An Efficient Personalized Video Recommendation Algorithm Based on Mixed Mode

摘要

The collaborative filtering recommendation algorithm based on k-means clustering is more accurate in the case of large amount of users' ratings, and it is not suitable for the situations that users' ratings are relatively small. The video gene recommendation algorithm based on linear regression uses the opposite scenario. In order to solve the problem of general application of the single recommendation algorithm, this paper proposes a hybrid recommendation algorithm based on collaborative filtering and video gene. This algorithm first constructs user project matrix, calculates user similarity, and then cluster to get a recommendation list by k-means clustering. Then the genetic structure of the video is analysed, the gene preference is combined by style preference and regional preference, and the weight of the gene is determined by linear regression, at last, a recommendation list is obtained by selecting the object with high gene preference. Finally, the final recommendation results are obtained by weighting the recommended results in two recommendation lists. The experimental results show that the proposed algorithm has higher accuracy no matter when the number of users' ratings is large or small.