Authors:
Viktorija Gribermane
1
and
Erika Nazaruka
2
Affiliations:
1
Institute of Applied Computer Systems, Riga Technical University, 1 Setas St., Riga, Latvia
;
2
Department of Applied Computer Science, Riga Technical University, 1 Setas St., Riga, Latvia
Keyword(s):
Natural Text Processing, Knowledge Extraction, Unrestricted Language, Domain Model, Software Model.
Abstract:
Text processing techniques are critical for automated analysis of domain documentation. Proper domain analysis may include analysis of a huge number of documents that may describe business procedures, policies, organizational structures, regulations, etc., as well as minutes of discussions with domain experts. The results of analysis are also presented as documents with or without supporting graphics, e.g., domain models. Automated composition of domain models (including software models) should decrease the time necessary for analysis and would provide traceability to the original documentation. The goal of this research is to understand how text processing techniques can be applied for composition of those models and what are the current trends in this field. The result of analysis of 15 approaches showed that Natural Language Processing features are just the starting point in document processing. The main difficulty is proper analysis of dependencies among words in sentences and am
ong sentences themselves. The obtained results indicate two directions in identification of patterns of those dependencies and a complete diversity in their applications.
(More)