Datasets:
id
int64 783k
26.6M
| langvar
int64 1
1
| txt
stringlengths 1
29
| txt_degr
stringlengths 1
20
| meaning
int64 40.2k
37.6M
| langvar_uid
stringclasses 1
value |
---|---|---|---|---|---|
26,111,254 | 1 | qafar af | qafaraf | 36,299,577 | aar-000 |
26,111,255 | 1 | qafar | qafar | 36,299,577 | aar-000 |
26,238,566 | 1 | Afaraf | afaraf | 36,867,197 | aar-000 |
26,238,566 | 1 | Afaraf | afaraf | 36,867,198 | aar-000 |
1,684,561 | 1 | dʼa | da | 19,080,613 | aar-000 |
1,453,573 | 1 | tah | tah | 2,675,740 | aar-000 |
26,069,335 | 1 | Eroppah | eroppah | 36,229,630 | aar-000 |
1,452,951 | 1 | abe | abe | 2,675,921 | aar-000 |
1,452,951 | 1 | abe | abe | 2,675,924 | aar-000 |
1,453,204 | 1 | ellel | ellel | 2,675,879 | aar-000 |
1,689,216 | 1 | áf | af | 19,080,854 | aar-000 |
1,453,024 | 1 | amoyta | amoyta | 2,676,263 | aar-000 |
1,453,024 | 1 | amoyta | amoyta | 2,676,284 | aar-000 |
1,453,024 | 1 | amoyta | amoyta | 2,676,362 | aar-000 |
18,715,653 | 1 | 3Ng~u | 3ngu | 20,033,955 | aar-000 |
18,725,077 | 1 | 3Xub | 3xub | 20,033,958 | aar-000 |
18,730,646 | 1 | 3bul | 3bul | 20,033,961 | aar-000 |
18,599,285 | 1 | 3nu | 3nu | 20,031,129 | aar-000 |
18,657,467 | 1 | 3r3b | 3r3b | 20,033,932 | aar-000 |
18,699,874 | 1 | 3r3b3 | 3r3b3 | 20,033,948 | aar-000 |
18,754,267 | 1 | 3ro | 3ro | 20,033,976 | aar-000 |
18,604,033 | 1 | 3tu | 3tu | 20,031,130 | aar-000 |
18,651,926 | 1 | 3yuft3 | 3yuft3 | 20,033,929 | aar-000 |
23,303,833 | 1 | Afrikah | afrikah | 36,229,626 | aar-000 |
23,303,833 | 1 | Afrikah | afrikah | 28,508,376 | aar-000 |
4,540,577 | 1 | Almazán | almazan | 1,189,814 | aar-000 |
3,723,643 | 1 | Arabic | arabic | 4,444,606 | aar-000 |
3,727,298 | 1 | English | english | 4,444,708 | aar-000 |
6,097,848 | 1 | Enqlizxsh - English | enqlizxshenglish | 1,673,301 | aar-000 |
3,728,248 | 1 | French | french | 4,444,731 | aar-000 |
3,726,575 | 1 | German | german | 4,444,687 | aar-000 |
3,730,652 | 1 | Italian | italian | 4,444,794 | aar-000 |
6,097,849 | 1 | My Happy Ending | myhappyending | 1,641,679 | aar-000 |
5,895,368 | 1 | Simon Doull | simondoull | 1,542,123 | aar-000 |
19,667,027 | 1 | United States | unitedstates | 31,737,076 | aar-000 |
18,663,119 | 1 | X3b3l | x3b3l | 20,033,934 | aar-000 |
18,786,519 | 1 | X3lE | x3le | 20,033,990 | aar-000 |
1,453,056 | 1 | a saku | asaku | 2,675,661 | aar-000 |
1,453,056 | 1 | a saku | asaku | 28,095,618 | aar-000 |
1,621,690 | 1 | ab | ab | 19,081,215 | aar-000 |
1,621,690 | 1 | ab | ab | 19,081,216 | aar-000 |
1,621,690 | 1 | ab | ab | 19,081,586 | aar-000 |
1,621,690 | 1 | ab | ab | 19,082,248 | aar-000 |
18,663,121 | 1 | ab3la | ab3la | 20,033,934 | aar-000 |
1,452,946 | 1 | abatasa | abatasa | 2,675,628 | aar-000 |
1,452,946 | 1 | abatasa | abatasa | 28,082,280 | aar-000 |
17,463,567 | 1 | abaːl | abal | 19,081,631 | aar-000 |
1,452,948 | 1 | abba-bada | abbabada | 2,675,828 | aar-000 |
1,452,948 | 1 | abba-bada | abbabada | 2,675,830 | aar-000 |
1,452,948 | 1 | abba-bada | abbabada | 28,138,052 | aar-000 |
1,452,948 | 1 | abba-bada | abbabada | 28,138,118 | aar-000 |
1,452,949 | 1 | abbah-ina | abbahina | 2,675,979 | aar-000 |
1,452,950 | 1 | abbire | abbire | 2,675,655 | aar-000 |
23,065,322 | 1 | abeesa | abeesa | 28,030,564 | aar-000 |
1,452,953 | 1 | able | able | 2,676,421 | aar-000 |
1,452,954 | 1 | aboyya | aboyya | 2,675,977 | aar-000 |
1,452,956 | 1 | abu | abu | 2,675,636 | aar-000 |
1,452,956 | 1 | abu | abu | 2,676,154 | aar-000 |
1,452,956 | 1 | abu | abu | 28,219,762 | aar-000 |
1,452,957 | 1 | abuke | abuke | 2,676,120 | aar-000 |
18,730,647 | 1 | abul | abul | 20,033,961 | aar-000 |
1,452,958 | 1 | abur | abur | 2,676,358 | aar-000 |
1,452,959 | 1 | aburo | aburo | 2,675,998 | aar-000 |
1,452,959 | 1 | aburo | aburo | 2,676,409 | aar-000 |
1,720,679 | 1 | abäs | abas | 19,081,273 | aar-000 |
1,697,941 | 1 | abäsáː | abasa | 19,080,809 | aar-000 |
2,292,288 | 1 | abəːsə́ːmaː | abəsəma | 19,080,684 | aar-000 |
17,463,691 | 1 | abəːyáː | abəya | 19,082,154 | aar-000 |
1,716,512 | 1 | abə́la | abəla | 19,080,826 | aar-000 |
17,463,690 | 1 | abə́ːyaː | abəya | 19,082,153 | aar-000 |
1,452,952 | 1 | abʼha | abha | 2,676,345 | aar-000 |
17,463,448 | 1 | adar | adar | 19,081,450 | aar-000 |
17,463,448 | 1 | adar | adar | 19,081,458 | aar-000 |
17,463,448 | 1 | adar | adar | 19,081,460 | aar-000 |
1,452,963 | 1 | adarras | adarras | 2,675,918 | aar-000 |
1,452,966 | 1 | addah | addah | 2,675,889 | aar-000 |
1,452,967 | 1 | addal | addal | 2,675,819 | aar-000 |
1,452,970 | 1 | adige | adige | 2,676,309 | aar-000 |
1,452,972 | 1 | admo | admo | 2,675,756 | aar-000 |
1,452,973 | 1 | admo abe | admoabe | 2,675,757 | aar-000 |
1,452,975 | 1 | adure | adure | 2,676,276 | aar-000 |
1,652,805 | 1 | af | af | 19,080,854 | aar-000 |
1,652,805 | 1 | af | af | 19,081,737 | aar-000 |
1,652,805 | 1 | af | af | 19,081,929 | aar-000 |
1,733,474 | 1 | af maː-liː | afmali | 19,080,963 | aar-000 |
17,463,477 | 1 | affaʼra | affara | 19,081,487 | aar-000 |
1,703,015 | 1 | agaboːytá-t angəːg | agaboytatangəg | 19,080,889 | aar-000 |
1,715,806 | 1 | agaboːytát ábbaː | agaboytatabba | 19,080,697 | aar-000 |
1,614,621 | 1 | agaboːytát-iná | agaboytatina | 19,080,699 | aar-000 |
1,613,161 | 1 | agaboːytáː | agaboyta | 19,080,650 | aar-000 |
1,613,161 | 1 | agaboːytáː | agaboyta | 19,080,660 | aar-000 |
1,681,354 | 1 | agaboːytáː ak raːbtä́ nəːm | agaboytaakrabtanəm | 19,080,711 | aar-000 |
1,452,980 | 1 | agabu | agabu | 2,675,931 | aar-000 |
17,463,705 | 1 | agam | agam | 19,082,215 | aar-000 |
1,452,981 | 1 | agda | agda | 2,676,246 | aar-000 |
1,695,672 | 1 | aggaʼf | aggaf | 19,080,934 | aar-000 |
1,452,982 | 1 | aggile | aggile | 2,675,736 | aar-000 |
1,452,982 | 1 | aggile | aggile | 2,675,831 | aar-000 |
1,452,982 | 1 | aggile | aggile | 2,675,839 | aar-000 |
1,452,984 | 1 | agime | agime | 2,676,001 | aar-000 |
Dataset Card for panlex-meanings
This is a dataset of words in several thousand languages, extracted from https://panlex.org.
Dataset Details
Dataset Description
This dataset has been extracted from https://panlex.org (the 20240301
database dump) and rearranged on the per-language basis.
Each language subset consists of expressions (words and phrases). Each expression is associated with some meanings (if there is more than one meaning, they are in separate rows).
Thus, by joining per-language datasets by meaning ids, one can obtain a bilingual dictionary for the chosen language pair.
- Curated by: David Dale (@cointegrated), based on a snapshot of the Panlex database (https://panlex.org/snapshot).
- Language(s) (NLP): The Panlex database mentions 7558 languages, but only 6241 of them have at least one entry (where entry is a combination of expression and meaning), and only 1012 have at least 1000 entries. These 1012 languages are tagged in the current dataset.
- License: CC0 1.0 Universal License, as explained in https://panlex.org/license.
Dataset Sources [optional]
- Original website: https://panlex.org/
- Paper: Kamholz, David, Jonathan Pool, and Susan M. Colowick. 2014. PanLex: Building a Resource for Panlingual Lexical Translation. Proceedings of the 9th International Conference on Language Resources and Evaluation (LREC 2014).
Uses
The intended use of the dataset is to extract bilingual dictionaries for the purposes of language learning by machines or humans.
The code below illustrates how the dataset could be used to extract a bilingual Avar-English dictionary.
from datasets import load_dataset
ds_ava = load_dataset('cointegrated/panlex-meanings', name='ava', split='train')
ds_eng = load_dataset('cointegrated/panlex-meanings', name='eng', split='train')
df_ava = ds_ava.to_pandas()
df_eng = ds_eng.to_pandas()
df_ava_eng = df_ava.merge(df_eng, on='meaning', suffixes=['_ava', '_eng']).drop_duplicates(subset=['txt_ava', 'txt_eng'])
print(df_ava_eng.shape)
# (10565, 11)
print(df_ava_eng.sample(5)[['txt_ava', 'txt_eng', 'langvar_uid_ava']])
# txt_ava txt_eng langvar_uid_ava
# 7921 калим rug ava-002
# 3279 хІераб old ava-001
# 41 бакьулълъи middle ava-000
# 9542 шумаш nose ava-006
# 15030 гӏащтӏи axe ava-000
Apart from these direct translations, one could also try extracting multi-hop translations (e.g. enrich the direct Avar-English word pairs with the word pairs that share a common Russian translation). However, given that many words have multiple meaning, this approach usually generates some false translations, so it should be used with caution.
Dataset Structure
The dataset is split by languages, denoted by their ISO 639 codes. Each language might contain multiple varieties; they are annotated within each per-language split.
To determine a code for your language, please consult the https://panlex.org webside. For additional information about a language, you may also want to consult https://glottolog.org/.
Each split contains the following fields:
id
(int): id of the expressionlangvar
(int): id of the language variaetytxt
(str): the full text of the expressiontxt_degr
(str): degraded (i.e. simplified to facilitate lookup) textmeaning
(int): id of the meaning. This is the column to join for obtaining synonyms (within a language) or translations (across languages)langvar_uid
(str): more human-readable id of the language (e.g.eng-000
stands for generic English,eng-001
for simple English,eng-004
for American English). These ids could be looked up in the language dropdown at https://vocab.panlex.org/.
Dataset Creation
This dataset has been extracted from https://panlex.org (the 20240301
database dump) and automatically rearranged on the per-language basis.
The rearrangement consisted of the following steps:
- Grouping together the language varieties from the
langvar
table with the samelang_code
. - For each language, selecting the corresponding subset from the
expr
table. - Joining the selected set with the
denotation
table, to get themeaning
id. This increases the number of rows (for some languages, x5), because multiple meannings may be attached to the same expression.
Bias, Risks, and Limitations
As with any multilingual dataset, Panlex data may exhbit the problem of under- and mis-representation of some languages.
The dataset consists primarily of the standard written forms ("lemmas") of the expressions, so it may not well represent their usage within a language.
Citation
Kamholz, David, Jonathan Pool, and Susan M. Colowick. 2014. PanLex: Building a Resource for Panlingual Lexical Translation. Proceedings of the 9th International Conference on Language Resources and Evaluation (LREC 2014).
BibTeX:
@inproceedings{kamholz-etal-2014-panlex,
title = "{P}an{L}ex: Building a Resource for Panlingual Lexical Translation",
author = "Kamholz, David and
Pool, Jonathan and
Colowick, Susan",
editor = "Calzolari, Nicoletta and
Choukri, Khalid and
Declerck, Thierry and
Loftsson, Hrafn and
Maegaard, Bente and
Mariani, Joseph and
Moreno, Asuncion and
Odijk, Jan and
Piperidis, Stelios",
booktitle = "Proceedings of the Ninth International Conference on Language Resources and Evaluation ({LREC}'14)",
month = may,
year = "2014",
address = "Reykjavik, Iceland",
publisher = "European Language Resources Association (ELRA)",
url = "http://www.lrec-conf.org/proceedings/lrec2014/pdf/1029_Paper.pdf",
pages = "3145--3150",
abstract = "PanLex, a project of The Long Now Foundation, aims to enable the translation of lexemes among all human languages in the world. By focusing on lexemic translations, rather than grammatical or corpus data, it achieves broader lexical and language coverage than related projects. The PanLex database currently documents 20 million lexemes in about 9,000 language varieties, with 1.1 billion pairwise translations. The project primarily engages in content procurement, while encouraging outside use of its data for research and development. Its data acquisition strategy emphasizes broad, high-quality lexical and language coverage. The project plans to add data derived from 4,000 new sources to the database by the end of 2016. The dataset is publicly accessible via an HTTP API and monthly snapshots in CSV, JSON, and XML formats. Several online applications have been developed that query PanLex data. More broadly, the project aims to make a contribution to the preservation of global linguistic diversity.",
}
Glossary
To understand the terms like "language", "language variety", "expression" and "meaning" more precisely, please read the Panlex documentation on their data model and database design.
- Downloads last month
- 471