Research On The Application Of Big Data Mining In Book Beading Recommendation

The advent of the era of big data has brought us a huge impingement and challenge in every vocation in the world,big data is affecting and changing our lives.It's becoming much more difficult to quickly filter out garbage resources and accurately find the resource we need in this era of information overloaded and exploded.So personalized recommendation arises at the historic moment,it is an effective way that we can quickly find the information we need from the recommendation,and save the time of searching.Similarly,personalized book recommendation can help readers to filter out many books they are not really interested from the massive library resources,and can quickly guide readers to accurately obtain the high quality books they are highly required,save the time for the readers,and also improve the utilization rate of the library collection.In general situation,personalized recommendation is generated by the recommender system.In the recommender system,the most critical part is the recommendation algorithm,the most widely used algorithm is the Collaborative Filtering Algorithm,mainly includes User-Based and Item-Based.This paper tries to use the borrow data mining readers evaluation,using the algorithms to generate personalized book recommendation,and that attempt to discuss the feasibility of using big-data mining technology to implement personalized book recommendation.The main research contents are as follows:1.Study on the current development and application situation of big data as well as the recommender system,study on the distributed framework of big data mining Apache Hadoop and Apache Mahout.2.Have a deep study on the two Collaborative Filtering Algorithms,detailed analyzed the principle and calculation steps,compared the advantages and disadvantages of the algorithms.3.Conduct an experiment with real borrow data derived from a university library management system.Using the K-Means clustering algorithm for clustering calculation according to the length of readers' borrowing time,which is regarded as the rating score of the books borrowed by readers.4.With Apache Mahout,the two Collaborative Filtering Algorithms are all coded and implemented,and conduct experiments with the borrow data,then generate recommendations.Also have a distribute implementation of the Item-Based Collaborative Filtering Algorithm on Hadoop.