Design And Implementation Of A Recommender System With Alternating Least Square Matrix Factorization

On social media,e-commerce,streaming platforms or online shop business,recommendation systems play a very important role.The main objective of a recommendation system is to help a specified user to get a list of items he might be interested in,on the platform he is interacting with,depending on his inputs.It has been proven that having a good recommendation system can increase revenues drastically.Moreover,a good recommendation system improves the user experience and satisfaction.Many recommendation systems such has the item-based collaborative filtering have been proposed in the past.The problem is even with the progress that was made in the area,most of them are still not robust and powerful enough to accomplish the task as there should do.In this paper,we propose a movie recommendation system that merges the capabilities of the item-based collaborative filtering with one of the matrix factorization algorithm called the Alternating Least Square Matrix Factorization in order to recommend users only products or items they might like.At first,the system will evaluate the relationship or the distance between items by using the formal KNN item-based collaborative filtering,and secondly,it will decompose the user-item interaction matrix into the product of two lower dimensionality rectangular matrices using the Alternating Least Square Matrix Factorization to form a correlation of users and items.The results of those two operations will be merged to form our recommendation system.Our findings show that the approach we used is good.Using two different recommendation techniques and combining them together makes a great difference and offers better recommendation than traditional systems.Further,the use of the Alternating Least Square algorithm will help us recommend items properly even in a very sparse domain.The use of the Alternating Least Square fixed most of the problems that were encountered with previous systems.