Comprendere un documento con il computer

In this talk, I shall deal with the complex process of computer reading and understanding of text documents as the result of the interleaving of a number of levels of processing, both linguistic and extra-linguistic, such as parsing, classifying, learning and knowing. In real tasks, there exists no parsing without classifying, no classifying without learning, no learning without knowing and eventually no knowing without "doing things with words". Only through robust integration and co-operation of less than optimal components and inter-disciplinary cross-fertilization we can hope to develop general and comprehensive solutions which are more than the sums of their parts.