Classification and Evaluation of Attributes in Trustworthy Software: Analytical Approach to Questionnaires and Defect Reports

摘要

The issue about trustworthy software has been widely concerned, which is no longer limited to security or reliability. The concept of software trustworthiness in this paper includes five important attributes: availability, reliability, security, safety, and maintainability. It attempts to employ questionnaire survey and defect reports from the users to validate the importance and influence of these attributes to trustworthy software. In addition, an evaluation index set is explored for evaluating the trustworthiness of a software product, which is incorporated with a fuzzy comprehensive evaluation approach. Once the attributes are determined, the defect reports will be used to justify and analyze the degree of importance related to software trustworthiness via a text mining method. A case study has been conducted to show the effectiveness and usefulness of this proposed classification and evaluation method.