Research And Application Of User Collaborative Filtering Video Recommendation System Based On Hadoop Architecture

In recent years,with the rapid development of the technologies and application such as mobile Internet,cloud computing,multimedia,interactive television is entering a digital age,providing personalized programming services which increasingly penetrate into people's lives.Mutualism,individuation and informatization has become the main features of the interactive television.Film and television recommendation system has become an important research direction of interactive television.Targeting at specific user groups,based on mining user's interest preferences,the recommendation system provides users with accurate information and product recommendations.Nowadays,under the big data environment,the original recommendation system,with its old information retrieval and information filtering technology,has been unable to meet the needs of processing massive data under a fast,real-time circumstance,which restricting the development of recommended systems.Taking interactive television recommended as the research object,combined with Hadoop and the collaborative filtering algorithm,this paper has not only analyzed and built a distributed architecture recommendation system,but also optimized and improved the existing recommendation algorithm,reaching to the goals which are anticipated.The main research contents are as follows:(1)Based on Hadoop technologies,an automatic and intelligent interactive television recommendation system has been built.This paper has distributed Hadoop architecture into the field of interactive television.By analyzing the issues of users' interest change,cold start,data sparseness and some others existing in the recommendation system,this paper has designed a big data recommendation system in interactive television industry.(2)Aimed at the current demand of personalized,accurate and efficient TV program recommendation,this paper has constructed a collaborative filtering recommendation model based on TV user's behavior,which is used in interactive TV program recommendation.Based on user's demand behavior data,content information and basic information in interactive TV,using the K-Means clustering algorithm to cluster users,combining with the characteristics of interactive television industry,this paper established the model of collaborative filtering recommendation.(3)The system is designed and implemented by big data technologies,such as Hadoop distributed file system,HBase column database and Mahout,etc.Finally,the recommended model has not only been validated on a test environment,but also applied on the actual system,which achieved good practical results,providing more appropriate services and interactive TV experience for users.