Artificially Evolved Chunks for Morphosyntactic Analysis

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F19%3A10427089" target="_blank" >RIV/00216208:11320/19:10427089 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.aclweb.org/anthology/W19-7815" target="_blank" >https://www.aclweb.org/anthology/W19-7815</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Artificially Evolved Chunks for Morphosyntactic Analysis

Popis výsledku v původním jazyce

We introduce a language-agnostic evolutionary technique for automatically extracting chunksfrom dependency treebanks. We evaluate these chunks on a number of morphosyntactic tasks,namely POS1 tagging, morphological feature tagging, and dependency parsing. We test the utilityof these chunks in a host of different ways. We first learn chunking as one task in a shared multitaskframework together with POS and morphological feature tagging. The predictions from thisnetwork are then used as input to augment sequence-labelling dependency parsing. Finally, weinvestigate the impact chunks have on dependency parsing in a multi-task framework. Our resultsfrom these analyses show that these chunks improve performance at different levels of syntacticabstraction on English UD treebanks and a small, diverse subset of non-English UD treebanks.

Název v anglickém jazyce

Artificially Evolved Chunks for Morphosyntactic Analysis

Popis výsledku anglicky

We introduce a language-agnostic evolutionary technique for automatically extracting chunksfrom dependency treebanks. We evaluate these chunks on a number of morphosyntactic tasks,namely POS1 tagging, morphological feature tagging, and dependency parsing. We test the utilityof these chunks in a host of different ways. We first learn chunking as one task in a shared multitaskframework together with POS and morphological feature tagging. The predictions from thisnetwork are then used as input to augment sequence-labelling dependency parsing. Finally, weinvestigate the impact chunks have on dependency parsing in a multi-task framework. Our resultsfrom these analyses show that these chunks improve performance at different levels of syntacticabstraction on English UD treebanks and a small, diverse subset of non-English UD treebanks.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

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

Projekt

—

Návaznosti

—

Rok uplatnění

2019

Kód důvěrnosti údajů

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