Detecting Programmer Profiling Quality

摘要

This paper presents the test of a software quality support tool, an automatic audit tool of java source code. This approach uses software metrics and computational intelligence techniques to evaluate java source code and to determinate programmers profile. For this, it employs traditional and new source code metrics for modeling its content in context. It employs an Artificial Neural Network for data classification and an Expert System on the recommendation build phase. The main objective of this work is to test the effectiveness of the tool to calculate the metrics and the selection of developer profiles. For this purpose, it was prepared a manual audit of some metrics and compared to the automatic results, and the allocation of samples in clusters made by the IJA Neural Network was compared to the allocation of samples made by the Expectation Maximization Algorithm (EM).