A Distributed Hybrid Recommendation Framework to Address the New-User Cold-Start Problem

With the development of electronic commerce, recommender system is becoming an important research topic. Existing methods of recommender systems such as collaborative filtering and content analysis are suitable for different cases. Not every recommender system is capable of handling all kinds of situations. In this paper, we bring forward a distributed hybrid recommendation framework to address the new-user cold-start problem based on user classification. First, current user characteristics, user context and operating records are used to classify the user type. Then, suitable recommendation algorithms are dynamically selected based on the current user type, and executed in parallel. Finally, the recommendation results are merged into a consolidated list. In the experiment on movie ratings dataset Movie Lens, the proposed framework can enhance the accuracy of recommendation system. Also, a 100% coverage can be achieved in the case of new users. This shows the potential of the proposed framework in addressing new-user cold-start problem.