Processing Use Case Scenarios and Text in a Formal Style as Inputs for TFM-based Transformations

Erika Nazaruka
2020 Baltic Journal of Modern Computing  
TFM (Topological Functioning Model) based transformations start from text fragments as inputs and end with source code. Automated processing of use case scenarios is likely to be more predictable than text in a formal style thanks to their structure. The goal of the research is to understand whether the differences in processing these two text forms are essential for getting core elements of a TFM, or even a structured form has essential limitations. The theoretical results illustrate that use
more » ... ase specifications may have more structured and less structured formats. Even in the former format, use case steps may contain explanations and even text fragments in a formal style that increases unpredictability. Analysis of text in the both cases requires part-of-speech tagging, lemmas, constituency and dependency parsing, coreference resolution, and language pattern matching. Thus, structuring the initial documents is questionable but cases when they are to be managed in projects. Keywords: use cases, text in a formal style, natural language processing, topological functioning model, computation independent model, model transformation, knowledge acquisition Processing Use Case Scenarios and Text in a Formal Style 49 computation independent. They could be descriptions in prose (instructions, position descriptions, interview protocols), descriptions in structured text (e.g., use case scenarios, user stories, templates for requirements), graphical schemes (e.g., use case models, business process models in different notations such as BPMN (Business Process Model and Notation), structural models in different notation such as Entity-Relationship diagrams) and mathematical or physical formulas. All these means do not include any computation dependent information. An ideal CIM contains complete unambiguous knowledge. However, CIM completeness and unambiguity of representations are questionable since descriptions inherit natural language ambiguity, as well as schemes usually provide a fragmentary view on a problem domain or represent just one or several aspects of it. There must be a model that can serve as a ground onto which gathered knowledge could be projected and verified and as a starting point for further automated transformations. A model that has these abilities is a Topological Functioning Model (TFM). The TFM is based on principles of system theory and algebraic topology. It specifies a system in a holistic manner, showing its inner functionality and interaction with external systems at the high level of abstraction. The TFM can be manually (but according to precise rules) transformed into most used UML diagram types: class diagrams, activity diagrams, use cases with their specifications (Osis and Asnina, 2011a) and Topological UML (Donins et al., 2011) diagrams such as Topological Class diagrams, Topological Use Case diagrams, Activity diagrams, State Chart diagrams, Sequence and Communication diagrams (Osis and Donins, 2010) . Application of the principles of the theoretical foundation of the model leads to discovering complete knowledge and verifying its accuracy. Sources of knowledge for the TFM differ in their representation formats, structure or its absence, and have different readiness for automated processing and projecting to the TFM. Automated processing of input sources is crucial since additional modelling requires additional resources like staff, budget and time. Automated processing in comparison with manual allows reducing time needed. However, what knowledge and of what quality could be processed and projected to the TFM is a question that requires additional research. The aim of this research is to clarify what an input form of knowledge sources, a text in a formal style or a use case scenario, has essential advantages for the TFM at the present. In this research focus is put on automated processing that includes less parsing and transforming notation elements but more application of Natural Language Processing (NLP) for acquiring and verifying knowledge from the corresponding text. The paper is organized as follows. Section 2 presents brief overview of the TFM and its place and role in TFM-based transformations. Section 3 discusses initial theoretical results on natural language processing in prose in a formal style and in a numbered step form of use case scenarios. Section 4 gives a brief overview in related work. Conclusion presents main results and speculations on issues and further research. Topological Functioning Model Based Transformations Chain of Transformations The meta-picture of the TFM driven transformations (Figure 1) illustrates a general vision of the TFM driven transformations. There are two groups of the input, i.e.,
doi:10.22364/bjmc.2020.8.1.03 fatcat:hkjetiecmvfxrajux76f7ixt44