Completeness of Knowledge in Models Extracted from Natural Text

Proceedings of the 16th International Conference on Evaluation of Novel Approaches to Software Engineering (ENASE 2021) 2021
Viktorija Gribermane, Ērika Nazaruka

Requirements given in the form of text in natural language are a widely used way of defining requirements for software. Various domain modeling approaches aim to extract domain models from the given natural text with different goals and output models. The article focuses on evaluating 17 approaches for domain model extraction based on the completeness of the extracted knowledge of the resulting target models. Criteria for the evaluation have been defined and a comparison has been given, which highlights the importance of including all three - functional, behavioral and structural information, in order to retain the most complete extracted knowledge.

Keywords
Natural-language Requirements, Domain Modeling, Model Extraction, Natural Language Processing
DOI
10.5220/0010454301140125
Hyperlink
https://www.scitepress.org/Link.aspx?doi=10.5220/0010454301140125

Gribermane, V., Nazaruka, Ē. Completeness of Knowledge in Models Extracted from Natural Text. In: Proceedings of the 16th International Conference on Evaluation of Novel Approaches to Software Engineering (ENASE 2021), Czech Republic, Prague, 26-27 April, 2021. [S.l.]: SciTePress, 2021, pp.114-125. ISBN 978-989-758-508-1. ISSN 2184-4895. Available from: doi:10.5220/0010454301140125

Publication language
English (en)