A Mahout-based Collaborative Filtering Recommendation Engine: Research And Implementation

The rapid expansion of the Internet application gives users a lot of data information, wellmeet the demand of information to the user in the information age, especially the rapiddevelopment of the web and the rise of mobile Internet users to create and share content isbecoming more and more easy, the production of large Numbers of user generated contentimmediately. The existing information retrieval technology (such as search engine) to acertain extent solved the intensified contradiction, but not completely meet the demand of thesociety, when the user itself unable to provide effective keywords, or a user without a cleardemand, search engine passive driver shortcomings exposed.Then recommend as aninformation retrieval system emerging technology arises at the historic moment.It relies on itsintelligent mining user demand, actively push accurate information, soon caught the attentionof the researchers and the market.This project aims to explore building a Hadoop platform distributed recommendationengine, the use of open source frameworks Mahout collaborative filtering algorithm toachieve the traditional MapReduce programming model of the transplant. This paper firstintroduces the research background recommendation engine, significance of the topic,research status, elaborated the theory of knowledge and recommendation engine collaborativefiltering algorithm; Secondly, a detailed description of the overall architecture of therecommendation engine and recommendation engine algorithm design; Thirdly, focusing onthe engine of the recommendation engine is implemented, and finally Finally introduced therecommendation engine experimental results and preliminary application in reality.The main contribution of this paper include:1) Design and implementation of Hadoop-based collaborative filtering recommendationengine, to achieve the traditional collaborative filtering algorithm from single to MapReduceframework transplant.2) Design and implementation of the Web management engine management Hadoopplatform recommendation engine to support multi-tasking multi-strategy recommendedoperations.