Genre Annotation of Web Corpora: Scheme and Issues

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216224%3A14330%2F21%3A00118741" target="_blank" >RIV/00216224:14330/21:00118741 - isvavai.cz</a>

Výsledek na webu

<a href="https://link.springer.com/book/10.1007/978-3-030-63128-4" target="_blank" >https://link.springer.com/book/10.1007/978-3-030-63128-4</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/978-3-030-63128-4_55" target="_blank" >10.1007/978-3-030-63128-4_55</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Genre Annotation of Web Corpora: Scheme and Issues

Popis výsledku v původním jazyce

Unlike traditional corpora made from printed media in the past decades, sources of web corpora are not categorised and described well, thus making it difficult to control the content of the corpus. This paper presents an attempt to classify genres in a large English web corpus through supervised learning. A set of genres suitable for web corpora users is defined based on a research of related work. A genre annotation scheme with active learning rounds is introduced. A collection of web pages representing various genres that was created for this task and a scheme of consequent human annotation of the data set is described. Measuring the inter-annotator agreement revealed that either the problem may not be well defined, or that our expectations concerning the precision and recall of the classifier cannot be met. Eventually, the project was postponed at that point. Possible solutions of the issue are discussed at the end of the paper.

Název v anglickém jazyce

Genre Annotation of Web Corpora: Scheme and Issues

Popis výsledku anglicky

Unlike traditional corpora made from printed media in the past decades, sources of web corpora are not categorised and described well, thus making it difficult to control the content of the corpus. This paper presents an attempt to classify genres in a large English web corpus through supervised learning. A set of genres suitable for web corpora users is defined based on a research of related work. A genre annotation scheme with active learning rounds is introduced. A collection of web pages representing various genres that was created for this task and a scheme of consequent human annotation of the data set is described. Measuring the inter-annotator agreement revealed that either the problem may not be well defined, or that our expectations concerning the precision and recall of the classifier cannot be met. Eventually, the project was postponed at that point. Possible solutions of the issue are discussed at the end of the paper.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

60203 - Linguistics

Projekt

<a href="/cs/project/GA18-23891S" target="_blank" >GA18-23891S: Hyperintensionální usuzování nad texty přirozeného jazyka</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)<br>S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2021

Kód důvěrnosti údajů

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

Název statě ve sborníku

Proceedings of the Future Technologies Conference (FTC) 2020, Volume 1

ISBN

9783030631277

ISSN

2194-5357

e-ISSN

2194-5365

Počet stran výsledku

17

Strana od-do

738-754

Název nakladatele

Springer Nature Switzerland AG

Místo vydání

Vancouver, Canada

Místo konání akce

Vancouver, Canada

Datum konání akce

5. 11. 2020

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—