G-BRACKETS: A NEW GIS FORMALISM FOR FEATURE DATA ENCODING AND QUERYING

In this paper, we outline the bridges that connect the area of GIS knowledge representation and its querying languages. Our work is in fact grounded in a research framework which we had founded to come out with an appropriate GIS query language that is adapted enough to this kind of information systems having a very particular aspects. Such aspects are unfortunately not yet fully exploited by the existing solutions. In our study, we were led to consider the existing systems and standards for the representation of GIS information, and we quote in this case GML and ESRI. We have reached at the end of this study the conviction that it is impossible to build a query language dealing with our basic aim if we ignore the representation of the GIS database to query. Convinced of the robustness of the XML representation, but aware of his extremely verbose and difficult to read character, we had then to take inspiration from XML technology while correcting this verbosity problem and the underlying complexity. We present then in this paper a new formalism that is able to interest, engage and influence the business of geographic information. Our formalism "G-Brackets" has to respond strongly to our initial aims which can be summarized in the following goals: - A solid and hierarchical formalism with the Manner of XML technology. - An object oriented representation. - A knowledge based representation according to first-order predicates logic, ready to be requested by an inference engine able to go beyond an ordinary SQL engine. - In the end, a simple and less verbose symbolism that is readable and able to separate between the data model and its semantics. Such pattern has constituted for us a real extension of the MVC design pattern that we named MVX pattern.