Collaborative Filtering Algorithm in Pictures Recommendation Based on SVD

Under the environment of information exploding, how to find pictures from the massive information to meet their own needs, has been a major factor restricting the efficiency of UI designers. In order to reduce the cost of maintaining style consistence of UI for designers, this paper proposes a recommendation algorithm of pictures having the same style based on SVD collaborative filtering. This algorithm computes rating matrix by 0 or 1 according to whether the images are used or not, and it calculates the similarity of pictures which have been used and which have not during each new design process to predict whether they are in the same style, and optimize the calculation by SVD. It is shown in the experimental results that this algorithm is effective and it can recommend pictures of high-quality after being optimized by SVD.