News Recommendation Model Based on Improved Label Propagation Algorithm

Facing the ever-expanding scale of news information, how to filter and redundant information in complex and diverse data to accurately recommend information to users has become an important challenge. News recommendations have become a powerful tool for dealing with information overload. Scholars have done a lot of research on personalized news recommendation. However, due to the sparseness of user data, the traditional collaborative filtering algorithm is not only too time-consuming but also less accurate. Therefore, aiming at the characteristics of academic social networks, we propose a news recommendation model based on community detection, which builds a friend relationship community by using the users friend relationship and coherent neighborhood propinquity algorithm, and then integrates the collaborative filtering algorithm to recommend news to users. Through verification on the dataset of the academic social network SCHOLAT, we can prove that the recommendation model can achieve good accuracy while improving recommendation efficiency.