The use of symbolic information in automation of statistical treatment for ill-structured domains
Karina Gibert Oliveras · AI Communications · 1996
Klass is a clustering tool that can use semantic information to guide the classification process. It is parameterized on the aggregation criteria and the metrics to be used in the classification process. One of its most important features is the use of both qualitative and quantitative information in the object descriptions. Because of their intrinsic characteristics (coexistence of quantitative and qualitative variables the last ones with a great number of modalities additional expert knowledge on the domain structure) ill-structured domains are difficult problems for the actual statistical and artificial intelligence techniques. Briefly, construction of complete knowledge bases of the domain, to be used in diagnostic oriented systems, is almost unreachable due to the complexity of this kind of domains. The clustering (based on distances, which are, in fact, syntactic criteria) has also a poor behaviour; actually, standard statistical techniques were not specifically designed for simultaneous treatment of numerical variables and great quantities of qualitative information. The main goal of this work is to overcome some exhibited limitations of Statistics and Artificial Intelligence techniques referred to in this particular context. Among other results, one may highlight: