Project leader is Tibor VAMOS

Databases of fuzzy signatures are common tools in every problem where

qualitative human estimates are the sources of information. Due to the

fact that the estimates of domain experts are more valuable than row data,

in every case of medical diagnosis databases human evaluations of

people, economies, etc. cover a representative part of our knowledge

background. Human estimates are represented in fuzzy values and a

composition of those related to an item of the database (a patient, a

country, etc.) is a signature. A highly practical data mining algorithm was

developed that handles the information in a very flexible way for

experimentation with different viewpoints, classification and granulation

methods, weights, supports the discovery of clusters, i.e. common

phenomena and relations, seen from different considerations of

relevance. The system works for a special medical database related to

brain injuries, deficiencies of the newborn and helps in early diagnosis

and therapy with surprising improvement, habilitation results

(methodology Prof. Dr. Katona).