A DISTANCE MEASURE FOR DECISION-MAKING IN UNCERTAIN DOMAINS

A novel definition of syntactic distance between structural symbolic descriptions is proposed. It is based on a probabilistic interpretation of the canonical matching predicate. By means of this distance measure it is possible to cope with the problem of matching noise affected descriptions or imprecise rules. Furthermore, an extension of the syntactic distance which manages incomplete descriptions is presented. Finally, the application of the syntactic distance to the problem of classifying digitized office documents by using their page layout description is shown.