Summary of the paper
| Title | Learning Recursive Segments for Discourse Parsing |
|---|---|
| Authors | Stergos Afantenos, Pascal Denis, Philippe Muller and Laurence Danlos |
| Abstract | Automatically detecting discourse segments is an important preliminary steptowards full discourse parsing. Previous research on discourse segmentationhave relied on the assumption that elementary discourse units (EDUs) in adocument always form a linear sequence (i.e., they can never be nested).Unfortunately, this assumption turns out to be too strong, for some theories ofdiscourse, like the "Segmented Discourse Representation Theory" or SDRT, allowfor nested discourse units. In this paper, we present a simple approach todiscourse segmentation that is able to produce nested EDUs. Our approach buildson standard multi-class classification techniques making use of a regularizedmaximum entropy model, combined with a simple repairing heuristic that enforcesglobal coherence. Our system was developed and evaluated on the first round ofannotations provided by the French Annodis project (an ongoing effort to createa discourse bank for French). Cross-validated on only 47 documents (1,445EDUs), our system achieves encouraging performance results with an F-score of73% for finding EDUs. |
| Language | Semantics |
| Topics | Discourse annotation, representation and processing, Corpus (creation, annotation, etc.), Semantics |
Full paper  |
Learning Recursive Segments for Discourse Parsing |
| Bibtex | @InProceedings{AFANTENOS10.582, author = {Stergos Afantenos, Pascal Denis, Philippe Muller and Laurence Danlos}, title = {Learning Recursive Segments for Discourse Parsing}, booktitle = {Proceedings of the Seventh conference on International Language Resources and Evaluation (LREC'10)}, year = {2010}, month = {may}, date = {19-21}, address = {Valletta, Malta}, editor = {Nicoletta Calzolari (Conference Chair), Khalid Choukri, Bente Maegaard, Joseph Mariani, Jan Odjik, Stelios Piperidis, Mike Rosner, Daniel Tapias}, publisher = {European Language Resources Association (ELRA)}, isbn = {2-9517408-6-7}, language = {english} } |