摘要
为提高Web服务推荐算法的可靠性和时效性,提出一种Spark框架下基于对比散度的网络服务推荐算法。基于用户之间的直接信任关系,构建信任网络无向图模型,提出一种基于两层受限玻尔兹曼机的Web服务质量预测模型TLRBM(Two Layers Restricted Boltzmann Machine Model),并将该模型应用于Web服务质量预测。为提高算法处理Web服务大数据的能力,采用对比散度算法CD(Contrastive Divergence)来提高收敛速度,并采用Spark框架实现TLRBM的并行化执行,大幅度提升了Web服务推荐算法的计算速度。通过在Epinions数据集上的仿真测试,验证了该算法在Web服务推荐算法的可靠性和时效性上的性能优势。
To improve the reliability and timeliness of Web service recommendation algorithm,I proposed a recommendation algorithm based on contrast divergence in Spark framework.Based on the direct trust relationship between users,I constructed the undirected graph model of trust network,and proposed a prediction model based on two-layer restricted Boltzmann machine(TLRBM).The model was applied to the prediction of Web service quality.The Contrastive Divergence(CD) algorithm was adopted to improve the convergence speed,and it improved the algorithm ability of processing Web big data.The Spark framework was used to implement the parallel execution of TLRBM,which greatly improved the computing speed of Web service recommendation algorithm.The simulation tests on Epinions dataset verified that the performance advantages of the proposed algorithm in the reliability and timeliness of Web service recommendation algorithm.
作者
那勇
Na Yong(Jilin Distance Education Technology Innovation Center,Changchun 130022,Jilin,China;Department of Distance Education Technology,Jilin Radio and TV University,Changchun 130022,Jilin,China)
出处
《计算机应用与软件》
北大核心
2019年第8期293-299,共7页
Computer Applications and Software
基金
吉林省科技发展计划资助项目(20190902010TC)
关键词
Spark框架
并行化
WEB服务
玻尔兹曼机
推荐算法
云计算
大数据
Spark framework
Parallelization
Web services
Boltzmann machine
Recommendation algorithms
Cloud computing
Big data
