PCP Research on Intelligent Webs
Different participants in the Principia Cybernetica Project are researching algorithms to inject more intelligence into the web, and into distributed knowledge systems in general. The main idea is that the different documents available on the web can be seen as nodes in a network, connected by a number of links, along which variable numbers of users travel in search for information. The resulting directed graph can be enhanced with weights (that, e.g., measure the frequency of usage) and semantic link types, and reorganized in order to provide better access to the knowledge that it contains. This makes it possible to "mine" the implicit knowledge distributed over millions of documents, the connections between them, and the way they are used.
This knowledge can be used to improve the quality of the results that users get in return for their queries, point them towards relevant documents, reorganize the web to make it more efficient, and even discover wholly new concepts and relationships. Thus, the collective intelligence of the millions of users and authors of web information may be leveraged, turning the web from a mere repository of information into an adaptive, self-organizing knowledge system that actively "thinks" in interaction with its users.
Such an intelligent web can be seen as a first step towards the "global brain", which according to our philosophy is the likely next stage in evolution. More generally, this research is inspired by the theoretical concepts and principles from cybernetics, systems, complexity, cognition, evolution and self-organization that underly the whole of the Principia Cybernetica philosophy. As such, it allows us to to formalize and operationalize some parts of the theory, and to test it by experiments and practical applications.
This research is carried out mostly at the Center "Leo Apostel" at the Free University of Brussels (VUB), by Francis Heylighen, Johan Bollen and Alex Riegler, and at Distributed Knowledge Systems Team, by Cliff Joslyn, Luis Rocha and Johan Bollen, at Los Alamos National Laboratory (LANL). The two groups collaborate closely, and keep contact with a number of related groups and centers, such as the Global Brain Group, the Symbiotic Intelligence Project, and Cognitive Technologies, inc. It is possible to join one of these groups as a researcher at the PhD or PostDoc level.
Various algorithms and techniques are being explored within the general paradigm of a distributed, self-organizing knowledge network. One basic paradigm, first developed by the Brussels group, is the one of the learning web, which sees links beween web pages as analogous to synapses in the brain, which adapt their connection strength to the frequency of their usage (the Hebb rule for neural networks). Another application is Luis Rocha's Talkmine system inspired by methods from fuzzy set theory, Pask's conversation theory and collaborative filtering. The two approaches are presently being integrated in the project on Active Recommendation Systems for the Library Without Walls. A final approach developed by the Brussels group is the structuring of networks of concepts by "bootstrapping" the distinctions between concepts. This makes it possible to merge, differentiate or cluster different nodes in a network.
In addition to these systems developed by us, we are exploring other methods to turn the web into an associative network (basically, a weighted, directed graph). A number of methods use existing links or connectivity patterns to compute the importance or "authority" of a web page for a given query. This makes it possible to cluster related pages into "communities" that cover a given subject domain. It is also possible to derive similarity weights between pages by using co-citation of pages or co-occurrence of words. This analysis of web connectivity patterns is complemented by a number of techniques used in collaborative filtering, which uses the similarity between users' preferences to recommend particular pages to particular users.
Copyright© 1999 Principia Cybernetica -
Referencing this page