LIGM

The 'Models and Algorithms' Group of LIGM works on automatic processing of natural language texts. We emphasize the use of precise and explicit data: dictionaries, grammars, as compared to the approximation from uncertain data. Applications are numerous, but the most significant are related to information retrieval. The three directions that interest us now are:

Supervised machine learning with large-coverage NLP dictionaries and grammars;
Production of electronic dictionaries in other languages than those for which reliable tools are already available;
Intermediate treatment between lexical analysis and parsing for access to information in large databases of texts: recognition and indexing terminology, disambiguation, etc.;
In the longer term, the continued creation of lexicon-grammar tables, which consist of a systematic and formal description of the syntax of natural languages.

Research Topics

Lexicon-Grammar, Local Grammars, Parsing, Multiwords, Word Sense Disambiguation, Information Extraction, Named Entities, Dictionaries, Multilingualism, Terminology.

Members

Matthieu Constant, Takuya Namakura, Éric Laporte, Claude Martineau, Tita Kyriacopoulou, Cristian Martinez.