Automatic Processes in Database Building and Subsequent Automatic Abstractions

Abstract

One of the difficulties in building an effective system for spatial and attribute generalization is the predominance of the conventionally based generalization paradigm. This is exemplified by numerous attempts at reproducing rules derived from manual environments and re-building them in a GIS or CIS. What is often lacking is an understanding of what generalization achieves in its various forms. There is no doubt that the representation issue in generalization is of critical importance, since it is through the representation that we communicate visually. However, with access to databases and high-level programming tools, the quality, consistency, and reliability of the information warrants far greater emphasis than has been given in the past. Re-addressing generalization from a database perspective shifts the focus in model development from cartographic representation issues to geographic information. However, doing so places demands on operational mapping organizations for database building, data classification, and subsequent feature coding. Recently, techniques have been developed to automatically classify, code, and build databases. Since the impediment to generalization has rested partially in establishing a logical database, the approach discussed in this chapter is to identify techniques that allow users to create the database automatically, without having to code the features interactively. Three different data sets being used for experimentation include hydrography, transportation networks, and land-cover. The models differ somewhat, given the nature of the object categories. Hydrographic and transportation-network features can be classified and coded using a rule base and graph theory, while land cover can be handled using classification and aggregation hierarchies. The objective of this research is to load unclassified categories of features, derive a series of classifications automatically, build a database automatically, and subsequently generalize the data using the derived database to any smaller scale of representation. The objective of this chapter is to present the methodologies required to achieve the research goals and to illustrate the results.