Application of AI

Citation

Rudall, B.H. (1999), "Application of AI", Kybernetes, Vol. 28 No. 3. https://doi.org/10.1108/k.1999.06728caa.005

Application of AI

Keywords Automation, Cybernetics, Research, Technological developments

Abstract Gives reports and surveys of selected current research and development in systems and cybernetics. They include: Human cells may process information, Artificial brain, Genetics smartcard, Robotic revolution, Application of AI, New devices in medicine, Innovations in cybernetics and systems.

Application of AI

A neural network approach to thesaurus generation

A recent article in CC AI (the journal for the integrated study of artificial intelligence, cognitive science and applied epistemology), Vol. 15 No. 1-2, 1998, pp. 81-111, shows how a neural network approach is made to the generation of a thesaurus. The authors Dimitri G. Roussinov and Hsinchun Chen, both of the School of Management at the University of Arizona, USA, say that:

The rapid proliferation of texual and multimedia online databases, digital libraries, Internet servers, and intranet services has turned researchers' and practitioners' dream of creating an information-rich society into a nightmare of info-gluts. Many researchers believe that turning an info-glut into a useful digital library requires automated techniques for organizing and categorizing large scale information.

Their research is aimed at developing a scaleable texual classification and categorization system based on the Kohonen's self-organising feature map (SOM) algorithm. They show in their work how self-organization can be used for automatic thesaurus generation. The researchers took advantage of the sparsity of coordinates in the document input vectors in setting out their proposed data structure and algorithm and reduced complexity by, they say, several orders of magnitude. The detail of the algorithmic intuition and the mathematical foundation of their work and the algorithm's performances were also discussed by using the data from three benchmarking experiments. Of most interest to cyberneticians and systemists are the studies of the approaches: the serial, statistical approach; the parallel neural network approach; the self-organising map approach.

The neural network approach to the automatic generation is dealt with in some detail as are the benchmarking experiments. The latter experiments illustrate this neural approach and examine the algorithm's performance in: classification of electronic meeting comments; Internet homepages, and Compendex collection.

The article not only illustrates the application of the neural approach but also represents research in the area of full automatic creation of thesauri. It identifies how self-organisation can be successfully used in such generations.

Other articles in the same issue of the CC AI are concerned with advanced document management and the creation of electronic thesauri. They illustrate the different possible paradigms that are identifiable in the range of "knowledge-poor" and "knowledge-rich" approaches. The creation and maintenance remains the main problem of their usage.