Addressing Cold Start Challenges In Recommender Systems: Towards A New Hybrid Approach
Nouhaila Idrissi, Ahmed Zellou, Oumaima Hourrane, Zohra Bakkoury, El Habib Benlahmar · 2019
Along with the rapid expansion of data in information systems, managing and searching for personalized information has become a tedious task. Users are overwhelmed by a wide range of available choices. It then becomes necessary to have access to effective tools and techniques to filter data and make it usable for everyday operations. Recommender systems have been considered as key tools for assisting the user and providing her with more effective access to information through personalized recommendations based on prior feedback. However, this becomes difficult when the rating or purchase history is not available. This is known as the cold start problem. To fill these gaps, this paper proposes a new hybrid approach considering three main modules and relying on the combination of semantic recommendation, demographic recommendation, and Singular Value Decomposition-based collaborative recommendation.