Abstract
Information overload is one of the most serious problems in big data environment, recommendation systems is a way to effectively mitigate the problem. In order to make use of rich user feedback and social networks information and to further improve the performance of the recommendation system ,This thesis makes a improvement on the user-based collaborative filtering algorithm by normalization method, Meanwhile the algorithm could be run on the MapReduce in the Hadoop platform. The experimental results show that the algorithm on Hadoop platform can effectively improve the accuracy of the data to recommend and computational efficiency, so as to improve the satisfaction of users.
Highlights
The rapid development of the Internet have brought about great changes in the world, which cause information overload[1,2,3]
Information overload has two sides .For information users, searching for what they need from the vast amounts of information accurately is becoming more difficult .For information manufacturers, making information they produce conspicuous in the vast amounts of information is a problem need to be solved urgently
In 2003 Deng Ailin put forward a collaborative filtering recommendation algorithm based on project score predicts [4] in order to solve the sharply fallen recommendation quality rating caused by extremely thin data .In 2004, Gao Fengrong adopt the method of clustering and classification respectively by divided sparse score matrix, reducing the scope of neighbor and the need to predict the number of resources, improving the efficiency and scalability of the collaborative filtering recommendation algorithm [5]
Summary
The rapid development of the Internet have brought about great changes in the world, which cause information overload[1,2,3]. In 2003 Deng Ailin put forward a collaborative filtering recommendation algorithm based on project score predicts [4] in order to solve the sharply fallen recommendation quality rating caused by extremely thin data .In 2004, Gao Fengrong adopt the method of clustering and classification respectively by divided sparse score matrix, reducing the scope of neighbor and the need to predict the number of resources, improving the efficiency and scalability of the collaborative filtering recommendation algorithm [5]. Collaborative filtering recommendation methods are mainly based on matrix decomposition [9][10]. We analyzed deeply the traditional collaborative filtering recommendation algorithm based on user, run the improved the algorithmon on MapReduce in the platform of Hadoop ,which has made the effect more accuracy
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
