Content-based search of model repositories with graph matching techniques

Project repositories are important sources of knowledge, and their reuse and sharing is essential for improving the process of software design. Repositories contain all the data and documents for a software project, but they also contain business process models and application models. Such models explain in details the data structure, behavior, and component relationships of an application. They have a determined structure specified by the rules of the language. Given the importance of models in a software development and maintenance life-cycle, there is a need of enabling users to query model repositories. Through search engines, models can be reused according to a user's need, helping to achieve reduction of modeling time and costs. A possible solution for such a need is the adoption of search engines exploiting model similarity match techniques for content-based search, which may allow more expressive user queries, where the information need also considers the model structure. In this thesis, the problem of content-based search for a repository of models is investigated. The scope of the thesis is to provide thorough study of the current techniques and tools for project repository search, and to propose a general framework for content based-search of models. The framework uses querying on graphs, performing similarity search based on graph matching. Furthermore, a prototype implementation for a Domain Specific Language search engine is realized, and its performance is evaluated.