The SYNALP team developed two methods for automatically acquiring a Verbnet like classification of French verbs from three existing verb lexicons (Dicovalence, Treelex and the LADL tables): one method uses Formal Concept Analysis and combines it with statistical filters while the other uses the IGNGF clustering method (Incremental Growing Neural Gas with Feature maximisation)..

The classifications obtained using these two methods can be downloaded from the Download tab.

Both methods use subcategorisation frames extracted from a syntactic lexicon derived from Dicovalence, Treelex and the LADL tables. They classify about 4200 French verbs.

Similar to Verbnet, the classification groups together verbs, subcategorisation frames and thematic grids. However the classification was derived fully automatically and is therefore imperfect.

For more information on how these classifications were derived and on their evaluation, see:

I. Falk, C. Gardent and J.-C. Lamirel Classifying French Verbs using French and English Lexical Resources Proceedings of ACL 2012. Jeju Island Korea, July. PDF

I. Falk and C. Gardent Combining Formal Concept Analysis and Translation to assign Frames and Thematic Grids to French Verbs Proceedings of CLA 2011, the 8th International Conference on Concept Lattices and Their Applications PDF