IGC - evaluation set 20.09

  Authors

Barkarson, Starkaður ; Steingrímsson, Steinþór ; Andrésdóttir, Þórdís Dröfn and Hafsteinsdóttir, Hildur

  Item identifier

http://hdl.handle.net/20.500.12537/51

 Project URL

http://igc.arnastofnun.is

 Date issued

2020-09-14

 Type

corpus, text

 Size

101261 tokens

 Language(s)

Icelandic

 Description

The evaluation set contains 101.261 tokens and is divided into nine subcorpora: adjudications, books, educational websites, legal tests, news, opinions, parliamentary speeches, sport news and radio and tv news scripts. The texts were retrieved randomly from Icelandic Gigaword Corpus (version 2018) and pos-tagged with the ABL-tagger that had been trained on MIM-GOLD 20.05. It was then flagged by using four complementary methods and the flagged tags then manually checked. Each line contains a word and a pos-tag, separated by a tab. Each sentence is separated by a newline. Þetta prófunarsett inniheldur 101,261 tóka og skiptist í níu undirmálheildir: dóma, bækur, fræðslumiðla, lög, fréttir, skoðanir, þingræður, íþróttafréttir og handrit að sjónvarps- og útvarpsfréttum. Textarnir voru valdir af handahófi úr Risamálheildinni (2018) og markaðir með ABL-markaranum sem hafði verið þjálfaður á MIM-GOLD 20.05. Vafaatriði voru merkt með fjórum mismunandi aðferðum og yfirfarin handvirkt. Hver lína hefur að geyma orð og mark, aðgreind með dálkmerki. Hver málsgrein er sér í línu.

 Publisher

The Árni Magnússon Institute for Icelandic Studies

 Acknowledgement

Ministry of Education, Science and Culture

Project code: Icelandic Gigaword Corpus (G1)

Project name: Language Technology for Icelandic 2019-2023

 Subject(s)

evaluation set morphosyntactic tagging gigaword corpus pos-tagging

 Collection(s)

Clarin IS

Show full item record