A knowledge based reengineering approach via ontology and description logic.




Journal Title

Journal ISSN



Volume Title


De Montfort University


Thesis or dissertation

Peer reviewed


Traditional software reengineering often involves a great deal of manual effort by software maintainers. This is time consuming and error prone. Due to the knowledge intensive properties of software reengineering, a knowledge-based solution is proposed in this thesis to semi-automate some of this manual effort. This thesis aims to explore the principle research question: “How can software systems be described by knowledge representation techniques in order to semi-automate the manual effort in software reengineering?” The underlying research procedure of this thesis is scientific method, which consists of: observation, proposition, test and conclusion. Ontology and description logic are employed to model and represent the knowledge in different software systems, which is integrated with domain knowledge. Model transformation is used to support ontology development. Description logic is used to implement ontology mapping algorithms, in which the problem of detecting semantic relationships is converted into the problem of deducing the satisfiability of logical formulae. Operating system ontology has been built with a top-down approach, and it was deployed to support platform specific software migration [132] and portable software development [18]. Data-dominant software ontology has been built via a bottom-up approach, and it was deployed to support program comprehension [131] and modularisation [130]. This thesis suggests that software systems can be represented by ontology and description logic. Consequently, it will help in semi-automating some of the manual tasks in software reengineering. However, there are also limitations: bottom-up ontology development may sacrifice some complexity of systems; top-down ontology development may become time consuming and complicated. In terms of future work, a greater number of diverse software system categories could be involved and different software system knowledge could be explored.





Research Institute