Hybrid Recommendation Algorithm Combining Project Importance and Prediction Score

Aiming at the problem of long tail and poor recommendation quality, a hybrid recommendation algorithm combining item importance and predictive score is proposed. Firstly, the Personal Rank algorithm is used to calculate the item importance. Taking the project score as the edge weight, optimizing its graph structure; Increase the wandering probability of minority projects and high score projects to further capture the potential relationship between users and projects; Generate a set of candidate recommendations by ranking items by importance. Secondly, Slope One algorithm is used to predict the score of the items in the candidate recommendation set. In order to improve the accuracy of prediction, the deviation calculation method was optimized considering the similarity of user scores. The final prediction score was corrected using the credibility weight. Finally, the influence of project importance and predicted score value on recommendation is integrated, and the recommendation is completed by ranking the comprehensive recommendation value. The proposed algorithm was verified and analyzed on the real data set. The experimental results show that the proposed hybrid algorithm has improved the recommendation indexes compared with other literatures, which proves that the proposed algorithm effectively alleviates the long tail problem and improves the recommendation performance.