forest path Knowledge is defined (Oxford English Dictionary) variously as (1) expertise and skills acquired by a person through experience or education; (2) what is known in a particular field or in total; (3) awareness or familiarity gained by experience of a fact or situation. (see wikipedia).

With geographic knowledge, in its first sense, I would like to refer to the "common sense" expertise we gain by experience with objects on a geographical scale, like roads, mountains, buildings, lakes. In its second sense, the term can also be used to refer to empirical scientific theories about objects on a geographical scale, which is the central matter of the earth sciences, geography and regional science. I am particularly interested in the following two topics:

1) Learning and representing geographic knowledge. According to W.V.O. Quine ("The Roots of Reference"), the way to develop theories about our physical world inevitably involves learning basic "observation sentences" that provide the semantic intersubjective anchor of natural languages. These are basic "common sense" propositions about observations (which, if their referents are objects, also imply an agreement about what constitutes an object of a human category), that are shared by a community of speakers in the same situation. It is a key idea that through learning observation sentences, complex human categories as well as complex measurements are actually semantically grounded. Therefore, representing knowledge in terms of observation sentences could provide a solution to the symbol grounding problem of computer science. I'm working on these topics:

  • Practically suitable representation methods. Spatio-temporal geometry and logic.
  • Semantic grounding of symbolic representations. Measurement Theory.
  • Empirical concepts/categories and their philosophy

2) Incomplete data representations. How is geographic knowledge related to existing data models and representation schemes? Actually, geographic data can be seen as a result of (basic) observation. But often it will represent physical geographical objects only in a very incomplete way. So in what respect do these data representations and the common sense or scientific knowledge differ? In particular:

  • Finding reliable and operational (but necessarily incomplete) formal definitions of categories for databases. Compare concept learning.
  • Finding domain theoretical explanations/interpretations for structural properties of data base instances.
  • Which features in data to use? That is, which observations to make?
  • Using these methods to check data consistency and to semantically annotate data bases.

This website reports on my ongoing research in this field.