Final report on NLP analysis and normalization
Résumé
This document describes the ALVIS NLP line that has been developed. In ALVIS, it is used to analyze the crawled documents that are then indexed on semantic and domain-specific grounds. It is also to process training corpora in order to acquire the specialized linguistic and domain resources (in WP6) that are then exploited for the semantic analysis of larger document collections. Four languages are addressed in ALVIS: English, French, Slovene and Chinese. Some specific NLP modules have been developed and integrated for each language. The resulting NLP lines have been tested trough several experiments on different domains, on document collections of various size and with various degrees of NLP analysis.