Design And Implementation Of Recommender System Based On Stream Computing

With the development of information technology and Internet, people are in the information explosion era. Recommender system is an effective method to solute information overload, by means of analyze user behavior to recommend valuable information. Improve system’s real-time response capacity and massive data computing performance, meets the even growing business needs, these issues are not only the main technical problems faced by many internet enterprises, but also the attentional directions of many scientific research institutions.Stream computing technology is aimed at the real-time stream data computation, it mainly applied to produce enormous numbers of stream data and the fields where have a high requirement of real-time. Aiming at the above-mentioned problems which exist in the Recommender system, the author designed a recommender system based on Storm which combined the stream computing technology and recommendation algorithms. This system is based on one distributed stream computing platform called "Storm", besides, In order to enhance the efficiency of the algorithm, reduced the reaction time of the system, the system implemented User-based Collaborative Filtering Algorithm by Trident calculation method and Redis storage scheme. Moreover, it implemented message response of the cluster by means of Distributed RPC service. The system not only supported the calculation and analysis of massive data, but also met the real-time response requirements of users, and it had good extensibility as well. It reduced the complexity of system development and deployment effectively by the application of real time storm calculation which provided by the distributed stream computing platform-Storm.The author confirmed Recommender system based on Storm had many advantages, such as the performance of parallel computing, data real time feedback, system extensibility, over collaborative filtering recommender system based on traditional solution, hence this system have a certain reference value for practical application of the related industries.