Conflict detection in non-functional requirement using conceptual graph representation

The conflict of non-functional requirements (NFRs) is one of main problems in analysis and design phase in software engineering. Because NFR is hard to be understand and can be conflicted, contradicted, or interfered with another NFRs. Additionally, the main character of NFRs, they may be connected to the same quality attribute that leads to misunderstanding and cause of the requirement conflict problem. This dissertation introduces knowledge representation method for detecting the conflicts in NFRs using the conceptual graphs. This dissertation proposed the method to extract the NFR from the input software requirement specification document and detect conflict that may occur between the NFRs. The results from the billing system case study from the telecommunication company in Thailand indicates that there are total NFR xml frames extracted from the software specification document are 353 frames and it is valid frame in case of complete NFR, incomplete NFR frames, and Invalid NFR frames and the accuracy of extracting process in case of the precision is 0.36, the recall is 0.65 and F1 measure is 0.46. Moreover, these NFR xml frame can be detected as the candidate NFR conflict for 20 NFRs. and detected 2 NFRs as a NFR conflict, and report 2 NFRs as a potential NFR conflict or a bad smell NFR.