Optimizing Support Vector Machine for Classifying Non Functional Requirements

Problems faced in contemporary practice should be understood to improve requirements engineering processes. System requirements are descriptions of services provided by a system and operational constraints. NonFunctional Requirements (NFR) defines overall qualities/attributes of the system. NFR analysis is a significant activity in this branch of engineering. In this study, a methodology for classifying NFR is presented. Inverse Document Frequency is used for extracting the features from the NFR dataset and is classified by Support Vector Machine (SVM). The efficiency of the SVM depends upon the parameter used with Radial Basis Function. In this study, the RBF kernel is optimized by Artificial Bee Colony algorithm (ABC) to optimize the RBF parameters to improve performance.