What I think when I think about treebanks

Anders Søgaard

What I think when I think about treebanks

Abstract

In this opinion piece, I present four somewhat controversial suggestions for the design of futuretreebanks: a) Treebanks should be based on adversarial samples, rather than pseudorepresentativesamples. b) Treebanks should include multiple splits of the data, rather than justa single split, as in most treebanks today. c) They should include multiple annotations of eachsentence, whenever possible, instead of adjudicated annotations. d) There is no real motivationfor adhering to a notion of well-formedness, since we now have parsers based on deep learningthat generalize easily and perform well on any type of graphs, and treebanks therefore do not haveto limit themselves to trees or directed acyclic graphs.

Original language	English
Title of host publication	Proceedings of the 16th International Workshop on Treebanks and Linguistic Theories (TLT16),
Publisher	Association for Computational Linguistics
Publication date	2018
Pages	161-166
Publication status	Published - 2018
Event	16th International Workshop on Treebanks and Linguistic Theories (TLT16) - Prague, Czech Republic Duration: 23 Jan 2018 → 24 Jan 2018

Conference

Conference	16th International Workshop on Treebanks and Linguistic Theories (TLT16)
Country/Territory	Czech Republic
City	Prague
Period	23/01/2018 → 24/01/2018

Access to Document

http://www.aclweb.org/anthology/W17-7620Licence: CC BY

Cite this

@inproceedings{f9a2988225994f04b55fc27e63b10ddc,

title = "What I think when I think about treebanks",

abstract = "In this opinion piece, I present four somewhat controversial suggestions for the design of futuretreebanks: a) Treebanks should be based on adversarial samples, rather than pseudorepresentativesamples. b) Treebanks should include multiple splits of the data, rather than justa single split, as in most treebanks today. c) They should include multiple annotations of eachsentence, whenever possible, instead of adjudicated annotations. d) There is no real motivationfor adhering to a notion of well-formedness, since we now have parsers based on deep learningthat generalize easily and perform well on any type of graphs, and treebanks therefore do not haveto limit themselves to trees or directed acyclic graphs.",

author = "Anders S{\o}gaard",

year = "2018",

language = "English",

pages = "161--166",

booktitle = "Proceedings of the 16th International Workshop on Treebanks and Linguistic Theories (TLT16),",

publisher = "Association for Computational Linguistics",

note = "16th International Workshop on Treebanks and Linguistic Theories (TLT16) ; Conference date: 23-01-2018 Through 24-01-2018",

}

TY - GEN

T1 - What I think when I think about treebanks

AU - Søgaard, Anders

PY - 2018

Y1 - 2018

N2 - In this opinion piece, I present four somewhat controversial suggestions for the design of futuretreebanks: a) Treebanks should be based on adversarial samples, rather than pseudorepresentativesamples. b) Treebanks should include multiple splits of the data, rather than justa single split, as in most treebanks today. c) They should include multiple annotations of eachsentence, whenever possible, instead of adjudicated annotations. d) There is no real motivationfor adhering to a notion of well-formedness, since we now have parsers based on deep learningthat generalize easily and perform well on any type of graphs, and treebanks therefore do not haveto limit themselves to trees or directed acyclic graphs.

AB - In this opinion piece, I present four somewhat controversial suggestions for the design of futuretreebanks: a) Treebanks should be based on adversarial samples, rather than pseudorepresentativesamples. b) Treebanks should include multiple splits of the data, rather than justa single split, as in most treebanks today. c) They should include multiple annotations of eachsentence, whenever possible, instead of adjudicated annotations. d) There is no real motivationfor adhering to a notion of well-formedness, since we now have parsers based on deep learningthat generalize easily and perform well on any type of graphs, and treebanks therefore do not haveto limit themselves to trees or directed acyclic graphs.

M3 - Article in proceedings

SP - 161

EP - 166

BT - Proceedings of the 16th International Workshop on Treebanks and Linguistic Theories (TLT16),

PB - Association for Computational Linguistics

T2 - 16th International Workshop on Treebanks and Linguistic Theories (TLT16)

Y2 - 23 January 2018 through 24 January 2018

ER -