Trustrace: Improving Automated Trace Retrieval through Resource Trust Analysis

Abstract

Traceability is a task to create/recover traceability links among different software artifacts. It uses resources, such as an expert, source and target document, and traceability approach, to create/recover traceability links. However, it does not provide any guidance that how much we can trust on available resources. We propose Trustrace, a trust-based traceability recovery process, to improve expert trust on a recovered link and trust over the traceability inputs. Trustrace has three sub components, in particular, Link trust improver (LTI), traceability factor controller (TFC), and a hybrid traceability approach (HTA). LTI uses various source of information, such as temporal information, design documents, source code structure, and so on, to increase experts' trust over a link. To develop TFC, we will perform a systematic literature review and empirical studies to find out which factors impact the traceability-process inputs and document these factors in a trust pattern. TFC trust pattern will help practitioner and researchers to know which steps they can take to avoid/control these factors to improve their trust on these inputs. In the HTA, we will combine different traceability recovery approaches. All approaches have different positive and negative points, we will combine all the positive points of different approaches to increase experts' trust over the HTA. In Trustrace, HTA will implement the LTI model following TFC instructions to improve the expert trust over recovered link as well as precision and recall.