Using similarity measures for test scenario selection

Abstract

Specification based testing involves using specification as the basis for generating test cases. The Unified Modeling Language (UML) consists of diagrams to capture the static and dynamic behaviour of a system. Generating test scenarios using UML activity diagrams produces all possible scenarios which is impossible to test exhaustively. Test scenario selection involves selecting an effective subset of scenarios for testing. This paper presents a new metric based on common subscenario between scenarios, their lengths, and weights based on the position of the common subscenario in the scenario. The aim of this strategy is to select scenarios that are least similar and at the same time provide high coverage. The method is compared with results of random selection to study effectiveness of our technique.