D. de Kok, J. Ma, C. Dima, and E. Hinrichs. Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers, page 311--317. Valencia, Spain, Association for Computational Linguistics, (April 2017)
Abstract
Prepostitional phrase (PP) attachment is a well known challenge to parsing. In this paper, we combine the insights of different works, namely: (1) treating PP attachment as a classification task with an arbitrary number of attachment candidates; (2) using auxiliary distributions to augment the data beyond the hand-annotated training set; (3) using topological fields to get information about the distribution of PP attachment throughout clauses and (4) using state-of-the-art techniques such as word embeddings and neural networks. We show that jointly using these techniques leads to substantial improvements. We also conduct a qualitative analysis to gauge where the ceiling of the task is in a realistic setup.
%0 Conference Paper
%1 de-kok-etal-2017-pp
%A de Kok, Daniël
%A Ma, Jianqiang
%A Dima, Corina
%A Hinrichs, Erhard
%B Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers
%C Valencia, Spain
%D 2017
%E Lapata, Mirella
%E Blunsom, Phil
%E Koller, Alexander
%I Association for Computational Linguistics
%K imported myown
%P 311--317
%T PP Attachment: Where do We Stand?
%U https://aclanthology.org/E17-2050
%X Prepostitional phrase (PP) attachment is a well known challenge to parsing. In this paper, we combine the insights of different works, namely: (1) treating PP attachment as a classification task with an arbitrary number of attachment candidates; (2) using auxiliary distributions to augment the data beyond the hand-annotated training set; (3) using topological fields to get information about the distribution of PP attachment throughout clauses and (4) using state-of-the-art techniques such as word embeddings and neural networks. We show that jointly using these techniques leads to substantial improvements. We also conduct a qualitative analysis to gauge where the ceiling of the task is in a realistic setup.
@inproceedings{de-kok-etal-2017-pp,
abstract = {Prepostitional phrase (PP) attachment is a well known challenge to parsing. In this paper, we combine the insights of different works, namely: (1) treating PP attachment as a classification task with an arbitrary number of attachment candidates; (2) using auxiliary distributions to augment the data beyond the hand-annotated training set; (3) using topological fields to get information about the distribution of PP attachment throughout clauses and (4) using state-of-the-art techniques such as word embeddings and neural networks. We show that jointly using these techniques leads to substantial improvements. We also conduct a qualitative analysis to gauge where the ceiling of the task is in a realistic setup.},
added-at = {2023-11-27T12:11:53.000+0100},
address = {Valencia, Spain},
author = {de Kok, Dani{\"e}l and Ma, Jianqiang and Dima, Corina and Hinrichs, Erhard},
biburl = {https://puma.ub.uni-stuttgart.de/bibtex/2c86082e41e61dfb92d1d9a75299315c7/gdima},
booktitle = {Proceedings of the 15th Conference of the {E}uropean Chapter of the Association for Computational Linguistics: Volume 2, Short Papers},
editor = {Lapata, Mirella and Blunsom, Phil and Koller, Alexander},
interhash = {a70855de724cd83c32c4ece3eae780c7},
intrahash = {c86082e41e61dfb92d1d9a75299315c7},
keywords = {imported myown},
month = apr,
pages = {311--317},
publisher = {Association for Computational Linguistics},
timestamp = {2023-11-27T12:17:31.000+0100},
title = {{PP} Attachment: Where do We Stand?},
url = {https://aclanthology.org/E17-2050},
year = 2017
}
