KIT | KIT-Bibliothek | Impressum | Datenschutz

Using BERT for the Detection of Architectural Tactics in Code

Keim, Jan; Kaplan, Angelika; Koziolek, Anne; Mirakhorli, Mehdi

Quality-driven design decisions are often addressed by using architectural tactics that are re-usable solution options for certain quality concerns. However, it is not suficient to only make good design decisions but also to review the realization of design decisions in code. As manual creation of traceability links for design decisions into code is costly, some approaches perform structural analyses to recover traceability links. However, architectural tactics are high-level solutions described in terms of roles and interactions and there is a wide range of possibilities to implement each. Therefore, structural analyses only yield limited results. Transfer-learning approaches using language models like BERT are a recent trend in the field of natural language processing. These approaches yield state-of-the-art results for tasks like text classification. We intent to experiment with BERT and present an approach to detect architectural tactics in code by fine-tuning BERT. A 10-fold cross-validation shows promising results with an average F1-Score of 90%, which is on a par with state-of-the-art approaches. We additionally apply our approach to a case study, where the results of our approach show promising potential but fall behind the state-of-the-art. ... mehr

Open Access Logo

Volltext §
DOI: 10.5445/IR/1000121031
Veröffentlicht am 08.07.2020
Cover der Publikation
Zugehörige Institution(en) am KIT Institut für Programmstrukturen und Datenorganisation (IPD)
Publikationstyp Forschungsbericht/Preprint
Publikationsjahr 2020
Sprache Englisch
Identifikator ISSN: 2190-4782
KITopen-ID: 1000121031
Verlag KIT, Karlsruhe
Umfang 16 S.
Serie Karlsruhe Reports in Informatics ; 2020,2
Schlagwörter Software Architecture, architectural tactics, natural language processing, transfer learning, traceability, language modeling, BERT
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page