基于WordNet的通用服务分类方法  

A common automatic service classification method based on WordNet

作　　者：何佳[1] 赵海燕[1] 陈庆奎[1] 席丽娜[1] 曹健[2] 

机构地区：[1]上海理工大学光电信息与计算机工程学院上海现代光学系统重点实验室,上海200093 [2]上海交通大学计算机科学与技术系,上海200030

出　　处：《计算机工程与科学》2013年第9期157-161,共5页Computer Engineering & Science

基　　金：国家自然科学基金资助项目(61073021;61272438;60970012);上海市科委资助项目(12511502704;11511500102;10DZ1200200);上海交通大学医工交叉资助项目(YG2011MS38);上海市教委科研创新资助项目(13ZZ112)

摘　　要：随着服务技术的发展,越来越多的组织将业务功能作为服务通过网络对外发布。服务的增多导致人工对这些服务进行分类的成本越来越高。将文本挖掘、语义技术和机器学习技术相结合,提出了一个基于WordNet的服务自动分类方法。首先,利用文本挖掘技术和语义消歧技术,从服务的描述文档、社会化标注等获得可描述每个服务的一组有确切语义的Sense向量,本文选取的Sense向量是对每个API进行社会化标注的一组Tags。然后,利用K-均值聚类方法完成相应的分类。最后,以Programmable Web上的服务作为测试数据进行了实验,实验表明本方法具有较好的分类效果。Abstract：With the development of ness function as service through Internet. services manually is becoming more and the service technology, more and more organizations publish their busi- With the quick increase of service number, the cost of classifying these more expensive. An automatic service classification method based on WordNet is proposed by combining text mining, semantic technique and machine learning technique. In this ap- proach, text mining and semantic disambiguation techniques are used to obtain a set of Sense vectors with exact word meaning from description documents and social annotation of the service. These Sense vectors can describe a service. They are a set of tags of socially annotating each API. Then, a K-means algorithm is used to classify these services. At last, using the services on Programmable Web as test data, experiments demonstrate that our classification method has good effect.

关 键 词：服务 语义消歧 社会化标注 相似度 WORDNET 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]