Knowledge Representation of Cartographic Generalization

Data, information and knowledge are the three inseparable levels in info-flow. Data are a serial of descriptions of discrete facts and geospace from the ontological view, which are considered as records in database. Information that conducts user’s viewpoint to specific matter is a collection of facts or data. While knowledge exist in our brains which are the fusion of experience, information and apprehension. We consider that knowledge is the highest description of geospatial phenomena, processes and concepts, and we use it to interpret, predict and communicate. Knowledge representation is an important branch of computer automation, which is the foundation of knowledge reasoning and AI. Map generalization is a decision process with many artificial intervenes. How to represent spatial knowledge and how to convert the export’s knowledge and experience in brain to formal knowledge is a key of the implementation of automated map generalization, and it is still a long run. Knowledge about map generalization is experience and summarization about the abstraction, generalization and characterization of spatial information. The paper presents the approaches of knowledge representation of map generalization and the principles for expert system, then analysis the type of map generalization knowledge and divide it into three classes: description knowledge and facts, reasoning and process knowledge, evaluation knowledge. The paper illustrates representation of knowledge about map generalization based on rules and models. Rule representation uses the selection and action operations to simulate the process of cartographer’s manipulations with some geometric and semantic constraints. Model representation deals with map through the essential information transmission about map models, and is divided into deep model generalization (map model) and shallow model generalization (symbol representation). Finally the paper develops the organization of knowledge and expert system in map generalization in details.