Improved hybrid recommendation algorithm based on joint interpolation

Abstract

Aiming at the sparsity problem of the underlying score matrix of the collaborative filtering algorithm, to solve the problem of poor filling effect when the existing filling method has a large difference in the scores of items between neighbors, an improved hybrid recommendation algorithm based on joint interpolation is proposed. The algorithm first uses joint interpolation to fill in the user’s rating matrix, and then uses the similarity between the filled data and the user and item information to predict the user’s rating of the item, and then compares the item’s rating with the user’s scores of similar items, impose penalties on scores that are far apart, and finally recommend the penalized scores from high to low. Experimental results show that the algorithm has a better recommendation effect.