Conceptual Model Generation from Requirements Model: A Natural Language Processing Approach

摘要

In the conceptual modeling stage of the object-oriented development process, the requirements model is analyzed in order to establish the static structure (conceptual model) and dynamic structure (sequence diagrams, state diagrams) of the future system. The analysis is done manually by a requirements engineer, based upon his experience, the system's domain and certain rules of extraction of classes, objects, methods and properties. However, this task could get complicated when large systems are developed. For this reason, a method to automatically generate a conceptual model from the system's textual descriptions of the use case scenarios in Spanish language is presented. Techniques of natural language processing (NLP) and conceptual graphs are used as the basis of the method. The advantage of the proposed method, in comparison to other approaches [1, 2, 3], is that it makes exhaustive use of natural language techniques to face the text analysis. This allows us to consider text with a certain level of ambiguity and to cover relevant linguistic aspects, like composition of nouns, and verbal periphrases.