Toward Gamified Personality Acquisition in Travel Recommender Systems

摘要

This paper proposes a novel method for user profiling in recommender systems (RS). RS have emerged as a key tool in information filtering. But despite their importance in our lives, systems still suffer from the cold-start problem: the inability to infer preferences of a new user who has not rated enough items. Up till now, only limited research has focused on optimizing user profile acquisition processes. This paper addresses that gap, employing a gamified personality-acquisition system based on the widely used Five Factor Model (FFM) for assessing personality. Our web-based system accurately extrapolates a user's preferences by guiding them through a series of interactive and contextualized questions. This paper demonstrates the efficacy of a gamified user profiling system that employs story-based questions derived from explicit personality inventory questions. The Gamified Personality Acquisition (GPA) system was shown to increase Mean Absolute Error (MAE) and Receiver Operating Characteristic (ROC) sensitivity in a travel RS while mitigating the cold-start problem in comparison to rating-based and traditional personality-based RS.