Authors:
David Šenkýř
and
Petr Kroha
Affiliation:
Faculty of Information Technology, Czech Technical University in Prague and Czech Republic
Keyword(s):
Requirements Specification, Text Processing, Grammatical Inspection, Incompleteness, Domain Model.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Knowledge Management and Information Sharing
;
Knowledge-Based Systems
;
Requirements Engineering
;
Symbolic Systems
Abstract:
In this contribution, we investigate the incompleteness problem in textual requirements specifications. Incompleteness
is a typical problem that arises when stakeholders (e.g., domain experts) hold some information
for generally known, and they do not mention it to the analyst. A model based on the incomplete requirements
suffers from missing objects, properties, or relationships as we show in an illustrating example. Our
presented methods are based on grammatical inspection, semantic networks (ConceptNet and BabelNet), and
pre-configured data from on-line dictionaries. Additionally, we show how a domain model has to be used
to reveal some missing parts of it. Our experiments have shown that the precision of our methods is about
60–82 %.