Design And Implementation Of VOD Platform With Personalized Recommendation Function

With the advent of the Internet era,people's entertainment activities are increasingly rich,and the sources of information are increasingly extensive.Watching video has gradually become a mainstream way of entertainment and an important way to obtain information.However,because there are too many types of videos and there are a lot of videos on each platform,it is very important to quickly recommend users' favorite videos to users according to their preferences.Cloud on demand platform with personalized recommendation function can enable users to obtain and enable cloud on demand services in the cloud,and intelligently recommend videos to users.The existing recommendation system has many shortcomings and is not perfect,so the content of this thesis is aimed at the problems existing in the current situation of video recommendation system,using storm distributed solution to solve the problem of big data processing,and combining with the video recommendation algorithm,proposes a user-defined recommendation algorithm,realizes personalized recommendation on the video on demand platform,and solves the problems of inaccurate and untimely recommendation.This thesis completes the collection of video feature data and the construction of its video feature vector set,records the user operation log,analyzes the user's interest,and establishes the user's interest feature vector.Combined with the user's interest feature vector,it extracts the video attribute features associated with the behavior to construct,and uses the method of weighting the feature vector to establish the personalized real-time recommendation algorithm,which is finally implemented in the mobile app Combined with the video recommendation system established on the VOD platform,the personalized real-time video recommendation is realized.In the process of establishing vector set,two different schemes are adopted according to the different characteristics of user interest vector and video feature vector,that is,a fixed number of similar vectors and a fixed similarity value are used to establish the set.After completing the similar set of user interest and video feature vector,users can be recommended.Two ideas are used to recommend to users.When using the user based collaborative filtering algorithm,the cloud on demand platform recommends the video content corresponding to the vectors in the similar set of user interest vectors to users.When using the article based collaborative filtering algorithm,the cloud on demand platform recommends the video content in the similar set of video feature vectors to users.