The Research On Test Case Generation Based On Uncertain Requirements

The first step in software engineering is requirement analysis,and an accurate analysis of requirements is of great significance to software development.In reality,the requirements will be uncertain,the reasons for this come from different aspects,including the incomplete understanding of the system in the process of collection and refinement of requirements,the uncertainty of the environment in related fields,and the vagueness of language expressions,and so on.Therefore,it is natural that the model constructed for the uncertain requirements will have uncertain factors,and the test cases generated according to the model will also have the uncertainty.At present,the research into uncertain requirements has attracted considerable attentions.However,current research studies mainly focus on testing,managing the uncertainty and establishing the requirements model based on the change of requirements,and only few of them focus on how to directly generate test cases based on uncertain requirements.In this paper,we mainly study how to generate test cases for software testing based on the uncertain software requirements.We study one typical uncertain requirements,which implies ”multiple possibilities”.Our test case generation approach makes use of partial models based on labeled transition systems(LTS)to describe the uncertain requirements.Based on the characteristics of LTS and partial models,we propose an automatic test case generation algorithm.Finally,test cases are prioritized based on the weights,which relate to the uncertainty properties of the corresponding test cases,in order to provide recommendations for executing these test cases.The contributions of this paper are as follows:1.A method to model the uncertain requirements is proposed.This method is a formal modeling method based on LTS,which is able to describe the state change and path migration of the software system and is a good basis for the generation of path-based test cases.2.An automated test case generation algorithm is given.The essence of this algorithm is a path-based traversal algorithm,and it aims to generate a list of test cases from the formal model of uncertain requirements.Due to the uncertainty described by the formal model,the resulting test cases also contain information related to the uncertain requirements,which is then used for prioritizing test cases to assist the executions of these test cases.3.Mechanisms for handling of loops.Considering the complexity of loops in a model,we provide a mechanism to handle it in order to avoid the state explosion problem.