Error Mining on Syntactic Parser Output

We introduce an error mining technique for automatically detecting errors in resources that are used in parsing systems. We applied this technique to parsing results produced on several million words by two distinct parsing systems, which share the syntactic lexicon and the pre-parsing processing chain. We are thus able to identify incorrectness and incompleteness sources in the resources. In particular, by comparing both systems’ results, we are able to isolate problems coming from shared resources from those coming from grammars. MOTS-CLÉS : analyse syntaxique, lexique syntaxique, fouille d’erreurs.