Information retrieval chatbots based on conceptual models

Abstract

Customer support systems based on chatbots gain an increasing popularity. Chatbots are becoming more and more important to a plethora of applications not only for social services. Modern information retrieval (IR) chatbots are based on simple queries to a database and do not ensure intelligent dialogues with users. In this paper we propose an IR-chatbot model that incorporates a concept-based knowledge model and an index-guided traversal through it to ensure the discovery of information relevant for users and coherent to their preferences. The proposed approach not only supports a search session, but also helps users to discover properties of items and sequentially refine an imprecise query.