International Conference on Computer and Knowledge Engineering

Home / 14th International Conference on Computer and Knowledge Engineering

Automated software design using Machine Learning With Natural Language Processing

Authors :

Fahimeh Khedmatkon¹ Seyed Mohammad Hossein Hasheminejad² Jaleh Shoshtarian Malak³

1- Faculty of Engineering, Alzahra University, Tehran, Iran 2- Faculty of Engineering, Alzahra University, Tehran, Iran 3- Information Technology Department, Tehran University of Medical Sciences, Tehran, Iran

Keywords :

automated software design،Machine Learning،Natural language processing (NLP)

Abstract :

Designing is one of the most difficult tasks in the software development process. Researchers trying to increase the level of automation in the software development process. This is because of the daily increase of demanding software production; the quality growth of automotive software design makes designers respond more quickly to the refreshed demands. One of the good designing methods is to use of data mining. In this research, the decision tree algorithm method is used to identify classes and detect relationships between classes automatically. The data input is the proposed method is the use case descriptions software and its output is the list of classes with relationships between them. This method has three 3 stages: First, the use case description scenarios are analyzed by natural language processing tools. In the next step, by setting rules, the data set is created, and then the decision tree algorithm is executed on the dataset. output is the list of classes along with the type of relationship between them. Extracting of association relations one-to-one, one-to-many, and many-to-many and association relationships. To evaluate the proposed method, 4 case studies have been used: ATM case study, Cinema Booking System case study, Graduate Development Program case study, and Select Cruises case study. in this research, an attempt has been made to provide a method that has better results than previous methods. The proposed method on the ATM dataset has an accuracy of 90% and an F-score of 0.93 which is a value of accuracy and F-score is better than similar research and has better results.