An Interactive System for Stem-Suffix Discrimination in Italian Words

Today, owing to the growing diffusion of on-line processing facilities, the automatic processing of textual data, i.e. of information expressed in natural language, becomes more and more important both for: 1) applications concerning specifically the processing of textual fragments, (studies in linguistic analysis, in information retrieval and question-answering systems, etc.); 2) the implementation of a viable communication process within interactive systems also for problems not specifically linguistic, (such as computer aided instruction, computer aided design, etc.), so that non specialized people can access and use the system without the burden of some more or less rigidly coded command language. A typical application which combines both of the above aspects can be found in a hospital where the collection and the analysis of clinical data is to be performed automatically. This will be accomplished by a combined hardware-software system capable to process and analyze clinical data expressed in narrative form, e.g. the patients' anamneses; it should furthermore enable the sanitary staff of the hospital to use some form of on-line communication language as explicit as possible. Any significant processing of information expressed in natural language requires one be able to perform some kind of content analysis on the input data, and to infer some significant associations between these data and particular actions which are considered pertinent in a given context. For instance, in the case of the hospital system mentioned before, the content analysis of patients' anamneses should make it possible to relate the textual data composing an anarmlesis with the clinical aspects considered useful for the possible diagnosis and therapy.