| In this article, study and application research around the automation of service extraction in SOA (Service Oriented Architecture) was performed. Currently, the SOA-lization of the enterprise IT architecture has begun. However, there is an obvious lack of concrete SOA service extraction procedures, and, most analysis methods neglect the semantic features of the analyzed objects. Aimed at this issue, a new business semantics covering SOA service extraction procedure was stated in this article.First of all, the procedure in this article used processes associated with the supporting IT systems as the analysis basis. By introducing the famous industry standard eTOM framework, signification of the enterprise process could be defined and the definitions could be linked to the IT systems supporting the corresponding process.Then, in order to bring the semantic definitions into the analysis of IT architecture and SOA service extraction, the article defined a sequence of quantification algebra and weight values. Meanwhile, the N.G. algorithm in statistics was improved so that the data exchange relationship and semantic features between IT systems could be taken into account and the whole analysis procedure could be more scientifically applicable to enterprise IT architecture. After that, this article used the improved N.G. algorithm to cluster the enterprise IT data exchange flow diagram and completed the extraction of SOA services.At last, for the sake of evaluation of the algorithm improvement and the eventual service extraction, a case study was performed and some qualitative as well as quantitative analysis conducted. During the research, a dedicated software supporting the entire analysis procedure was implemented, which guaranteed the operatability of the procedure in this article.The value of this article is the realization of partial automation of SOA service extraction. Meanwhile, compared to pure mathematical analysis, this article stressed the importance of business semantics. |
PDF Full Text Request