The JRC-Acquis: A Multilingual Aligned Parallel Corpus with 20+ Languages

By Ralf Steinberger, Bruno Pouliquen, Anna Widiger, Camelia Ignat, Tomaz Erjavec, Dan Tufis, D. Varga. International Conference on Language Resources and Evaluation, 2006

707 citations recorded by the source corpus.

We present a new, unique and freely available parallel corpus containing European Union (EU) documents of mostly legal nature. It is available in all 20 official EU languages, with additional documents being available in the languages of the EU candidate countries. The corpus consists of almost 8,000 documents per language, with an average size of nearly 9 million words per language. Pair-wise paragraph alignment information produced by two different aligners (Vanilla and HunAlign) is available for all 190+ language pair combinations. Most texts have been manually classified according to the EUROVOC subject domains so that the collection can also be used to train and test multi-label classification algorithms and keyword-assignment software. The corpus is encoded in XML, according to the Text Encoding Initiative Guidelines. Due to the large number of parallel texts in many languages, the JRC-Acquis is particularly suitable to carry out all types of cross-language research, as well as to test and benchmark text analysis software across different languages (for instance for alignment, sentence splitting and term extraction).

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/1d9a8f3556c31eee9e0f77de6d04bd18978f8473

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Improving Translation Model by Monolingual Data (2011, methodology): “JRC-Acquis We used the JRC-Acquis 3.0 corpus (Steinberger et al., 2006) mainly because of the number of available languages.”
Search Engine Guided Non-Parametric Neural Machine Translation (2017, methodology): “We evaluate the proposed search engine guided non-parametric neural machine translation on three language pairs (En-Fr, En-De, and En-Es, in both directions) from JRC-Acquis Corpus (Steinberger et al., 2006) which consists of documents from a legal domain.”
Findings of the 2021 Conference on Machine Translation (WMT21) (2021, background): “Released corpora for the Spanish–Portuguese language pair included parallel datasets: Europarl v10 (Koehn, 2005), News Commentary v16, Wiki Titles v3, Tilde MODEL (Rozis and Skadin, š, 2017), JRC-Acquis (Stein- 20http://opus.nlpl.eu/ berger et al., 2006), and monolingual corpora: Europarl v10 (Koehn”
DGT-TM: A freely available Translation Memory in 22 languages (2012, methodology): “The number of highly multilingual parallel collections is relatively low, with EuroParl (Koehn 2005; 21 languages in the 2011 update), DGT-TM (2007) and JRC-Acquis (Steinberger et al., 2006) (22 languages each) being among the most highly multilingual ones.”
South-East European Times : A parallel corpus of Balkan languages , Francis Tyers and (2010, methodology): “We present some of the most frequent errors found in the JRC-Acquis corpus.”
The United Nations Parallel Corpus v1.0 (2016, methodology): “For individual documents, it was decided to follow the TEI-based format of the JRC-Aquis parallel corpus (Steinberger et al., 2006).”
A Parallel Corpus for Evaluating Machine Translation between Arabic and European Languages (2017, methodology): “By definition, translations of this document collection are therefore available in all ofﬁcial EU languages (Steinberger et al., 2006).”
Augmenting Neural Machine Translation with Knowledge Graphs (2019, methodology): “0 (Steinberger et al., 2006), Europarl (Koehn, 2005) and OpenSubtitles2013 (Tiedemann, 2012), obtaining a parallel training corpus of two million sentences, containing around 38M running words.”
Character-Based Pivot Translation for Under-Resourced Languages and Domains (2012, background): “On the other hand, there (Steinberger et al., 2006) is in-domain data for other languages like Danish that may act as an intermediate pivot.”
A massively parallel corpus: the Bible in 100 languages (2014, background): “…Centre of the European Commission has released multiple corpora in more than 20 languages, including the sentence-aligned JRC-Acquis (22 languages, Steinberger et al. 2006) and the paragraph-aligned DGT-Acquis (23 languages); the InterCorp corpus (Cˇerma´k and Rosen 2012), a collection of texts…”
The repetition rate of text as a predictor of the effectiveness of machine translation adaptation (2014, methodology): “For what concerns the legal domain, the publicly available JRC-Acquis collection (Steinberger et al., 2006) was used, which mostly includes EU legislative texts translated into 22 languages.”
SYSTRAN's Pure Neural Machine Translation Systems (2016, background): “Also available open-source corpora are domain speciﬁc: Europarl (Koehn, 2005), JRC (Steinberger et al., 2006) or MultiUN (Chen and Eisele, 2012) are legal texts, ted talk are scientiﬁc presentations, open subtitles (Tiedemann, 2012) are colloquial, etc.”
Sparse canonical correlation analysis (2009, background): “– Two paired English-French and English-Spanish corpora from the jrc-acquis dataset (Ralf et al. 2006).”
English-Portuguese Biomedical Translation Task Using a Genuine Phrase-Based Statistical Machine Translation Approach (2016, methodology): “The translations were extracted automatically from several corpora, including Europarl (Koehn and Monz, 2005), JRC-Acquis (Steinberger et al., 2006), OPUS EMEA (Tiedemann, 2009) and others, using a combination of complementary alignment and extraction methods: GIZA (Och and Ney, 2003), Anymalign (La”
Search Engine Guided Neural Machine Translation (2018, methodology): “Data We use the JRC-Acquis corpus(Steinberger et al., 2006) for evaluating the proposed SEG-NMT model.”