Research And Design On Video Recommendation Technology Based On Hadoop And Mahout

Based on the recommendation and retrieval of video program data,this paper carries out the research of video recommendation system.The goal is to establish a complete video recommendation system.Based on the existing research results,a video recommendation system suitable for processing large-scale video data is studied and designed to enhance the user experience of video recommendation system and improve the accuracy and recall rate of video recommendation.The main work of this paper is as follows:(1)This paper introduced required theory,related technical achievements and recommended algorithm on the construction of Video Recommender System in-depth.After in-depth analysis and research on previous research results and corresponding video recommendation algorithm,based on the open source machine learning framework Mahout and Hadoop cloud computing platform,an improved hybrid recommendation model was proposed,and the design and implementation of video recommendation system was carried out.(2)The video recommendation system designed in this paper is "a Hybrid Recommendation Model for Video Resources Based on Hadoop and Mahout ".Including collaborative filtering recommendation algorithms,recommendations of popular videos,recommendations for new users,and cross-references of video resources.In this paper,the video recommendation lists generated by various recommendation algorithms are proportionally extracted form a total recommendation list,which is de-weighted,sorting based on the user's ranking and top-N selection processes,and finally the recommended video recommendation list is formed to the user.(3)Based on the above recommended model,this paper completes the construction of video recommendation system.Video recommendation system is divided into three parts: namely the display interface,recommended engine and video data log system.The log system is responsible for the storage of video data and user behavior.The recommendation engine is responsible for the user's video recommendation and the display interface to interact with the user.The three modules cooperate with each other to achieve the purpose of video recommendation.(4)The test and evaluation indexes of video recommendation system are used to evaluate the performance of video recommendation system.We can find this Video Recommendation System not only improved the video recommended accuracy and recall rate,but also solved the problem "cold start" and "sparse",and improved the video recommended efficiency and user experience degree.