基于Word2Vec和LDA主题模型的Web服务聚类方法

来源期刊:中南大学学报(自然科学版)2018年第12期

论文作者:曹步清 肖巧翔 张祥平 刘建勋 李晏新闻

文章页码:2979 - 2986

关键词:Web服务;Word2Vec;LDA主题模型;K-means算法;Web服务聚类

Key words:Web services; Word2Vec; LDA topic model; K-means algorithm; Web service clustering

摘    要:为高效地发现满足用户需求的Web服务,针对Web服务的描述文本较短、缺乏足够有效信息的问题,提出一种基于Word2Vec和LDA主题模型的Web服务聚类方法。该方法首先将Wikipedia语料库作为扩充源,使用word2vec对Web服务描述文档内容进行扩充,再将扩充后的描述文档利用主题模型进行特征建模,将短文本主题建模转化为长文本主题建模,更准确地实现服务内容主题表达,最后根据文档的主题分布矩阵寻找相似的服务并完成聚类,使用从ProgrammableWeb收集的真实数据进行实验。研究结果表明:本文方法与TFIDF-K,LDA,WT-LDA和LDA-K方法相比,F分别提高419.74%,20.11%,15.60%和27.80%,利用扩充后的Web服务的描述文档进行聚类的方法能够有效提高Web服务聚类的效果。

Abstract: Considering that the description text of Web service is short and lack of enough effective information, a Web service clustering method was proposed based on Word2Vec and LDA topic model in order to find the Web service that meets user’s needs efficiently. Firstly, Wikipedia corpus was used as an extension source, and Word2Vec was used to extend the content of Web service description document, and then the expanded description document was modeled using the topic model. The short text topic modeling was transformed into a long text topic modeling, which achieved the topic of service content expression more accurately. Finally the similar service was found based on the topic distribution matrix of the document and the clustering was completed. Real data from ProgrammableWeb was used to carry out experiments. The results show that F obtained by the method increases by 419.74%, 20.11%, 15.60%, 27.80%, respectively, compared with those using TFIDF-K, LDA, WT-LDA and LDA-K. The use of extended Web service description documents clustering method can effectively improve the effectiveness of Web service clustering.

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号