From Natural Language Text to Visual Models: A survey of Issues and Approaches

Abstract

Over the last 20 years, research groups focused on automating the process of extracting valuable information from Natural Language text in order to discover data and process models. In this context, several tools and approaches have been proposed. The overall objective of this survey is to examine existing literature works that transform textual specifications into visual models. This paper aims to give a comprehensive account of the existing tools meant to discover data and process models from natural language text. Our analysis focuses on approaches of these tools in the model extraction process and highlight issues of each proposed approach. In the case of object oriented software modelling of data models extraction we analyze the degree of automation, efficiency and completeness of the transformation process. Regarding process models extraction, the study is not limited only to business process discovery, but it also provides case studies from several fields such as medical or archaeological. Even if not all the tools developed are clearly depicting a Natural Language Processing technique, a review of each approach is presented.

Guiding Feature Models Synthesis from User-Stories: An Exploratory Approach

“…The interest of an automatic technique for requirement analysis with natural language processing is shown in [2]. We used some equivalent methods and tools to transform textual specifications into visual models as detailed in [26] to design our approach. A Systematic Literature Review (SLR) on 13 studies highlights the systematic use of NPL techniques for feature identification on requirements described in natural language [5].…”

Extracting TFM Core Elements From Use Case Scenarios by Processing Structure and Text in Natural Language

“…The overview of existing solutions in the field of UML model creation from textual requirements and business process model creation from textual documents [27] showed that existing means allowed creating class diagrams, object diagrams, use case diagrams, and several of them provide composition of sequence, collaboration and activity diagrams. However, all the solutions have certain limitations, for example, some require user intervention, some cannot perform analysis of irrelevant classes, some require structuring a text in a certain form before processing, and some cannot correctly determine several structural relationships between classes.…”

“…However, all the solutions have certain limitations, for example, some require user intervention, some cannot perform analysis of irrelevant classes, some require structuring a text in a certain form before processing, and some cannot correctly determine several structural relationships between classes. The only approach mentioned by the authors [27] that allows deriving a complete business process model is presented by Friedrich, Mendling and Puhlmann [28].…”