The planning work prior to implementing an SOA migration project is very important for its success. Up to now, most of this kind of work has been manual work. An SOA migration planning approach based on intelligent information processing methods is addressed to semi-automate the manual work. This thesis will investigate the principle research question: 'How can we obtain SOA migration planning schemas (semi-) automatically instead of by traditional manual work in order to determine if legacy software systems should be migrated to SOA computation environment?'. The controlled experiment research method has been adopted for directing research throughout the whole thesis. Data mining methods are used to analyse SOA migration source and migration targets. The mined information will be the supplementation of traditional analysis results. Text similarity measurement methods are used to measure the matching relationship between migration sources and migration targets. It implements the quantitative analysis of matching relationships instead of common qualitative analysis. Concretely, an association rule and sequence pattern mining algorithms are proposed to analyse legacy assets and domain logics for establishing a Service model and a Component model. These two algorithms can mine all motifs with any min-support number without assuming any ordering. It is better than the existing algorithms for establishing Service models and Component models in SOA migration situations. Two matching strategies based on keyword level and superficial semantic levels are described, which can calculate the degree of similarity between legacy components and domain services effectively. Two decision-making methods based on similarity matrix and hybrid information are investigated, which are for creating SOA migration planning schemas. Finally a simple evaluation method is depicted. Two case studies on migrating e-learning legacy systems to SOA have been explored. The results show the proposed approach is encouraging and applicable. Therefore, the SOA migration planning schemas can be created semi-automatically instead of by traditional manual work by using data mining and text similarity measurement methods.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:585441 |
| Date | January 2012 |
| Creators | Zhang, Zhuo |
| Publisher | De Montfort University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://hdl.handle.net/2086/9020 |
Page generated in 0.002 seconds