Semi-Supervised Never-Ending Learning in Rhetorical Relation Identification. Maziero, E. G., Hirst, G., & Pardo, T. A. S. In Recent Advances in Natural Language Processing (RANLP-2015), pages ???--???, Hissar, Bulgaria, September, 2015. abstract bibtex Some languages do not have enough labeled data to obtain good discourse parsing, specially in the relation identification step, and the additional use of unlabeled data is a plausible solution. A workflow is presented that uses a semi-supervised learning approach. Instead of only a pre-defined additional set of unlabeled data, texts obtained from the web are continuously added. This obtains near human perfomance (0.79) in intra-sentential rhetorical relation identification. An experiment for English also shows improvement in accuracy using a similar workflow.
@inproceedings{Maziero2015RANLP,
author = {Erick Galani Maziero and Graeme Hirst and Thiago A. S. Pardo},
title = {Semi-Supervised Never-Ending Learning in Rhetorical Relation Identification},
address = {Hissar, Bulgaria},
booktitle = {Recent Advances in Natural Language Processing (RANLP-2015)},
pages = {???--???},
year = {2015},
month = {September},
download = {http://ftp.cs.toronto.edu/pub/gh/Maziero-etal-2015-RANLP.pdf},
abstract = {Some languages do not have enough labeled data to
obtain good discourse parsing, specially in the
relation identification step, and the additional use
of unlabeled data is a plausible solution. A
workflow is presented that uses a semi-supervised
learning approach. Instead of only a pre-defined
additional set of unlabeled data, texts obtained
from the web are continuously added. This obtains
near human perfomance (0.79) in intra-sentential
rhetorical relation identification. An experiment
for English also shows improvement in accuracy using
a similar workflow. }
}
Downloads: 0
{"_id":"eHSSCjckKGH2SQHqW","bibbaseid":"maziero-hirst-pardo-semisupervisedneverendinglearninginrhetoricalrelationidentification-2015","downloads":0,"creationDate":"2016-10-19T19:03:54.036Z","title":"Semi-Supervised Never-Ending Learning in Rhetorical Relation Identification","author_short":["Maziero, E. G.","Hirst, G.","Pardo, T. A. S."],"year":2015,"bibtype":"inproceedings","biburl":"http://www.cs.toronto.edu/compling/all_bib.bib","bibdata":{"bibtype":"inproceedings","type":"inproceedings","author":[{"firstnames":["Erick","Galani"],"propositions":[],"lastnames":["Maziero"],"suffixes":[]},{"firstnames":["Graeme"],"propositions":[],"lastnames":["Hirst"],"suffixes":[]},{"firstnames":["Thiago","A.","S."],"propositions":[],"lastnames":["Pardo"],"suffixes":[]}],"title":"Semi-Supervised Never-Ending Learning in Rhetorical Relation Identification","address":"Hissar, Bulgaria","booktitle":"Recent Advances in Natural Language Processing (RANLP-2015)","pages":"???--???","year":"2015","month":"September","download":"http://ftp.cs.toronto.edu/pub/gh/Maziero-etal-2015-RANLP.pdf","abstract":"Some languages do not have enough labeled data to obtain good discourse parsing, specially in the relation identification step, and the additional use of unlabeled data is a plausible solution. A workflow is presented that uses a semi-supervised learning approach. Instead of only a pre-defined additional set of unlabeled data, texts obtained from the web are continuously added. This obtains near human perfomance (0.79) in intra-sentential rhetorical relation identification. An experiment for English also shows improvement in accuracy using a similar workflow. ","bibtex":"@inproceedings{Maziero2015RANLP,\n author = {Erick Galani Maziero and Graeme Hirst and Thiago A. S. Pardo},\n title = {Semi-Supervised Never-Ending Learning in Rhetorical Relation Identification},\n address = {Hissar, Bulgaria},\n booktitle = {Recent Advances in Natural Language Processing (RANLP-2015)},\n pages = {???--???},\n year = {2015},\n month = {September},\n download = {http://ftp.cs.toronto.edu/pub/gh/Maziero-etal-2015-RANLP.pdf},\n abstract = {Some languages do not have enough labeled data to\n obtain good discourse parsing, specially in the\n relation identification step, and the additional use\n of unlabeled data is a plausible solution. A\n workflow is presented that uses a semi-supervised\n learning approach. Instead of only a pre-defined\n additional set of unlabeled data, texts obtained\n from the web are continuously added. This obtains\n near human perfomance (0.79) in intra-sentential\n rhetorical relation identification. An experiment\n for English also shows improvement in accuracy using\n a similar workflow. }\n}\n\n\n","author_short":["Maziero, E. G.","Hirst, G.","Pardo, T. A. S."],"key":"Maziero2015RANLP","id":"Maziero2015RANLP","bibbaseid":"maziero-hirst-pardo-semisupervisedneverendinglearninginrhetoricalrelationidentification-2015","role":"author","urls":{},"downloads":0},"search_terms":["semi","supervised","never","ending","learning","rhetorical","relation","identification","maziero","hirst","pardo"],"keywords":[],"authorIDs":[],"dataSources":["2vBSdbWEoTEQZtb6g"]}