An Improved and Hybrid Collaborative Filtering for Microblog Recommendation

Abstract

As the Internet grows quickly, individuals face overloads of information. Microblog has become one of the most popular social platforms with the advent of mobile Internet. It has become a hot and urgent problem to offer users with accurate microblogs. To address this issue, several techniques of recommendations are developed. The major contributions of the techniques of recommendation presented in this article are: (1) the suggested enhanced user clustering collaborative filtering recommendation method. Based on the user rating matrix, the clustering is carried out according to user similarities. This reduces the search area of the algorithm for recommendations. (2) A decline function is added, considering that the interest of the user may vary dynamically over time. (3) A real time recommendation technique based on the subject for micro blogs is developed to successfully implement a real-time suggestion by maintaining intermediate user similitude variables. Experiments indicate that in terms of running time, precision and the suggested methods have been improved.