Collaborative filtering based on voting scores has been known to be the most successful recommendation technique and has been used in a number of different applications. Collaborative filtering system collects human judgments for items and matches together people who share the same needs or the same tastes. However, since customer seldom votes on products they used, this technique suffers from the sparsity problem. To overcome the problem, this paper establishes overall similarity degree by considering customers'' personal features to improve the original similarity degree in collaborative filtering. Genetic algorithm-based approach is utilized to determine the weight value of each feature of a customer. Experiments result shows this method has better performance on recommendation effect.
展开▼
