The Evolution of Complexity - Abstracts.

Algorithms for the evolution and development of knowledge networks that use common semantics.

By Johan Bollen

  • Vrije Universiteit Brussel
  • Fakulteit Psychologie en Opvoedkunde
  • Principia Cybernetica Project
  • Pleinlaan 2
  • 1050 Brussel
  • tel. + 32 629 25 25
  • fax + 32 629 24 89
  • E-mail:
  • Abstract:

    Knowledge like all other systems evolves by the laws of variation and selection. Ideas, chunks of isolated knowledge, survive in competition with other ideas through the principles of mutation, recombination and reproduction. A strict precondition for the survival and reproduction of ideas and knowledge is communication among its carriers. The Internet as a world wide system of information storage and communication is as such the ultimate 'breeding' ground for the mutation and recombination of ideas. The most popular Internet paradigm at present is that of the World Wide Web: a hypertext and hypermedia protocol in which knowledge is represented by a large, distributed and world wide network of nodes and their interconnections. The evolution of knowledge and ideas on the WWW takes place by the continuous process of changing and replacing of links between related nodes. The knowledge stored and communicated is not fixed and rigid, but continuously changing and adapting. This process is not autonomous but relies almost completely on the interventions of human web designers.

    This paper discusses our attempts to devise algorithms that can make knowledge networks as the WWW autonomously reorganise the knowledge they contain, using and finally absorbing the common semantics of its users. These algorithms consist of a number of very simple and locally applied learning rules that operate on the connections between the nodes in the network. The paths human browsers follow are used to update the strength of the links in the hyper-network, so that an evolutionary process of variation and selection can take place. Every user of the network contributes only a small amount of his/her semantics to the network which will finally resemble the common semantics of all its users.

    We believe these algorithms can make the Internet and more specifically the WWW a truly self-organising system that will finally encompass the whole of human knowledge. This body of knowledge can develop itself, not only by the explicit engineering of its contributors and designers but by the spontaneous and complex interaction among the users and the self-organising knowledge network. We also believe this kind of networks can develop and cultivate original ideas. They will not only store and distribute information, they will actively participate in the process of the formation and evolution of knowledge.