knowledge graph with semantic network extensions, and demonstrate its effectiveness, scalability, and real-time performance relative to state-of-the-art approaches. We utilize a novel vacation booking use case, which we illustrate using a large Airbnb dataset, a conversational Artificial Intelligence (AI) component based on the open-source RASA [1] framework, and commercial-grade Neo4j [2] knowledge graph and graph-based recommendation algorithms.