Processing rhetorical, morphosyntactic, and semantic features from corporate technical documents for identifying organizational domain knowledge

摘要

During the requirements elicitation (RE) process, transformations among languages occur from natural language--in which the stakeholders express their domain and needs--to a controlled language. One source of domain information to be used by the transformation process is related to technical documents belonging to the organizations (e.g. technical reports, legacy documents, and procedure manuals). Some properties of such documents are: different representation formats, high degree of ambiguity, and particular linguistics elements. The analysis and processing of such documents becomes complex because of these properties and, in turn, the complexity makes difficult both identifying the domain knowledge and understanding the associated processes. As a solution for an automated transformation, in this paper we define linguistics features to enable identification and composition of information units from a procedure manual, as a central task for the language transformation process in RE. The identified features are classified into rhetorical, morphosyntactic, and semantic features. Also, the features can be used to identify and represent organizational domain knowledge.