DETECTION OF CROSS-PROJECT BENEFICIAL CLONES
Kavitha Esther Rajakumari, T. Jebarajan · 2014
Duplicate codes are also known as code clones. They are considered as one of the main factors that deteriorate the quality of software. They are usually discarded by using automatic clone detection tools. In this paper the clones are detected using a data mining approach. The clones are well analyzed and the beneficial code clones are retained. These clones are maintained separately and are used in software maintenance. The beneficial clones will definitely help in reducing the overall time spent in maintenance phase. In this paper a method is proposed to detect and to store the beneficial clones. The clones are detected based on data mining algorithm and stored separately in a table. These good clones are purported as a database and used efficiently. 2. Literature Survey Most of the researchers have discussed about the negative aspects of clones. Below are the summary of techniques and views expressed about clones by various authors, in the field of code clones. Pitts and Raoult and Guillemin describe the equivalence of programs in terms of operational semantics. Pits introduced a method proving contextual equivalence of ML functions. Result and Guillemin proves that two different ways of program equivalence are fixed-point semantics and operational semantics. They are discussed for recursive definitions. Ivanovo introduces a technique called program schemata. He has said that program transformation is to transform a program into a semantically equivalent one by applying only semantic- preserving transformations. Fischer tells whether a component can be reused in a given context or not without any modification. The basic idea of the reuse approach is that a component satisfies a query with precondition and post condition. Podgurski and Pierce introduce a method called behavior sampling for the automated retrieval of components from a software component library for the purpose of reuse. The components are organized in a classification. The user is prompted with a choice between two different detections and its behavior. They are static and dynamic similarity detection.