Microblogging Recommended System, Based On Cloud Computing

We are at the beginning of the multicore era.The bottlenecks of performance encountered by the single CPU node become more and more obvious.However,with the use of cloud computing technology,we can complete the mass data storage and computing tasks very efficiently.Under the technical background,this paper study specificly three main parts,including the set-up of the Hadoop cloud coumputing platform and the migration of the traditional algorithm;the crawling of microblogging data sets;the analysis,improvement and optimization about the principle of collaborative filtering(CF) algorithm.In the cloud computing part,this paper bridfly introduces the concept,deployment patterns,SPI service model and the current research status of the cloud computing.Then by actively trying,paper successfully builds the hadoop clusters which provides the infrastructure protection for the following experiments.In the crawling of microblogging data sets part,this paper gives the specific crawling solution and carries on corresponding data preprocessing.Furthermore the work of the statistical properties analysis on the data set provides a solid scientific basis for the later classification algorithm.For the third part,based on the cloud coumputing environment,paper research the CF recommendation on the microblogging users,introduces several classic CF algorithm.throw the experiments,we found these algorithms is difficult to make an accurate recommendation, so paper designs a new recommending algorithm,EssCF which is more suitable giving the recommendation on very large items data set.At the same time, paper transforms EssCF algorithm into the MapReduce of the parallel programming framework.Finally,In the laboratory's distributed clusters,this paper makes a series of experiments with EssCF functional testing experiment,cluster testing experiment and the performance comparison experiment.Based on these experimental data, the paper analyzes the strengths and the weaknesses of EssCF recommending algorithm,summarizes the work of full text and gives the prospects about the next work.