Research On Optimization Of Testing Suite Based On Rough Set And D-S Theory

Today with the rapid development of information technology, computer technology has penetrated into all trades and all professions, becoming an essential means in people’s life, so there is a gradually increasing demand for people on the size and complexity of the software system. For the reason, the loss the software failure brought is greater, leading to corresponding increasing software cost and the software garbage, thus that software test to ensure software quality is more urgent. The software test is a significant method of ensuring software quality, and in recent years to which the computer academia,software engineers,and the software industry pays great attention. The software test prongs a whole life cycle of software development, from demand analysis to ultimate operating maintenance, in fact the software test occupy more than60%of the software development workload. From the above description, since the software test is an important method of ensuring software quality, an effective and feasible way of decreasing cost of the software development is to decrease cost of the software test.As a new theory of coping with uncertain problems, rough set is one of an effective way of data mining. In recent years, rough set is broadly applied in the field of data mining, artificial intelligence,and patter recognition. The important feature of rough set is, under the premise of employing information provided by the data itself without any knowledge to maintain the classified capability of knowledge unchanged, exports classification rules through knowledge reduction,provides strategic decision,and reveals the potential law of things.DS theory was born in1960s after mathematician A.P.Dempster had solved much value mapping problem using upper and lower probability and issued a series of correlative theses. Afterwards, A.P.Dempster’student made further study on the D-S theory, and promoted the theory to develop, forming a set of theory system of coping with uncertain reasoning based on "evidence"and"combination", and making great achievements.The rough set theory and DS theory are powerful tools of coping with inaccurate, uncertain information. Formation rule of DS theory itself has a disadvantage of subjectivity of basic probability distribution rules and so on, while the rough set theory does not need prior knowledge and enriched reduction algorithm, with the relationship between the two theory, the disadvantage of formation rule of DS theory could be solved better, which thus improves applicability and effects of compound algorithm.The main work of the paper summarized as follows:First, the paper analysis the present situation of test case reduction, the characteristics of rough set and DS theory and the relationship model between them.Second, against the shortcomings of the existing test case reduction algorithm, there is a new reduction algorithm, based on the concept of attributes set in the rough set theory, making the reduction algorithm more accurate when the value of fitness function is the same.Third, there is a new information fusion algorithm, based on rough set and evidence theory, to make a merge sort processing which information come from different sources in the regression testing. According to the relationship between rough set and evidence theory, the algorithm optimize the test cases in order to reduce costs.