Building Recommendation Systems Using the Algorithms KNN and SVD

Abstract

Recommendation systems are used successfully to provide items (example:movies, music, books, news, images) tailored to user preferences.Among the approaches proposed, we use the collaborative filtering approachof finding the information that satisfies the user by using thereviews of other users. These ratings are stored in matrices that theirsizes increase exponentially to predict whether an item is interestingor not. The problem is that these systems overlook that an assessmentmay have been influenced by other factors which we call the cold startfactor. Our objective is to apply a hybrid approach of recommendationsystems to improve the quality of the recommendation. The advantageof this approach is the fact that it does not require a new algorithmfor calculating the predictions. We we are going to apply the two Kclosestneighbor algorithms and the matrix factorization algorithm ofcollaborative filtering which are based on the method of (singular valuedecomposition).