An efficient and secure feature location approach based on data fusion and data mining

Abstract

In software systems, a feature signifies a functionality which is newlinedefined in respect of requirements and accessibility to the users and newlinedevelopers. It is basically a maintenance activity handled by developers since newlineit is the chief part of the incremental change process. Nowadays, several newlineapplications are outsourced but such applications have no strong newlineincorporation of software security. Consequently, security issues are now newlinebecoming a problem for the growth of business and satisfying requirements of newlineits customers. The recent Feature Location (FL) techniques use textual and newlinedynamic approach but provide less security. To overcome this drawback, this newlinethesis proposed two contributions. newlineIn first contribution, a novel secure approach for FL utilizing data newlinefusion and data mining is proposed. It comprises five steps. Originally, the newlinerepeated Test Cases (TCs) are eradicated as of the labeled test cases. Next, newlinefrom the removed labeled test cases, select important attributes using AFO newlinealgorithm. Then, perform Association Rule Mining (ARM) to ascertain closed newlineattributes. Subsequently, encrypt the closed attributes utilizing CC-RSA newlinealgorithm. After that, find the score value of the closed attributes counts newlineutilizing entropy calculation. Finally, the score value is provided as input to newlinethe normalized-K-Means (N-(K-Means)) algorithm where the score value is newlinenormalized utilizing min-max normalization and then grouped utilizing newlineK-Means Algorithm (KMA). It proffers better result for FL in the Source newlineCode (SC). The performance proffered by the proposed N-(K-Means) is newlinecontrasted with the prevailing KMA and LSI methods. newline