Title: Resolving ambiguity in natural language specification to generate UML diagrams for requirements specification
Authors: Unnati S. Shah; Devesh C. Jinwala
Addresses: Department of Computer Engineering, C.K. Pithawalla College of Engineering and Technology, Near Malvan Mandir, Dumas Road, Surat-395007, India ' Department of Computer Engineering, S.V. National Institute of Technology, Ichchhanath, Surat-395007, India
Abstract: Requirements engineering activity often starts with the vaguely defined requirements. Hence, it is essential that when the requirements are analysed, the ambiguities are resolved to the extent possible. One of the attractive alternatives in resolving ambiguities is to convert the informal natural language requirements (NLRs) into their formal or semi-formal counterpart that ensures precision and orthogonality. Towards meeting this goal, the Unified Modelling Language (UML) notations can be exploited to an advantage. As per our literature survey, there exist numerous tools that take NLR as input and output the UML counterpart. However, most of the tools do not support normalising and resolving ambiguities. In this paper, we propose an architecture viz. ARUgen that is aimed to resolve ambiguities, find object oriented terms and relationship to generate accurate UML diagrams with Java code. Our tool ARUgen mainly deals with ambiguities viz. pronoun anaphora, verb anaphora and coordination.
Keywords: requirements engineering; natural language processing; NLP; ambiguity; Unified Modelling Language; UML diagrams; requirements specifications; object oriented terms; object oriented relationships; Java code; pronoun anaphora; verb anaphora; coordination.
International Journal of Software Engineering, Technology and Applications, 2015 Vol.1 No.2/3/4, pp.308 - 334
Received: 28 May 2015
Accepted: 30 Oct 2015
Published online: 30 Mar 2016 *