The Video Recommendation System Based On Data Mining Modeling Research

Video related service is developed by all kinds of network operator because of itsvivid image of display form which attracts so many users. Not only broadcast network,but also telecommunication network and Internet, video related service is their keydeveloping direction. In face of the explosion of video related service and videocontents, the information overloading makes users feel like a fish out of water. Thenvideo recommendation technology is emerged and will benefit both users and networkoperators.This article will analyze the state of art of video recommendation and the maintechnologies used in video recommendation. And then, carry out the system modelingresearch of video recommendation based on association rules and semantics tags.The system modeling research based on association rules will mainly focused onthe association study of items which including video resource and users. So there aretwo methods, one gets the recommendation results from the videos which are morerelated with the watching history of target user, the other one bases on the other users’prefers and these users are quite related to the target user. Based on the user associationstudy, this article gives a promoting method to improve the classic score predictalgorithm Slope One which is also used to system modeling research of videorecommendation.The system modeling research based on semantics tags will firstly design a set oftags which can greatly describe the characteristics of the video and then model the videoand preference of the users’. At last, this article gives three methods to generaterecommendation result and merge their results together, which try its best to give thebest and comprehensive results.What this article focuses on has a great engineering significance, whose main valueand innovations are as following:Detailed analyze the recommendation system model based on association ruleswhich includes item association rules and user association rules, at the same time, do alot of experiment analysis. Get a preferably more than ten percent score prediction accuracy improvementto Slope One algorithm which is perfectly used in recommendation system modeling.The system modeling research based on semantics tags gives a whole modelingsolution to video recommendation system which possesses preferable feasibility andengineering significance.