DOI: 10.5176/2251-2039_IE17.38
Authors: Chen-Ya Chang, J. H. Chou
Abstract: In order to enhance the performance of software engineers, an index is needed to evaluate the coding efficiency when they use libraries. This need arises because the software engineers must use a large number of libraries when they are coding. In this work, we develop a dependence index ???? by using the data mining approach to evaluate and to obtain the status about how some library modules are used in different teams or different software products. We adopted the commercial software programs from the financial industry for this purpose. They are two sets of Java codes containing 536 programs and 601 programs, respectively. The performance of two groups of software engineers was evaluated and compared. The outputs report that several libraries are imported together more frequently and some libraries are used by the two teams for different software products. The index values provide two major contributions to software teams. First, it shows how the libraries are being grouped and used more in codes, implying relevance of the libraries. Second, the status of libraries used in different software teams can be clearly shown; higher Si values show that programmers follow the company’s policy of using more common libraries. The policy is typically what being imposed on programmers to follow to enhance the productivity of software development as individual efforts can be drastically reduced and common experience can be continuously built. We understand that we need to evaluate more by using more specific library sets so that the universal applicability of our method can be demonstrated and this task is being pursued.
Keywords: Software Engineering; software team; productivity; data mining ;
Updating...