The ANTHEM representation formalism for the alphabetic index of ICD.

This paper describes a formalism in which the knowledge of the Alphabetic Index of ICD is expressed unambiguously in the ANTHEM prototype. The Alphabetic Index may be viewed as a collection of hypotheses on which ICD-codes should be tried first for a given diagnostic expression. The hypotheses generation can be based upon characteristics of the semantic representation of the diagnostic expression. The formalism is described using an "operational" semantics by referring to the processes that have to operate on the expressions in the knowledge base. A close integration between a sound linguistic model of diagnostic expressions and the formalism itself is realized.