Hybrid recommendation algorithm based on Hamming clustering for user's access log and weighted user behavior

When the Hamming clustering is applied to personalized recommendations for the user access logs, since it does not reflect the user's preference for the project, so we proposed a hybrid recommendation algorithm based on Hamming clustering and weighted user behavior. This algorithm improves Hamming clustering in two aspects-sparse matrix processing and also improves the number of hits on the Web page, thus preventing the accidental clicks from affecting the user's clustering result. And we used the log-likelihood ratio to calculate the similarity of the users, and then filter the recommendation content through the weighted user behavior model. This model includes three aspects: the similarity between the users, the weighted user behavior and the visit time for the users. According to the data from the Github Archive website, it indicates that the proposed algorithm has a higher recommendation in terms of quality and efficiency.