Enhancing CASE Environments by Using Linguistics

In this paper it is argued that CASE environments could and should be enhanced considerably by using theories and knowledge from linguistics. The environments should 'know' about the language of their users and the domains they are used for. By basing the modeling techniques supported by the CASE tool on linguistic theories and by incorporating Natural Language parsing and generating tools, the CASE environment is able to handle the users' language in an accurate way. More specifically, the CASE environment deals with the meaning of words, instead of the meaningless strings themselves. These meanings, which are retrieved from an online lexicon, are linked to the words used in both the requirements documents as well as in the conceptual models in order to achieve a certain degree of consistency between the two of them. The base structure of these models is automatically derived by analyzing the textual requirements documents that describe the domain under consideration. This Natural Language analysis consists of parsing the texts and retrieving the word meanings that corresponds to concepts that may be of interest for modeling this domain accurately. Furthermore, the resulting models can be validated by people who are not familiar with the modeling notations used, by paraphrasing the models to Natural Language sentences. This paper mainly focuses on the profits gained by using linguistic knowledge in CASE environments, on the philosophy behind this approach and on three specific Natural Language components: the lexicon, Natural Language analysis and text generation for requirements validation. This article is based on the paper "Truly Intelligent CASE Environments Profit from Linguistics" [7], which was presented during the SEKE conference in Madrid (June 1997).