Section: Overall Objectives
The aim of the LED team is to investigate semantic phenomena (broadly construed) in natural languages from a computational perspective. Concretely, LED's goal is to develop wide coverage grammars (with a special emphasis on French) with a semantic dimension, to explore the linguistic and computational issues involved in putting these grammars to work (for both analysis and generation), and in particular to investigate the interplay between representation and inference.
Typical of the questions we wish to address are: what types of representations are needed (for example, to detect paraphrases or to represent temporal information)? What level of representation is needed (for some purposes, only shallow representations in weak logics are called for, while for others powerful logics seem to be required)? What kind of knowledge is needed to support the identified inferences? What type of inference is needed to guide the representation-building and reasoning processes? For example, when is statistical inference enough? When does it suffice to work with weak (decidable) logics? When are more powerful forms of inference (such as first- or higher-order theorem proving and model building) required?
To put it another way, LED focuses on computational semantics . This has always been an important theme at LED but, partly as a result of personnel changes in the team, its role has become increasingly dominant during last years. Two recent changes have further accentuated this trend. First, in September 2006, Laurent Romary left the team to work in Berlin, and in December 2006 Bertrand Gaiffe left to work at ATILF (the center for ``Analyse et Traitement Informatique de la Langue Fran caise,'' a large research center in Nancy, with which LED has good contacts, devoted to the development of lexical resources).