Software validation based on prioritization using concurrent activity diagram

Quality of developed software totally relies upon three factors the time, effort and testing technique used for testing. Normally in large organizations, the development team allocates a high portion of estimated development time for software testing. Therefore, efficient algorithm needed for designing optimized test scenarios. Proposed approach can be apply on large and complex software. One of the most crucial and tedious task in SDLC is the generation of test scenario specially for large and complex problems. Generation as well as to execution of large number of test cases consumes high portion of effort and duration of total development effort and duration respectively. Therefore automatic testing has become the necessity of software industry specially large scale software development organization to reduce the testing cost to develop qualitative product. Also its very impractical to execute complete set of test case due to limited time and cost, the prioritization of test case is the solution to improve the software quality. Paper proposes a modelling based testing approach to generate test scenarios that uses UML activity diagram (AD). To prioritized the test cases average percentage fault detection (APFD) metrics is used. the proposed approach carries two phases, In the first phase specification information of AD is transferred into an arbitrary and testable graph called activity interaction graph using proposed parser. To execute the second phase a algorithm named TSPACO: the combination of DFS and BFS is proposed. In second phase TSPACO is applied to generate test scenarios with respect to decision and concurrent criteria to prioritize the test scenarios. The proposed model generates prioritized test scenarios according to strength values of different types of activity diagram which are- forks, joins and merge point’, developed by using the proposed TSPACO algorithm. Using the APFD metric, effectiveness of the generated test scenarios is computed. The experimental results shows that test cases generated by proposed approach have 14% more effectiveness than the other existing approaches.