Personalization of interactive recommender systems for expert users

Abstract

Interactive recommender systems involve users in the process of recommendations. Users with different skill levels have various preferences in their type of interactions with interactive recommender systems. Adapting users' type of interactions with recommender systems according to their experience levels seems to be a promising improvement to enhance recommendation process and user satisfaction with interactive recommender systems. In this paper, expert users are automatically recognized based on their interaction with an interactive recommender system. Shopr software, an interactive recommender application for smartphones, is used to track users' interactions. Users are grouped in two categories of expert and novice users based on their interactions with Shopr software in our user study. Task completion time is used to recognize users' skill level. The result of the study show that all users recognized as expert users are indeed expert but some expert users who spent more time in the recommender system to improve their choice of shopping cannot be detected by this algorithm. The result of this research can be used to offer personalized interaction for expert users in interactive recommender systems.