本文選題：知識推理 + MapReduce��； 參考：《電子科技大學學報》2016年03期

【摘要】：海量網(wǎng)絡流量數(shù)據(jù)的處理與單一節(jié)點的計算能力瓶頸這一矛盾導致數(shù)據(jù)分類效率低,無法滿足現(xiàn)實需求。為解決這一問題,結(jié)合本體與MapReduce技術各自在海量異構數(shù)據(jù)描述與處理方面的優(yōu)勢,提出一種基于本體的并行網(wǎng)絡流量分類方法。該方法基于MapReduce并行計算架構,根據(jù)網(wǎng)絡流量本體結(jié)構,對網(wǎng)絡流量本體并行化構建;通過并行知識推理完成基于流量統(tǒng)計特征的網(wǎng)絡流量分類。實驗結(jié)果表明,集群環(huán)境下基于MapReduce的網(wǎng)絡流量本體構建效率明顯高于單機環(huán)境,而且適當增加計算節(jié)點使得加速比線性提升;并行知識推理的分類方法能夠有效地提高大規(guī)模網(wǎng)絡流量的分類效率。
[Abstract]:The contradiction between the processing of massive network traffic data and the bottleneck of computing power of a single node leads to the low efficiency of data classification and can not meet the actual needs. To solve this problem, combining the advantages of ontology and MapReduce in describing and processing massive heterogeneous data, a parallel network traffic classification method based on ontology is proposed. The method is based on the MapReduce parallel computing architecture, constructs the network traffic ontology parallelization according to the network traffic ontology structure, and accomplishes the network traffic classification based on the traffic statistics by parallel knowledge reasoning. The experimental results show that the efficiency of constructing network traffic ontology based on MapReduce in cluster environment is obviously higher than that in single computer environment, and the increase of computing nodes makes the speedup linear. The classification method of parallel knowledge reasoning can effectively improve the classification efficiency of large scale network traffic.
【作者單位】： 桂林電子科技大學認知無線電與信息處理省部共建教育部重點實驗室;桂林電子科技大學廣西高校云計算與復雜系統(tǒng)重點實驗室;桂林電子科技大學廣西可信軟件重點實驗室;
【基金】：國家自然科學基金(61163058,61363006) 廣西可信軟件重點實驗室開放課題(KX201306) 廣西高校云計算與復雜系統(tǒng)重點實驗室開放課題(14104)
【分類號】：TP393.06
，

本文編號：2058264