MDI4SE 2019 Abstracts

Full Papers Short Papers
Full Papers
Paper Nr: 2
Title:

Extracting Core Elements of TFM Functional Characteristics from Stanford CoreNLP Application Outcomes
Authors:

Erika Nazaruka, Jānis Osis and Viktorija Griberman
Abstract: Stanford CoreNLP is the Natural Language Processing (NLP) pipeline that allow analysing text at paragraph, sentence and word levels. Its outcomes can be used for extracting core elements of functional characteristics of the Topological Functioning Model (TFM). The TFM elements form the core of the knowledge model kept in the knowledge base. The knowledge model ought to be the core source for further model transformations up to source code. This paper presents research on main steps of processing Stanford CoreNLP application results to extract actions, objects, results and executors of the functional characteristics. The obtained results illustrate that such processing can be useful, however, requires text with rigour, and even uniform, structure of sentences as well as attention to the possible parsing errors.
Download
Paper Nr: 3
Title:

Identification of Causal Dependencies by using Natural Language Processing: A Survey
Authors:

Erika Nazaruka
Abstract: Identification of cause-effect relations in the domain is crucial for construction of its correct model, and especially for the Topological Functioning Model (TFM). Key elements of the TFM are functional characteristics of the system and cause-effect relations between them. Natural Language Processing (NLP) can help in automatic processing of textual descriptions of functionality of the domain. The current research illustrates results of a survey of research papers on identification and extracting cause-effect relations from text using NLP and other techniques. The survey shows that expression of cause-effect relations in text can be very different. Sometimes the same language constructs indicate both causal and non-causal relations. Hybrid solutions that use machine learning, ontologies, linguistic and syntactic patterns as well as temporal reasoning show better results in extracting and filtering cause-effect pairs. Multi cause and multi effect domains still are not very well studied.
Download
Short Papers
Paper Nr: 4
Title:

Vision of the TFM-driven Code Acquisition
Authors:

Vladislavs Nazaruks and Jānis Osis
Abstract: Code acquisition from the system (domain) model completely depends on quality of the model. This paper presents the general vision of the TFM-driven code acquisition. The TFM (Topological Functioning Model) keeps knowledge about the system (domain) functioning, behavior and structure obtained from verbal descriptions of the system (domain). The open question is how this knowledge covers source code constructs. The result shows that, indeed, the final code contain this knowledge, but constructs for representation may differ corresponding to the architectural decisions.
Download