A Collaborative Filtering Hybrid Recommendation Algorithm for Attribute and Rating
Lan Dongmei · Computer Technology and Development · 2013
Traditional collaborative filtering algorithm exists poor recommendation quality for recommending to the user based solely on sparse rating matrix.Propose a collaborative filtering hybrid recommendation algorithm for attribute and rating.The algorithm computes similarity based on attribute between item by category attributes of item,takes user's interests change over time into account,builds exponential function based on weight of rating time,and applies to Pearson correlation similarity between item.Weighted similarity based on attribute between item and Pearson correlation similarity between item by weighting factor,then calculated rating prediction based on item attribute.Depict professional classification tree,builds profession similarity model,and gets similarity of user's combined attribute by weighted sex,then obtains rating prediction based on user's attribute.At last,compute combined rating prediction by integrated the two above.Under experimental data set of Movielen,experimental results show that the proposed algorithm has a better mean absolute error.