Datasets:

Sprakbanken
/

ngram

first string	second string	third float64	lang float64	year int64	count int64
!	!	null	null	1,800	96
!	!	null	null	1,801	81
!	!	null	null	1,802	88
!	!	null	null	1,803	65
!	!	null	null	1,804	71
!	!	null	null	1,805	69
!	!	null	null	1,806	81
!	!	null	null	1,807	84
!	!	null	null	1,808	136
!	!	null	null	1,809	93
!	!	null	null	1,810	95
!	!	null	null	1,811	54
!	!	null	null	1,812	95
!	!	null	null	1,813	77
!	!	null	null	1,814	302
!	!	null	null	1,815	119
!	!	null	null	1,816	119
!	!	null	null	1,817	114
!	!	null	null	1,818	110
!	!	null	null	1,819	117
!	!	null	null	1,820	112
!	!	null	null	1,821	151
!	!	null	null	1,822	131
!	!	null	null	1,823	78
!	!	null	null	1,824	119
!	!	null	null	1,825	99
!	!	null	null	1,826	203
!	!	null	null	1,827	122
!	!	null	null	1,828	144
!	!	null	null	1,829	142
!	!	null	null	1,830	102
!	!	null	null	1,831	168
!	!	null	null	1,832	187
!	!	null	null	1,833	75
!	!	null	null	1,834	177
!	!	null	null	1,835	177
!	!	null	null	1,836	155
!	!	null	null	1,837	511
!	!	null	null	1,838	587
!	!	null	null	1,839	779
!	!	null	null	1,840	1,345
!	!	null	null	1,841	860
!	!	null	null	1,842	764
!	!	null	null	1,843	1,033
!	!	null	null	1,844	1,090
!	!	null	null	1,845	630
!	!	null	null	1,846	693
!	!	null	null	1,847	906
!	!	null	null	1,848	2,396
!	!	null	null	1,849	1,373
!	!	null	null	1,850	1,479
!	!	null	null	1,851	3,140
!	!	null	null	1,852	3,130
!	!	null	null	1,853	2,558
!	!	null	null	1,854	2,769
!	!	null	null	1,855	2,141
!	!	null	null	1,856	2,459
!	!	null	null	1,857	2,729
!	!	null	null	1,858	4,732
!	!	null	null	1,859	8,496
!	!	null	null	1,860	5,599
!	!	null	null	1,861	1,591
!	!	null	null	1,862	2,423
!	!	null	null	1,863	3,694
!	!	null	null	1,864	1,698
!	!	null	null	1,865	2,147
!	!	null	null	1,866	1,196
!	!	null	null	1,867	2,672
!	!	null	null	1,868	1,953
!	!	null	null	1,869	3,248
!	!	null	null	1,870	4,312
!	!	null	null	1,871	3,109
!	!	null	null	1,872	4,100
!	!	null	null	1,873	4,032
!	!	null	null	1,874	6,182
!	!	null	null	1,875	10,653
!	!	null	null	1,876	11,691
!	!	null	null	1,877	12,388
!	!	null	null	1,878	6,635
!	!	null	null	1,879	9,936
!	!	null	null	1,880	8,465
!	!	null	null	1,881	6,319
!	!	null	null	1,882	9,157
!	!	null	null	1,883	10,211
!	!	null	null	1,884	12,706
!	!	null	null	1,885	14,653
!	!	null	null	1,886	19,077
!	!	null	null	1,887	31,610
!	!	null	null	1,888	26,242
!	!	null	null	1,889	29,461
!	!	null	null	1,890	26,938
!	!	null	null	1,891	46,693
!	!	null	null	1,892	67,628
!	!	null	null	1,893	75,979
!	!	null	null	1,894	103,480
!	!	null	null	1,895	93,465
!	!	null	null	1,896	66,745
!	!	null	null	1,897	73,184
!	!	null	null	1,898	122,259
!	!	null	null	1,899	87,415

End of preview. Expand in Data Studio

Dataset Card for N-grams from NBdigital

This dataset contains n-grams (uni-, bi- and trigrams) from all books and newspapers digitized by the National Library of Norway before 2022-07-15. The N-grams are made based on approximately 610 000 books and 4 000 000 newspapers. In total, it's about 138.5 billion "tokens" (i.e. words and punctuation).

Dataset Details

Dataset Description

This dataset contains frequency information on word and punctuation usage in Norwegian. In particular, it contains uni- bi and trigram counts for each year.

Curated by: The Norwegian Language Bank at the National Library of Norway
Shared by: The Norwegian Language Bank at the National Library of Norway
Language(s) (NLP): nob, nno
License: CC-0

Dataset Sources

Source dataset: https://www.nb.no/sprakbanken/ressurskatalog/oai-nb-no-sbr-76/
Paper: Birkenes MB, Johnsen LG, Lindstad AM, Ostad J. From digital library to n-grams: NB N-gram. InProceedings of the 20th Nordic Conference of Computational Linguistics (NODALIDA 2015) 2015 May (pp. 293-295).
Demo: https://dh.nb.no/run/corpus-webapp/app/

Dataset Structure

The N-gram dataset is separated into several subsets for different corpora (digavis and digibok). The digavis dataset contains N-gram counts from newspapers and the digibok datasets contains N-gram counts from books. Moreover, the dataset contains the following fields:

first: The first token in the N-gram
second (optional): The second token in the N-gram (for bi- and trigrams)
third (optional): The third token in the N-gram (for trigrams)
lang (optional): The language of the documents the N-grams are counted from (only for digibok)
year: The year
count: The number of times the given N-gram occures in the given year

For example, in the 2022-digavis-unigram dataset, we have the row

first="hei", year=1972, count=4460

This means that the word "hei" occured 4460 times in the National Library of Norway's digitized newspapers from 1972. Similarly, if we consider the row

first="hei", second="til", lang="nob", year=1990, count=60

in the 2022-digibok-trigram dataset, then that means that the phrase "hei til" occured 60 times in the National Library of Norway's digitized books in Norwegian Bokmål from 1990.

Dataset Creation

This data stems from all books and newspapers in the National Library of Norway's digitized collection, which consists of almost all books and newspapers published in Norway. Note that much of the text stems from automatic text recognition of scanned book and newspaper pages, which may contain mistakes.

Citation

BibTeX:

@inproceedings{birkenes2015digital,
  title={From digital library to n-grams: NB N-gram},
  author={Birkenes, Magnus Breder and Johnsen, Lars G and Lindstad, Arne Martinus and Ostad, Johanne},
  booktitle={Proceedings of the 20th Nordic Conference of Computational Linguistics (NODALIDA 2015)},
  pages={293--295},
  year={2015}
}

Dataset Card Contact

sprakbanken [at] nb [dot] no

Downloads last month: 887

Size of downloaded dataset files:

43.1 GB

Size of the auto-converted Parquet files:

43.1 GB

Number of rows:

18,009,973,760