Ensemble Clustering Approaches Applied in Group-based Collaborative Filtering Supported by Multiple Users' Feedback.

In this article, we extend our previous work based on group collaborative filtering to improve the quality of groups generated through clustering algorithms with different types of feedback. On the Web, users can interact with content in different ways, such as clicking, commenting or rating and recommender systems should be able to process and use all available information. A pre-processing step using ensemble clustering can be used to combine all this information to create a stronger recommender. In this work, we propose the use of two ensemble clustering approaches to consider different types of feedback and to  improve the quality of the recommendations. Experimental results on two different datasets demonstrate that with our approaches the recommendation accuracy is significantly improved compared to well-known recommender algorithms and with our previous work.