Research And Application Of Recommendation System Based On Hadoop

At present, Internet has penetrated into people’s lives, people for informationacquisition is mainly derived from the network. In recent years, with the advent of theinformation age, there are a lot of information every day, thus the information overloadproblem is appeared. Now the user demand for personalized information is more andmore high, how to find information from these massive to meet their demand ofinformation is an urgent problem to be solved, so scholars carried out many relatedresearch including an effective technical scheme named recommendation system.As main tool for solving the problem of information filtering, it can activelyreferring users to meet user’s demand information, it is not only basis for thisrecommendation, but also according to the user in the daily life of some informationinterest, to achieve the win-win between information providers and users demand.itdoesn’t need to put forward their needs key to look up on Internet resources. At presentthe key factors affecting the development of recommendation technology lies inincreasing order of magnitude of existing resources, data sparseness and so on.In order to better solve the recommendation system in the development of theseproblems, improving the accuracy of collaborative filtering algorithms, this paper solvesthese problems as the goal, the following three aspects are researched:①This paper focuses on the collaborative filtering algorithm in recommendersystems, presents data sparsity lead to collaborative filtering recommendation accuracyis not high problem, the concept of hybrid recommendation algorithm is proposed.②In this paper, cloud computing platform is used to solve the traditionalcollaborative filtering algorithms in process scalability and real-time problems, Afterfully understand the relevant literature Hadoop technology, this paper analyzes the keyissues constraining the performance of Hadoop, put forward the improvement platformSHadoop,which improves Hadoop on large-scale data processing capabilities fullyembodies the cloud computing and recommendation system combining the advantages.③This paper completed a prototype of a movie recommendation system, proposethe improved evaluation system of recommendation, then using the proposed hybridrecommendation algorithm compared with the traditional recommendation algorithms.This paper studies the above content discussed in detail, These experiments weredivided into two groups, the first group of experimental results show improvements indata collection and evaluation criteria, hybrid algorithm can effectively improve theaccuracy of collaborative recommendation algorithm. The second group is SHadoopwith traditional Hadoop massive data sets in a job situation comparative, Experimentresults show that compared with traditional Hadoop. SHadoop improve the ability oflarge-scale data processing, and provide good support for the recommendation systemdata services, and provides follow-up studies with a new research direction.