Authors:
André Di Thommazo
1
;
Rafael Rovina
2
;
Thiago Ribeiro
2
;
Guilherme Olivatto
2
;
Elis Hernandes
3
;
Vera Werneck
4
and
Sandra Fabbri
3
Affiliations:
1
IFSP - São Paulo Federal Institute of Education, Science and Technology, Federal University of São Carlos and UFSCar, Brazil
;
2
IFSP - São Paulo Federal Institute of Education and Science and Technology, Brazil
;
3
Federal University of São Carlos and UFSCar, Brazil
;
4
Rio de Janeiro State University (UERJ), Brazil
Keyword(s):
Requirements Management Techniques, Requirements Engineering, Software Engineering, Fuzzy Logic.
Related
Ontology
Subjects/Areas/Topics:
Enterprise Information Systems
;
Information Systems Analysis and Specification
;
Requirements Analysis And Management
;
Software Engineering
Abstract:
One of the most commonly used ways to represent requirements traceability is the requirements traceability matrix (RTM). The difficulty of manually creating it motivates investigation into alternatives to generate it automatically. This article presents two approaches to automatically creating the RTM using artificial intelligence techniques: RTM-Fuzzy, based on fuzzy logic and RTM-N, based on neural networks. They combine two other approaches, one based on functional requirements entry data (RTM-E) and the other based on natural language processing (RTM-NLP). The RTMs were evaluated through an experimental study and the approaches were improved using a genetic algorithm and a decision tree. On average, the approaches that used fuzzy logic and neural networks to combine RTM-E and RTM-NLP had better results compared with RTM-E and RTM-NLP singly. The results show that artificial intelligence techniques can enhance effectiveness for determining the requirement’s traceability links.