Towards an Intelligent Web

Apart from simple knowledge elicitation, the most obvious practical application of the various network models for knowledge structuring which we have reviewed lies in the World-Wide Web, the hypermedia interface to the information stored on the global Internet computer network. As said, the hypertext organization maps directly onto the directed graphs used in entailment nets or semantic networks. The web is a perfect example of a distributed and collectively constructed system of knowledge. The wealth of information provided by the millions of web documents, however, is offset by the almost total lack of structure in the way these documents are linked. This makes it far from trivial to find the precise information one is looking for, even when that information is within easy reach.

The structuring algorithms which we discussed could be used to make the web simpler, more efficient, less redundant, and more complete, by pointing out lacking or superfluous links and nodes, and by suggesting better connection patterns. In particular the associative learning algorithms we sketched could be applied rather straightforwardly to the web as it now exists (Bollen & Heylighen, 1996; Heylighen & Bollen, 1996), resulting in a dramatic reduction in the average number of links a user needs to traverse in order to find a specific item. The proposed node merging and conceptual clustering techniques may prove useful in creating automatic indexes or review documents, grouping links to similar documents. The semantic network organization should also be able to simplify web browsing by providing a simple and unified set of link types, so that users can be more selective in which links they need to explore. For example, if you are interested in the general class to which a concept belongs, it would be meaningless to spend time exploring its instances.

While such a bootstrapping structuration of the web would help a user navigate through hyperspace, entailment nets may support information retrieval in an even more direct way. The directionality of entailment nets allows non-trivial inferences. Such inferences can be used to answer queries. Queries can be formulated either in a well-structured, formal or in an associative, fuzzy way (cf. Heylighen, 1991a), depending on how clear the problem is for the user. Formal queries are aimed at determining the presence of a particular semantic relation between two or more concepts. For example, "Can a penguin fly?", is a "yes-no" query that needs to be resolved by determining the presence or absence of an entailment from penguin to can fly. An open-ended query like "Which birds cannot fly?" should produce a list of all concepts that entail both bird and cannot fly. Associative queries, on the other hand, do not ask for the presence of specific relations, but for the concepts that are in the most general way associatively related with the concepts the user has in mind, e.g. bird, ice, fish, cute.

Both types of queries can be tackled in a knowledge network by the mechanism of spreading activation (Jones, 1986; Salton & Buckley, 1988; Chen & Ng, 1995): nodes or concepts that are linked to the concepts in the query are "activated". The activation spreads from those nodes through their links to neighbouring nodes, and the nodes which have received the highest activation are brought forward as candidate answers to the query. If none of the proposals are acceptable, those that seem closest to the answer are again activated and used as sources for a new process of spreading. This process is repeated, with the activation moving in parallel from node to node via their links, until a satisfactory solution is found.

In the case of associative queries, the only constraint on spreading activation is the strength of the intervening links: the activation passed through a link is proportional to its strength. The activation arriving in a node is the weighted sum of the activations arriving through all input links of that node. If the associative network is represented by a matrix of link strengths, spreading activation can be implemented by repeatedly multiplying an input vector representing the initial activation for all nodes in the network (typically with values of 1 for the query terms and 0 for all the others) with the matrix, and then selecting the nodes with the highest activation values from the different output vectors. (Chen & Ng, 1995, explore the usefulness of some other spreading activation algorithms for concept retrieval).

We have implemented such a spreading activation program that uses the associative data produced by our learning web experiment (Bollen & Heylighen, 1996b). The program often manages to mimic the "intuitive" reactions of a human subject trying to guess a word from various clues. For example, the input of the clue words control and society produces the word government as most highly activated, while the words room, work and paper produce office. This is similar to the way thoughts diffuse in the brain, moving along intuitive, fuzzy pathways, rather than retrieving exact matches like traditional search engines. Such "inferences" could obviously never have been achieved through logical deductions, since there is no way in which office could have been defined by a Boolean combination of the above query terms.

In the case of formal queries, spreading activation will not be a continuously diffusing intensity, but a discrete state of activation which can only follow certain paths. For example, a query looking for birds that cannot fly should not follow links of the "has part" or "causes" type, but only of the "is a" type. The typical implementation in a semantic network will activate the concepts in the query and let the activation follow all possible links of the right type until the activated paths intersect. This is quite inefficient for proving negative connections, though, since all possible paths need to be explored between penguin and can fly in order to show that none exist. For such situations, it seems better to allow negative entailments or "inhibitory" links, of the type "if penguin, then not (can fly)".

It seems that the introduction of these different types of flexible inference and discovery mechanisms could turn the rapidly developing World-Wide Web from a huge, static repository into an active processor and creator of knowledge (Heylighen & Bollen, 1996). The best metaphor for capturing the collective intelligence formed by millions of users interacting with such a self-organizing, "thinking" web may be the one of a global brain (Russell, 1995). Although many issues still need to be resolved, work is starting in different quarters to turn this science-fiction-like vision into a concrete reality (Goertzel & Pritchard, 1997; Mayer-Kress & Barczys, 1995; Heylighen, 1997b).

Such an implementation at the planetary scale was probably not envisaged by Gordon Pask. Yet, his conversational systems and the present vision of an intelligent web share their view of knowledge as a collective construction striving to achieve coherence, rather than a mapping of external objects. By abandoning the correspondence epistemology and its reliance on fixed primitives, bootstrapping approaches open the way to a truly flexible, adaptive and creative knowledge system. Of course, the systems sketched here are still in their infancy, and need to be thoroughly tested under diverse circumstances, and implemented on a sufficiently large scale to show their practical usefulness. This will obviously require a very large effort. I hope that the work of Gordon Pask, myself and our colleagues will provide sufficient inspiration for other researchers to take up these challenges.

Acknowledgments:

I have been supported during this research by the Fund for Scientific Research-Flanders (FWO), as a Senior Research Associate. I wish to thank Johan Bollen, Cliff Joslyn and others for collaboration and discussions on many of the ideas reviewed in this paper.

References

BAKKER R. R. (1987). Knowledge Graphs: representation and structuring of scientific knowledge, (Ph.D. Thesis, Dep. of Applied Mathematics, University of Twente, Netherlands).

BICKHARD M. & TERVEEN L. (1995). Foundational Issues in Artificial Intelligence and Cognitive Science, (Elsevier, ).

BOLLEN J. & HEYLIGHEN F. (1996) "Algorithms for the Self-organisation of Distributed, Multi-user Networks. Possible application for the future World Wide Web", in: Cybernetics and Systems '96 R. Trappl (ed.), (Austrian Society for Cybernetics), pp. 911-916.

BOLLEN J. & HEYLIGHEN F. (1996b). Finding words through spreading activation [http://pespmc1.vub.ac.be/SPREADACT.html].

BOLLEN J. & HEYLIGHEN F. (1997). "Dynamic and adaptive structuring of the World Wide Web based on user navigation patterns". Proceedings of the Flexible Hypertext Workshop (ACM) (in press).

BRACHMAN, R. J. (1977). "What's in a Concept: Structural Foundations for Semantic Nets", Int. J. Man-Machine Studies 9, 127-152.

CHEN H. & NG T. (1995). "An Algorithmic Approach to Concept Exploration in a Large Knowledge Network (Automatic Thesaurus Consultation): Symbolic Branch-and-bound Search vs. Connectionist Hopfield Net Activation", Journal of the American Society for Information Science 46: 5, 348-369.

CZIKO G. (1995). Without Miracles. Universal Selection Theory and the Second Darwinian Revolution, (MIT Press, Cambridge, MA).

FISHER D.H. (1987). "Knowledge Acquisition via Incremental Conceptual Clustering", Machine Learning 2(7), 139-172.

FORD K.M. & HAYES P.J. (Eds.) (1991). Reasoning Agents in a Dynamic World: The Frame Problem , (JAI Press, Greenwich CT).

GOERTZEL B. & PRITCHARD J. (1997) "About webMind", comoc "white paper" #1 [http://www.comoc.com/About-Webmind.html].

HARNAD S. (1990). "The Symbol Grounding Problem", Physica D 42, 335-346.

HEYLIGHEN F. & BOLLEN J. (1996) "The World-Wide Web as a Super-Brain: from metaphor to model", in: Cybernetics and Systems '96 R. Trappl (ed.), (Austrian Society for Cybernetics).p. 917-922.

HEYLIGHEN F. (1987). "Towards a General Framework for Modelling Representation Changes", in: Proc. 11th Int. Congress on Cybernetics, (Assoc. Internat. de Cybernétique, Namur), pp. 136-141.

HEYLIGHEN F. (1989). "A Support System for Structuring Complex Problem Domains", in: Proc. 12th Int. Congress on Cybernetics (Assoc. Intern. de Cybernétique, Namur), pp. 1025-1032.

HEYLIGHEN F. (1990a). "A Structural Language for the Foundations of Physics", International Journal of General Systems 18, 93-112.

HEYLIGHEN F. (1990b). Representation and Change. An Integrative Metarepresentational Framework for the Foundations of Physical and Cognitive Science, (Communication & Cognition, Gent).

HEYLIGHEN F. (1990c). "Classical and Non-classical Representations in Physics I", Cybernetics and Systems 21, 423-444.

HEYLIGHEN F. (1990d). "Relational Closure: a mathematical concept for distinction-making and complexity analysis", in: Cybernetics and Systems '90, R. Trappl (ed.), (World Science, Singapore), pp. 335-342.

HEYLIGHEN F. (1991a). "Design of a Hypermedia Interface Translating between Associative and Formal Representations", International Journal of Man-Machine Studies 35, 491-515.

HEYLIGHEN F. (1991b). "Structuring Knowledge in a Network of Concepts", in: Workbook of the 1st Principia Cybernetica Workshop, Heylighen F. (ed.) (Principia Cybernetica, Brussels), pp. 52-58. [http://pespmc1.vub.ac.be/Papers/StructuringKnowledge.html]

HEYLIGHEN F. (1992a). "From Complementarity to Bootstrapping of Distinctions: A Reply to Löfgren's Comments on my Proposed 'Structural Language", International Journal of General Systems 21(1), 99.

HEYLIGHEN F. (1992b). "Principles of Systems and Cybernetics: an evolutionary perspective", in: Cybernetics and Systems '92, R. Trappl (ed.), (World Science, Singapore), pp. 3-10.

HEYLIGHEN F. (1993). "Selection Criteria for the Evolution of Knowledge", in: Proc. 13th Int. Congress on Cybernetics (Association Internat. de Cybernétique, Namur), pp. 524-528.

HEYLIGHEN F. (1997a). "Objective, subjective and intersubjective selectors of knowledge", Evolution and Cognition [in press]

HEYLIGHEN F. (1997b). "The Global Brain Group", [http://pespmc1.vub.ac.be/GBRAIN-L.html]

HEYLIGHEN F., ROSSEEL E. & DEMEYERE F. (eds.) (1990). Self-Steering and Cognition in Complex Systems. Toward a New Cybernetics, (Gordon and Breach Science Publishers, New York).

HEYLIGHEN F., JOSLYN C. & TURCHIN V. (1997). Principia Cybernetica Web [http://pespmc1.vub.ac.be/].

JONES W. P. (1986). "On the Applied Use of Human Memory Models", International Journal of Man-Machine Studies 25, 191-228.

JOSLYN C., HEYLIGHEN F. & BOLLEN J. (1997). Systems Concepts [http://pespmc1.vub.ac.be/SYSCONC.html].

JOSLYN C., HEYLIGHEN F. & TURCHIN V. (1993). "Synopsys of the Principia Cybernetica Project", in: Proc. 13th Int. Congress on Cybernetics (Association Internationale de Cybernétique, Namur), pp. 509-513.

LöFGREN, L. (1991). "Complementarity in Language: Toward a General Understanding." In Carvallo, ed. Nature, Cognition and System II: Complementarity and Beyond. Dordrecht, Boston, London: Kluwer, pp. 73-104 .

MATURANA H. R., & VARELA F. J. (1992). The Tree of Knowledge: The Biological Roots of Understanding, (rev. ed.), (Shambhala, Boston).

MAYER-KRESS G. & C. BARCZYS (1995). "The Global Brain as an Emergent Structure from the Worldwide Computing Network", The Information Society 11 (1), 1-28.

MINSKY M. (1975): "A framework for representing Knowledge", in: The Psychology of Computer Vision, Winston P. (ed.), (McGraw-Hill, New York).

PASK G. & GREGORY D. (1986). "Conversational Systems", in: Human Productivity Enhancement (vol. II), Zeidner R. J. (ed.), (Praeger), pp. 204-235

PASK G. (1975). Cognition, Conversation and Learning, (Elsevier, Amsterdam).

PASK G. (1976). Conversation theory: applications in education and epistemology, Elsevier : Amsterdam.

PASK G. (1980). "Developments in Conversation Theory--Part 1", Int. J. Man-Machine Studies 13, 357-411.

PASK G. (1984). "Review of Conversation Theory and a Protologic (or Protolanguage) Lp", Educational Communications and Technology 32, no. 1, pp. 3-40.

PASK G. (1990). "Some Formal Aspects of Conversation Theory and Lp", in: Heylighen F., Rosseel E. & Demeyere F. (eds.). Self-Steering and Cognition in Complex Systems. Toward a New Cybernetics, (Gordon and Breach Science Publishers, New York), pp. 240-247.

PASK G. (1991). "The Foundations of Conversation Theory, Interaction or Actors Theory, all Cybernetic and Philosophically so", in: Heylighen F. (ed.), Workbook of the 1st Principia Cybernetica Workshop, (Principia Cybernetica, Brussels), pp. 15-18. [http://pespmc1.vub.ac.be/Papers/Workbook.html]

PIAGET J. (1937). La Construction du Réel chez l'Enfant, (Delachaux & Niestlé, Neuchâtel).

POPPER K. (1959). The Logic of Scientific Discovery, (Hutchinson, London).

RESCHER N. (1973) Coherence Theory of Truth, London, Oxford University Press.

ROCHA L. (1991). "Fuzzification of Conversation Theory", (presented at the 1st Principia Cybernetica Workshop, Free University of Brussels).

RUSSELL, P. (1995). The Global Brain Awakens: Our Next Evolutionary Leap (Miles River Press).

SALTON G. & BUCKLEY C. (1988). "On the Use of Spreading Activation Methods in Automatic Information Retrieval", Proc. 11th Ann. Int. ACM SIGIR Conf. on R&D in Information Retrieval (ACM), pp. 147-160.

SHASTRI L. (1988). Semantic Networks (Morgan Kauffman, Los Angeles).

SOWA J.F. (ed.) (1991). Principles of Semantic Networks: explorations in the representation of knowledge (Morgan Kauffman, Los Angeles)

SPENCER BROWN G. (1969). Laws of Form, (Allen & Unwin, London).

STOKMAN F.N. & DE VRIES P.H. (1988). "Structuring Knowledge in a Graph", in: Human-Computer Interaction, Psychonomic Aspects, G.C. van der Veer & G.J. Mulder (eds.), (Springer, Berlin).

THAGARD, P. (1989) Explanatory Coherence. Behavioral and Brain Sciences 12: 435-467.

TURCHIN V. (1993). "On Cybernetic Epistemology", Systems Research 10:1, 3-28.

VAN BRAKEL, J. (1992) The Complete Description of the Frame Problem. PSYCOLOQUY 3(60) frame-problem.2

VON FOERSTER H. (1996) Cybernetics of Cybernetics (2nd edition). (Future Systems, Minneapolis)

VON GLASERSFELD E. (1984). "An Introduction to Radical Constructivism", in: The Invented Reality, P. Watzlawick (ed.), (Norton, New York).