Analyzing similarity of software models using variant analysis approach

Software reuse has always been an integral part of software development, considering the cost and time benefits it brings in. Predominantly the significance of software reuse strategies pitches in only after the delivery of a diverse number of product variants. It has often been observed that, during the development of these variants the existing software models are cloned and tailored to meet the desired requirements. In practice, cloning brings in short term advantages such as reduction in development time and efforts; however cause long term maintenance challenges which pose to be barriers over time. Therefore other potential choices of improvising the cloning activity are being evaluated appreciably over the last few years in software development. A strong focus on the usage of a systematic reuse approach for software assets such as software models has been steadily growing importance to tackle the maintenance constraint introduced through the cloning activities. In order to systematically reuse software assets the degree of reuse potential of its elements has to be assessed. Typically, reuse potential is difficult to estimate due to the loss of similarity information among software models. Therefore the motivation of this dissertation is to provide the necessary similarity information among model variants to successfully support reuse migration. The research primarily focuses on introducing a reverse engineering approach for obtaining the similarity information from existing model variants using the variant analysis approach. However, considering the precision at which the similarity information has to be delivered, it seems a challenging task. Compared to the existing approaches the proposed framework supports simultaneous analysis of multiple model variants to deliver the necessary similarity information, allows improved correctness and completeness in the presented results and additionally reduces the response time in fetching the results. Furthermore the similarity information is provided at all level of abstraction thus making it a very efficient mechanism to estimate the potential reusable components. A generic analysis framework which can handle diverse model representations is implemented, thus ensuring instantiation in various contexts. The outcome centrally aims at addressing the need of similarity information among a set of cloned model variants to essentially overcome the maintenance complexity induced through cloning and hence making its way for software reuse.