0019/2009 - A Survey on Conceptual Modeling

Resumo

Conceptual modeling remains a relevant research topic, even though more than thirty years have passed since Peter Chen enunciated his Entity-Relationship Model. Methods and methodologies for the creation of conceptual models have been the subject of studies and projects which goal is to produce clearer, complete and easier-to-read models. Several methods, modeling languages and tools, have been proposed over the years, some of which aim at creating and/or reading such models automatically, which can imply a simplification that might oppose the idea of semantic accuracy and completeness. The common denominator among all proposals is that, for a conceptual model to be effective and useful, a designer must learn the language used in the Universe of Discourse to be modeled, along with its underlying concepts, and then represent such concepts in a modeling language. Also, no matter the source of information, the knowledge about the scenario to be modeled is always passed to the designer in a natural language. For the resulting model to be both detailed and unambiguous, the modeling language must convey the semantics of such environment, in a way that anyone who is literate in this language can, from reading the model, get the same understanding as from the description in a natural language. In other words, the modeling language must be as rich and generative as the natural language in which the Universe of Discourse concepts are described. Several projects that focus on conceptual modeling have turned to linguistics as a support for the modeling process itself, relating natural language constructs to those of the adopted modeling language; what they all have in common is that their work is done from the perspective of the (meta)model itself. This report presents some of the aforementioned studies and proposes that the linguistic approach, invaluable as it is, should actually be applied to the modeling process from the natural language perspective.