Profiting from Several Recommendation Algorithms Using a Scalable Approach

This chapter proposes the use of a scalable platform to run a complex recommendation system. We focus on a system made up of several recommendation algorithms which are run as an offline process. This offline process generates user profiles that represent which algorithm should provide the recommendations to a given user and item, and will be combined with a fuzzy decision system to generate every recommendation. Yet, given the amount of data that will be processed and the need to run that offline process frequently, we propose to reduce execution time by using Hadoop, a scalable, distributed and fault-tolerant platform. Obtained results shows how the main goal pursued here is achieved: the efficient use of computer resources which allows for a significant reduction in computing time.