Video Recommendaiton System Data Analysis And Decision-making Platform

In the video web site, recommend system is a kind of important application. It can helpusers to find the video which interested in on the magnitude of the video library, and also canhelp users discover their own interests. Tudou.com as one of China’s largest video website,have a large number of users to upload the video content.Due to the video library is very large,users watch video faced with serious information overload problem. To provide better videoservice for users, tudou.com developed its own recommendation system.The recommendationsystem is a hybrid recommendation system,include item-based collaborative filteringalgorithm,content-based algorithm and TopK vedio algorithm. How to determine an optimalrecommendtion sequence, is a problem which unable solved by the existing system and needfor a new support platform for doing analysis and decision-making. The videorecommendation system data analysis and decision-making platform which this articledescribed is such a platform.This paper introduces the video recommendtion system data analysis and decision-makingplatform mainly contains log processing, click rate statistics, recommend combinationdecision-making functions. The platform based on Hadoop distributed processingtechnology,using MapReduce programming which based the Key/Value technology rapidlyprocessing the massive logs. It use the multi-thread and regex techniques, efficient completionof the click-through rate statistics.In the recommendtion sequence of decision-making, withthe aid of statistical theory calculating the confidence interval of click rate, then generate anew recommendtion sequence.This paper research work of video recommendtion system data analysis anddecision-making platform has been used in the tudou.com, The ABTest shows that the overallclicks increase from15.33%to17.35%, practical application proves that the new solutionwhich calculated and analysed by the platform can effectively improve the overall CTR,increase the time of user online watching video, and increase the website’s overallcompetitiveness.