부산대학교 도서관

Every software contains numerous processes for building a program, and each step is significant for software requirements. As the globe expands and develops quickly, so does the demand for software. Categorization of requirements can be done manually however doing so requires a lot of human effort, time, money, and risk of inaccurate results. As a result, numerous earlier studies have suggested automating the classification process but consumes lot of time. Here several ways are introduced such that this time taking process can be automated and software requirements can be classified using several machine learning algorithms into various categories. In the process of achieving this there were several algorithms that were taken into consideration which includes KNN, SVM, DT, Naïve Bayes to train dataset and their evaluation metrics were studied.