A Context-based Information Retrieval Technique for Recovering Use-Case-to-Source-Code Trace Links in Embedded Software Systems

Post-requirements trace ability is the ability to relate requirements (e.g., use cases) forward to corresponding design documents, source code and test cases by establishing trace links. This ability is becoming ever more crucial within embedded systems development, as a critical activity of testing...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:EUROMICRO (IEEE Computer Society Press) S. 252 - 259
Hauptverfasser: Jiale Zhou, Yue Lu, Lundqvist, Kristina
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.09.2013
Schlagworte:
Accuracy
Context
context analysis
Context modeling
embedded system
functional requirements
Ice
post-requirements traceability
Software
trace link recovery
vector space model (VSM)
Vectors
ISSN:1089-6503
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Post-requirements trace ability is the ability to relate requirements (e.g., use cases) forward to corresponding design documents, source code and test cases by establishing trace links. This ability is becoming ever more crucial within embedded systems development, as a critical activity of testing, verification, validation and certification. However, semi-automatically or fully-automatically generating accurate trace links remains an open research challenge, especially for legacy systems. Vector Space Model (VSM), a notably known Information Retrieval (IR) technique aims to remedy this situation. However, VSM's low-accuracy level in practice is a limitation. The contribution of this paper is an improved VSM-based post-requirements trace ability recovery approach using a novel context analysis. Specifically, the analysis method can better utilize context information extracted from use cases to discover relevant source code files. Our approach is evaluated by using three different embedded applications in the domains of industrial automation, automotive and mobile. The evaluation shows that our new approach can achieve better accuracy than VSM, in terms of higher values of three main IR metrics, i.e., recall, precision, and mean average precision, when it handles embedded software applications.
ISSN:1089-6503
DOI:10.1109/SEAA.2013.30