In order to handle information into the natural language processing systems the morphological categorization of parts of speech POS must be represented in two kinds of forms: external - most human readable form and internal - most computer readable form. This document presents a General Model that contains an external form to represent morphological categorization based on attribute value trees AVT and two internal forms based on XML: one for general morphological information and one for particular information attached to a POS. Finally it is presented a LIR model that can efficiently be used for strong inflected languages to generate inflected forms of POS, to spell and annotate POS in a text and to full indexing a text.
展开▼