Optimization and Prioritization of Test Scenarios for Object Oriented Systems using Soft Computing Techniques

Abstract

There are three major verticals of software testing, i.e. test case generation, optimization and prioritization. At first, test cases are generated aiming at achieving maximum coverage. Then, test case optimization and prioritization are done on the generated test cases to minimize various factors like effort, cost etc. So, our work is concentrated on test suite management through generation, optimization, and prioritization of test scenarios. In our first work, an approach is proposed to optimize the test scenarios while keeping the percentage of coverage intact using Intelligent Optimization Agent (IOA). First, the System Under Test (SUT) is modelled using UML Activity Diagram which is then converted into an Activity Graph (AG). The graph is traversed and the optimized test cases are found out by using IOA. Further to make the optimized test suite more effective, a duplicate/redundancy node removal algorithm is proposed to remove the redundant nodes in optimized test scenarios. Then, the approach is extended by combining UML activity diagram with the sequence diagram. Addition of sequence diagram has increased the efficiency by detecting more faults like message dependency faults etc. Both the proposed approaches are compared with different existing approaches and found to be very effective i.e. the number of optimized test scenarios (37 for HMS case study) is better than other approaches. Our next work is proposing an approach for test scenario optimization using UML behavioural diagrams and cuckoo search algorithm. The system under test is modelled using the UML activity diagram and sequence diagram. An intermediate graph i.e. Activity Sequence Graph (ASG) is generated by considering the features of both the diagrams. Then, the cuckoo search algorithm is applied to optimize the test scenarios. The proposed work is also compared with some existing work and the number of optimized test scenarios (16 for HMS case study) is better than other approaches. Next, an approach is proposed for test scenario prioritization using UML behavioural diagrams and association rule mining. To improve the fault detection rate, an approach is proposed for prioritizing the test scenarios by using multiple modified functions and association rule mining. Here, UML activity diagram and sequence diagram are used to model the system and Activity Sequence Graph (ASG) is generated taking into account the combined features of both the diagrams. Then, test scenarios are generated by traversing the graph. The affected nodes and corresponding modified nodes are found out using a forward slicing algorithm. Then, Association Rule Mining (ARM) is applied to the historical data to generate the frequent pattern. Finally, test scenarios are prioritized based on Business Criticality Test Value (BCTV) and frequent pattern. The APFD metric is used to verify the effectiveness of proposed approach. The APFD value of our approach is 0.8258 (for HMS case study) and found to be better than other approach.