A SVM recommendation IoT model based on similarity evaluation and collaborative filtering of multi-angle knowledge units

Abstract

ABSTRACTThe traditional user-based collaborative filtering recommendation algorithm only considers the user's score when calculating the similarity between users, but ignores the differences between different projects. Aiming at the unsatisfactory performance of traditional methods in data sparseness, a collaborative filtering recommendation algorithm based on item label information is proposed for each user to select neighbors. Firstly, based on user rating matrix to determine the initial neighbor, calculate the target users for each target neighbor; when the minimum number of nearest neighbor scoring the goal of the project or not, consider adding the development from the tag information according to the nearest neighbor; for the target item rating prediction. The experimental results show that the algorithm improves the accuracy of similarity calculation, effectively alleviates the sparsity of user rating data, and improves the accuracy of prediction.