A Fundamental Problem
(or: a Unifying Theme)
for Knowledge Representation
& Reasoning

1. The task of a KRR system is to represent and reason about (and to enable action on) some domain (= "the world")

2. This might take the form of a computer system to help people
   • E.g., a geographic information system, a question-answering system, a knowledge base (i.e., an AI version of a database)

   or it might take the form of a computational cognitive agent.

3. In either case, there might be more than one source of information:
   • multiple senses
   • multiple informants

4. In the case of multiple senses, some of the information from one sense might originate in some domain-object, and some of the information from another sense might originate in the same domain-object:
   • Cf. the "binding" problem of neuroscience:
     • binding the color, shape, etc., of a physical object into an experience of a single object with many properties instead of experiences of multiple objects each with only one property

5. In the case of multiple informants, there are 2 problems:
   1. How to evaluate the trustworthiness of the informants
      • Leads to belief-revision/truth-maintenance issues
      • This also applies to sensory input, e.g., in color perception

   2. How to correlate their information
      • Cf. the binding problem
      • E.g., are the informants talking about the same thing?
        • Leads to issues in:
          • node merging/splitting
            • Maida, Anthony S., & Shapiro, Stuart C. (1982), "Intensional Concepts in Propositional Semantic Networks", Cognitive Science 6: 291-330; reprinted in Ronald J. Brachman & Hector J. Levesque (eds.), Readings in Knowledge Representation (Los Altos, CA: Morgan Kaufmann, 1985): 169-189.
            • Maida, Anthony S. (1991), "Maintaining Mental Models of Agents Who Have Existential Misconceptions", Artificial Intelligence 50(3): 331-383.

          • identifying/distinguishing perceptually (in)distinguishable objects
            • Santore, John F., & Shapiro, Stuart C. (2004), "A Cognitive Robotics Approach to Identifying Perceptually Indistinguishable Objects", in Alan Schultz (ed.), The Intersection of Cognitive Science and Robotics: From Interfaces to Intelligence, Papers from the 2004 AAAI Fall Symposium, Technical Report FS-04-05 (Menlo Park, CA: AAAI Press): 47-54.

6. Another name for all this: multi-source information fusion
   • Garfinkel, Simson L. (2008), "Information of the World, Unite!", Scientific American 299(3) (September): 82-87.
     • Note: This online HTML version is titled "Data Fusion: The Ups and Downs of All-Encompassing Digital Profiles" and omits all the graphics.
     • A password-protected PDF version of the original article with graphics is also available.

7. A related ontological problem:
   • What if 2 KRR systems (e.g., an information source and "your" KRR system)...
     • ...use 2 different ontologies?
     • ...use 2 different languages?

   • In both cases, need some sort of translation scheme.

   • Real-life e.g.: a problem for geographic information systems
     • "western" geographical-land-form ontology vs. indigenous Australian geographical-land-form ontology
     • See: Ethnophysiography
     • See:
       Mark, David M.; & Turk, Andrew G. (2003), "Landscape Categories in Yindjibarndi: Ontology, Environment, and Language", Conference on Spatial Information Theory (COSIT'03), in W. Kuhn, M. Worboys, & S. Timpf (eds.) Spatial Information Theory: Foundations of Geographic Information Science (Berlin: Springer Lecture Notes in Computer Science 2825).

8. Solutions:
   • Use an interlingua (especially for machine translation)
   • Use the Knowledge Interchange Format (KIF)
   • Use "ontological mediation"
     • Campbell, Alistair E., & Shapiro, Stuart C. (1998), "Algorithms for Ontological Mediation", in S. Harabagiu (ed.), Usage of WordNet in Natural Language Processing Systems: Proceedings of the Workshop, COLING-ACL 1998: 102-107.

   • Use negotiation
     • Rapaport, William J. (2003), "What Did You Mean by That? Misunderstanding, Negotiation, and Syntactic Semantics", Minds and Machines 13(3): 397-427. [PDF]