Automatic Detection and Resolution of Lexical Ambiguity in Process Models

Henrik Leopold, Jan Mendling, Fabian Pittke

Publikation: Wissenschaftliche FachzeitschriftOriginalbeitrag in FachzeitschriftBegutachtung

Abstract

System-related engineering tasks are often conducted using process models. In this context, it is essential that these models do not contain structural or terminological inconsistencies. To this end, several automatic analysis techniques have been proposed to support quality assurance. While formal properties of control flow can be checked in an automated fashion, there is a lack of techniques addressing textual quality. More specifically, there is currently no technique available for handling the issue of lexical ambiguity caused by homonyms and synonyms. In this paper, we address this research gap and propose a technique that detects and resolves lexical ambiguities in process models. We evaluate the technique using three process model collections from practice varying in size, domain, and degree of standardization. The evaluation demonstrates that the technique significantly reduces the level of lexical ambiguity and that meaningful candidates are proposed for resolving ambiguity.
OriginalspracheEnglisch
Seiten (von - bis)526 - 544
FachzeitschriftIEEE Transactions on Software Engineering
Jahrgang41
Ausgabenummer6
DOIs
PublikationsstatusVeröffentlicht - 2015

Zitat