4. Resources for automatic text processing

NLP systems rely on several types of resources:

• textual resources: textual data, known as corpora, are used to create test benches, to train learning systems, to extract lexical data, etc. ...;

• lexical resources: lexicons form the core of the linguistic information used by a system. They vary in nature depending on the application, and incorporate more or less complex information, from simple word lists to structured semantic resources. Given the cost involved in building a lexicon for a given application, the trend is towards reusability and automatic acquisition of lexical data;

• software resources: lemmatizers, segmenters and labelers are the basic building blocks of text processing. The complexity of a NLP application calls for the reusability of existing components; this...