Item category aware conditional restricted Boltzmann machine based recommendation

Abstract

Though Collaborative Filtering is one of most effective recommendation technique, the problem of dealing sparsity brings traditional collaborative filtering recommendation systems great challenge. In this paper, we propose an improved Item Category aware Conditional Restricted Boltzmann Machine Frame model for recommendation by integrating item category information as the conditional layer, aiming to optimise the model parameters, so as to get better recommendation efficiency. Experimental studies on the standard benchmark datasets of MovieLens 100k and MovieLens 1 M have shown its potential in improving recommendation accuracy.