GRASP Semantic Parsing of Natural Language Assistance Requests

Grasp is a natural language semantic analyser specifically designed to handle ordinary users assistance requests utterred in situations where they need help in front of various software applications ans services. This semantic analyser is based upon a corpus of ~11 000 assisting requests registered over the period 2005-2006 (see excerpt below). Grasp is currently under construction and is developped in Mathematica from Wolfram Research.

— GRASP is part of the DAFT project
— Page of François Bouchet

— Home page of the TreeTagger part of speech tagger of H. Schmid
— Home page of the Penn TreeBank project
— Home page of Wordnet Documentation

Demos pages

Sample demo of the GRASP analyser Samples taken from the Grasp Corpus requests and associated documentation (in French)
Interactive demo of the GRASP analyser Type and anlyse your own requests! (works only when the LIMSI server is running)

Documentation

Paper	F. Bouchet, Characterization of Conversational Activities in a Corpus of Assistance Requests, 21st European Summer School in Logic, Language and Information, Student Session, Bordeaux France, July 20-31, 2009
Paper	F. Bouchet, J-P Sansonnet, Tree kernel and Feature Vector Methods for Formal Semantic Requests Classification, Poster presentation, International Conference on Machine Learning and Data Mining MLDM´2009, Leipzig Germany,july 23-25 2009
Report	J-P. Sansonnet, The Grasp Analyzer, a corpus-based approach for a syntactical analyzer dedicated to the domain of the Function of Assistance, Technical report, LIMSI-AMI, july 2009
Report	F. Bouchet, J-P. Sansonnet, Characterization of Conversational Activities in a Corpus of Assisting Requests, Technical noteTN-G-07.03, LIMSI-AMI, march 2007
Slides	J-P. Sansonnet, Study of a Corpus of Requests for an Assisting Conversational Agent, LIPN Seminar, Paris XIII University, Paris 14 jan 2009
Report	J-P. Sansonnet, A tutorial about ordinary english analysis, Technical Report [obsolete but funny], july 2003
HowTo Report	J-P. Sansonnet, How to use TreeTagger with Mathematica,Technical Notebook, [obsolete but funny], (mail) june 2002.
HowTo Report	J-P. Sansonnet, How to use Wordnet with Mathematica,Technical Notebook (mail or see the full integration in Mathematica 6.0)

Documentation in french

Article	F. Bouchet, Caractérisation d’un Corpus de Requêtes d’Assistance, RECITAL'07, Toulouse, 5-8 juin 2007
Article	F. Bouchet, J-P. Sansonnet, Caractérisation de Requêtes d’Assistance à partir de corpus, MFI'07, Paris, 30 mai-1er juin 2007
Article	F. Bouchet, J-P. Sansonnet, Étude d’un corpus de requêtes en langue naturelle pour des agents assistants, Workshop WACA'06 Toulouse, oct 2006
Master	F. Bouchet, Conception d’un langage de requêtes pour un agent conversationnel assistant, Rapport M2R Informatique, Université de Paris XI, sept 2006
Rapport	F. Bouchet, Spécifications du langage DAFT 2.0, Rapport technique, LIMSI, Sept 2006
Slides	J.P. Sansonnet, Etude d’un Corpus de Requêtes d’Assistance en Langue Naturelle, séminaire GT ACA, 15 mars 2006

The Grasp corpora

Corpus "Grasp" J-P Sansonnet	Extrait du corpus Grasp\| [1000 phrases extraites aléatoirement à partir des ~8000] Exemples d'analyses sur le corpus Grasp [analyse des constituants de 100 phrases extraites aléatoirement à partir des ~8000] Extrait du fichier de définition de la sémantique lexicale (synsets) associée au corpus DAFT.
Corpus "Marco" J-P Sansonnet	Corpus aphabétique Corpus classé par thèmes Description des réactions de Marco La KBase de Marco construite à partir des questions déjà recueillies Les FAQ de Marco extraites des questions déjà recueillies
Corpus "WORDTEX"" J-P Sansonnet	Corpus aphabétique [2889 phrases recueillies à partir des FAQ des manuels et des sites webs des logiciels de traitement de texte : Microsoft Word® et Latex]