Recommender Systems based on biclustering and evolutionary simultaneous learning
Recommender Systems are tools which aid the customized selection of items according to user's preferences. Movies, music, and e-commerce are examples of well-known applications. Common characteristics of data on these domains are high-dimensionality, due to a large number of users and items, and high sparsity, since each user interacts with only a few items. In this setting, obtaining a model that represents the relationship between all users and all items is a challenging task. As proposed in the algorithm SCOAL (Simultaneous Co-Clustering and Learning), the use of biclustering and learning local models has emerged as a good solution for this problem. However, the number of local models is still a critical parameter that must be set manually. To optimize the search for the best value of this parameter, we propose the SCOAL-variant EvoSCOAL (Evolutionary Simultaneous Co-Clustering and Learning). Our preliminary results suggest that EvoSCOAL can obtain competitive predictive accuracy when compared to MSCOAL, a SCOAL-variant based on bisection. Also, we present improvements and different aspects which will be investigated in our approach in the continuation of this work.