Evolving mobile app recommender systems: An incremental multi-objective approach

Abstract

Existing recommender systems for mobile apps mainly focus on single objective which only reflects monotonous app needs of users. Therefore, we evolve the existing mobile app recommender systems leveraging the multi-objective approach. Moreover, to avoid risks introduced by dramatic system vibration, we realize the system evolution in an incremental manner. To achieve these two goals, we model the recommendation generation of the evolved system as a multi-objective optimization problem and propose a new rank aggregation based evolving scheme to gently evolve the systems. Furthermore, we propose a new recommending scheme for mobile apps based on Latent Semantic Analysis and leverage it to evolve the existing system. Real data evaluations have verified the effectiveness of our approach. © 2014 Springer-Verlag.