Traceability recovery by modeling programmer behavior

Abstract

Abstract

Traceability recovery in RAD software systems

“…With respect to software systems completely developed from scratch, such as those analyzed in [6], an industrial system may contain: In the following sub-section, each item will be analyzed, and relations to traceability issues established.…”

“…In a previous work [6], an approach to recover traceability links based on the conjecture of the sane programmer was presented. The underlying assumption is that sane programmers tend to process application-domain knowledge in a consistent way: program item names of different code regions related to a given high-level concept are likely to be the same or at least very similar.…”

“…However, in applying the method proposed in [6] to industrial software we have identified several affecting factors, related to techniques, tools and approaches adopted to increase productivity, that may have threatened traceability link recovery.…”

“…Moreover, compared to [6], the paper proposes an approach and a process, based on the sane programmer conjecture and the equations presented in [6], to recover traceability links in software systems developed with an extensive use of techniques aiming at reducing time to market. As in [6] the approach relies on Bayesian classification; however, it requires a process to filter information, accounting for factors affecting the traceability link recovery. For example, a GUI, automatically generated by a GUI builder tool, contains identifiers and comments not necessarily related to the high-level documents.…”

“…Section 2 summarizes background notions and equations presented in [6]. Section 3 introduces the traceability links affecting factors, while Section 4 presents the process to filter the information to recover traceability links.…”