Istituto di Scienza e Tecnologie dell'Informazione     
Sirovich F. Semantic memory in automatic learning of heuristic. Internal note IEI-B72-19, 1972.
The effectiveness of a problem solving system depends on the ability to provide it a symbolic representation of the environment in which it is called to operate, and on heuristics which control the action of the problem solving system depending on such a knowledge representation. Most of the work in Artificial Intelligence has followed the approach of selecting some• particular task and setting LIP a representation of the limited context in which the task is defined, in such a way that heuristics could effectively depend on the meaning of the occurring events as it is captured by the symbolic context representation. In these terms we can say that those systems understand the task and the context in which it is defined, and that they carry out an intelligent activity. When solving new problems an intelligent system is in addition. required to plan, adopt, test and modify problem solving procedures, and therefore some knowledge of them must effectively be present in the system (3): It is again a problem of representation. I find very fascinating the speculation presented by Minsky and Papert in (3) that learning is quite similar to any other intelligent activity and consists in finding a representation which satisfies some goal. Problem solving programs in the past represent a success in dealing with the issue of representing the micro-world in which they were operating, and the serious complaints• about them were their inflexibility and their very limited ability to learn. Coping with these limitations seems now a relevant and worth tackling problem. We would like to see an intelligent system able to generate a problem solving procedure to cope with a situation which is novel to it. Let us define the task of such a system in a less fuzzy way.

Icona documento 1) Download Document PDF

Icona documento Open access Icona documento Restricted Icona documento Private


Per ulteriori informazioni, contattare: Librarian http://puma.isti.cnr.it

Valid HTML 4.0 Transitional