Assessing content-based conformance between software R&D documents and design guidelines using relevance links

Abstract

Research-oriented software groups are groups that carry out research on original technology for software. They also develop prototype software to demonstrate the proof of concepts of their research. The prototype software gets more complex and grows rapidly along with the growth of software technologies. Software documents describing the characteristics of the software are also diverse, and their contents are becoming enormous. However, the groups lack proper documentation on their research and software. The main reason for this problem is that research documents are not well managed, even though they have significant influence on this group and their software. Therefore, a systematic approach for managing the generated documents is required on software R&D process. We propose a method that can reduce poor documentation related to software research and development. We construct design guideline models based on the best practices and represent each measurement of design guideline models by queries of semantics-aware traceability links. Then, we use a semi-automated method to assure the conformance of R&D documents to the guidelines. Finally, we provide an explanatory guideline for assessment results to explain the detail of the assessment and provide the advice for the quality improvement in the evaluated documents. We evaluated the documents generated from our previous R&D project to show the applicability of our method. As a result, our method can help improve the quality of software R&D project documentation in a short time.