Dependency Parsing as Sequence Labeling with Head-Based Encoding and Multi-Task Learning

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F19%3A10427138" target="_blank" >RIV/00216208:11320/19:10427138 - isvavai.cz</a>

Result on the web

<a href="https://www.aclweb.org/anthology/W19-7716" target="_blank" >https://www.aclweb.org/anthology/W19-7716</a>

DOI - Digital Object Identifier

—

Result language

angličtina

Original language name

Dependency Parsing as Sequence Labeling with Head-Based Encoding and Multi-Task Learning

Original language description

Dependency parsing as sequence labeling has recently proved to be a relevant alternative to thetraditional transition- and graph-based approaches. It offers a good trade-off between parsing accuracyand speed. However, recent work on dependency parsing as sequence labeling ignore thepre-processing time of Part-of-Speech tagging – which is required for this task – in the evaluationof speed while other studies showed that Part-of-Speech tags are not essential to achieve state-ofthe-art parsing scores. In this paper, we compare the accuracy and speed of shared and stackedmulti-task learning strategies – as well as a strategy that combines both – to learn Part-of-Speechtagging and dependency parsing in a single sequence labeling pipeline. In addition, we proposean alternative encoding of the dependencies as labels which does not use Part-of-Speech tags andimproves dependency parsing accuracy for most of the languages we evaluate.

Czech name

—

Czech description

—

Type

O - Miscellaneous

CEP classification

—

OECD FORD branch

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Project

—

Continuities

—

Publication year

2019

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů