ParaPat: The Multi-Million Sentences Parallel Corpus of Patents Abstracts

This repository contains the pointers to the ParaPat paper, a corpora of parallel sentences from patents extracted from Google Patents. The paper has been published in LREC2020.

Paper

The ParaPat paper can be found here: https://www.aclweb.org/anthology/2020.lrec-1.465/

For citation, please use the following BibTeX

@inproceedings{soares-etal-2020-parapat,
    title = "{P}ara{P}at: The Multi-Million Sentences Parallel Corpus of Patents Abstracts",
    author = "Soares, Felipe  and
      Stevenson, Mark  and
      Bartolome, Diego  and
      Zaretskaya, Anna",
    booktitle = "Proceedings of The 12th Language Resources and Evaluation Conference",
    month = may,
    year = "2020",
    address = "Marseille, France",
    publisher = "European Language Resources Association",
    url = "https://www.aclweb.org/anthology/2020.lrec-1.465",
    pages = "3769--3774",
    language = "English",
    ISBN = "979-10-95546-34-4",
}

Digital Object Identifier (DOI) and access to files

https://doi.org/10.6084/m9.figshare.12627632

Corpus Statistics:

Parallel corpora aligned into sentence level

Language Pair	# Sentences	# Unique Tokens
EN/ZH	4.9M	155.8M
EN/JA	6.1M	189.6M
EN/FR	12.2M	455M
EN/KO	2.3M	91.4M
EN/DE	2.2M	81.7M
EN/RU	4.3M	107.3M
DE/FR	1.2M	38.8M
FR/JA	0.3M	9.9M
EN/ES	0.6M	24.6M

Parallel corpora aligned into abstract level

Language Pair	# Abstracts
FR/KO	120,607
EN/UK	89,227
RU/UK	85,963
CS/EN	78,978
EN/RO	48,789
EN/HU	42,629
ES/FR	32,553
EN/SK	23,410
EN/PT	23,122
BG/EN	16,177
FR/RU	10,889

MT Experiments Results

Source Language	Target Language	BLEU Score
ES	EN	57.71
FR	EN	49.60
EN	RU	49.10
EN	FR	48.39
DE	EN	48.35
EN	DE	46.77
RU	EN	46.73
EN	ES	44.98
DE	FR	42.13
FR	DE	36.56
KO	EN	23.23
JA	EN	17.78
EN	JA	13.15
ZH	EN	13.01
EN	KO	11.93
EN	ZH	10.29
JA	FR	9.52
FR	JA	7.10