一个基于Burst检测的电影推荐算法

he recommendation algorithm research is crucial to the whole film recommendation system . But single recommendation algorithm has various drawbacks. How to use different recommendation alorithms to make the film personalized recommendation system more perfect, is the core issue of film recommendatin.The author puts forward a hybrid recommendation algorithm, which takes content-based recommendation algorithm and collaborative filtering recommendation algorithm into account. Firstly, according to the user-movie historical data and movie information, establish the initial user interest model. Then use the Burst Detection algorithm to discover user's new interests and build the user comprehensive interest model. Thirdly, use the agglomerative hierarchical clustering to cluster the users based on the comprehensive interest model. Last, use class as a unit to predict the score and the highest score movies turn into recommendation.