A robust linguistic infrastructure for efficient web content analysis: the ALVIS project

15 pagesThis paper focuses on the design and the development of a text processing architecture exploiting specialized NLP tools, to produce linguistically annotated documents. This architecture is instanciated using existing NLP modules and resources which need to be tuned to specific domains. Taking as an example the biological domain, we show how a syntactic analyser can be adapted to this domain. We focus on parsing since it exhibits various kinds of adaptation, ranging from unknown words analysis to specific vocabulary (terms, named entities) and structure identification