Personal Micro-blog Recommendation Based on User Characteristics and User Preference

In order to solve the problem that the recommendation algorithm is not personalized enough and the low accuracy, we propose a method of personalized micro-blog recommendation algorithm that combines user preference and characteristics. Firstly, we utilize the user's three major characteristics combine with time-effect function to calculate user similarity, and select the user's nearest neighbors; Then, we calculate the user preferences for different categories of real-time to obtain the categories which user preference. Finally, we combine the user nearest neighbor favorite tweet category and the user of the same preferences composition of the tweet together to produce recommended list, then based on improved forecasting score formula to predict the degree of interest in micro-blog recommended to users. Experimental results show that our method could solve the problem of algorithm not enough personal effectively and improve the accuracy of the recommendation.