Introduction to Artificial Inteligence - Spring 2001

Building a Knowledge Base

1. Desidarata for a knoweldge base
   • Clear and correct
   • Sufficiently general
   • Vocabulary and axioms usable multiple times
   • Without requiring overhaul for minor extensions!
   • Should be independent of inference procedure
2. Knowledge engineering
   • Decide what to talk about
   • Decide on vocabulary: predicates, functions, constants
   • Encode general knowledge about the domain
   • Debug: encode problem instance, pose queries
3. Combinational logic circuits domain
   • What to talk about: gates, wires, signals
   • Vocabulary: Gate Type(gate), Connected(x, y), function - Signal(wire)
   • Encoding a specific instance
   • Example queries
4. General ontology
   • Categories (is-a hierarchy)
   • Measures
   • Composite objects (partonomy)
   • Representation of change
   • Physical objects
   • Substances
   • Mental objects and belief
5. Example: grocery shopping world