Collaborative Filtering Recommendation Bottlenecks Review
Yiming Cao · IEEE Software · 2012
The personalized recommendation,through which users are free from the vast information retrieval,is an important way to obtain information after search engine.Collaborative filtering,a simple algorithm,is able to handle complicate objects and have good reommendations,becoming the most successful and the most widely used technology in personalized recommendation.However, collaborative filtering suffers serious challenges from data sparseness problem,cold start problem and scalability problem with the expansion of recommendation system.This article summarizes the traditional collaborative filtering recommendaion process,mainly researches solutions of problem of collaborative filtering bottleneck and analyzes the advantages and disadvantages of the solutions,In order to facilitate the follow-up selection of the solutions.