Integrating an Explicit Knowledge Model into Geographic Information Systems

Geographic Information Systems (GIS) require the power of sophisticated database management systems for efficiently managing the persistency and consistency of huge amounts of data. A pool of software tools takes care of the production process, which consists of editing, visualising and selecting geographic data. However, the quality of the data – its well-formedness and integrity – is not so easily ensured. Quality is represented by declarative expressions that constrain the relations between concepts of the geographic domain knowledge. However, there exists a mismatch between this implicit model which takes into account the spatial and multi-dimensional character of the data, and the format in which the geographic data is actually stored in the database. Current practices have shown that hard coding these quality ensuring constraints using software engineering results in a loss of expressiveness and even accuracy. Therefore, in this research, we envisioned a co-existence of the usual components of a GIS with an explicit representation of the geographic domain knowledge and constraints, for which we employed knowledge engineering techniques. In this paper we illustrate and validate our ideas at the hand of the GeoObjects System, a GIS we developed together with our industrial partner, TeleAtlas.