ArabicNLPWorld/arabic-russian-parallel-corpus

Arabic‑Russian Parallel Corpus

A parallel corpus for Arabic–Russian language pairs. Each record contains an Arabic sentence/phrase, its Russian translation, and the source of the pair.
The dataset has been cleaned, deduplicated, and source names normalized to lowercase.

📊 Dataset Statistics

Overview

Metric	Value
Total entries	116,393
Unique Arabic strings	116,124
Unique Russian strings	116,152
Unique sources	6

Data completeness

All fields are fully populated – no missing values.

Field	Empty records
Arabic (arabic)	0
Russian (russian)	0
Source (source)	0

Token and character statistics

Language	Total tokens	Average tokens per entry	Total characters
Arabic	2,156,789	18.5	8,423,567
Russian	1,854,322	15.9	6,112,445

Calculated after tokenization (Arabic: Unicode Arabic block, Russian: Cyrillic letters).

Origin of the data

The corpus was built by merging multiple sources:

Source	Entries	Percentage
religion (islam)	41,609	35.7%
dictionary	37,999	32.6%
bible	30,907	26.6%
tatoeba	4,649	4.0%
news	826	0.7%
conversation	403	0.3%

Note: “dictionary” includes both word‑lists and phrase dictionaries; “religion” consists of religious texts (Quran, Hadith, prayers); “bible” is biblical excerpts.

Cleaning & deduplication

Step	Count	Percentage
Raw pairs read	125,246	100%
Kept after cleaning	116,393	92.93%
Removed (filters)	8,617	6.88%
Duplicates removed	236	0.19%

Filters applied: minimum length 5 characters, maximum length 150,000 characters, minimum Arabic/Russian character ratio 0.3.

📁 Data Format

Each record contains the following fields:

Field	Type	Description
arabic	string	Sentence/phrase in Arabic (Arabic script)
russian	string	Corresponding translation in Russian (Cyrillic script)
source	string	Source category (religion, dictionary, bible, tatoeba, news, conversation)
arabic_token_count	int32	Number of Arabic tokens in the Arabic string
russian_token_count	int32	Number of Russian tokens in the Russian string
arabic_char_count	int32	Total characters in the Arabic string
russian_char_count	int32	Total characters in the Russian string

🚀 Usage Example

from datasets import load_dataset

dataset = load_dataset("ArabicNLPWorld/arabic-russian-parallel-corpus")
data = dataset["train"]

# First record
print(data[0])

# Filter by source
bible_entries = data.filter(lambda x: x["source"] == "bible")
print(f"Entries from Bible: {len(bible_entries)}")

# Get all unique sources
sources = set(data["source"])
print(f"All sources: {sources}")

Sample record

{
    "arabic": "السلام عليكم",
    "russian": "Мир вам",
    "source": "religion",
    "arabic_token_count": 2,
    "russian_token_count": 2,
    "arabic_char_count": 12,
    "russian_char_count": 7
}

🔬 Potential Applications

• Arabic–Russian machine translation
• Cross‑lingual word embedding learning
• Lexicography and bilingual dictionary building
• Language learning tools for Arabic and Russian speakers
• Religious and literary text analysis

📜 License

Creative Commons Attribution-ShareAlike 4.0 (CC BY-SA 4.0)

🤝 Citation

@dataset{arabic_russian_parallel_corpus_2026,
    title = {Arabic-Russian Parallel Corpus},
    author = {ArabicNLPWorld},
    year = {2026},
    publisher = {Hugging Face},
    url = {https://huggingface.co/datasets/ArabicNLPWorld/arabic-russian-parallel-corpus}
}

📬 Contact

For questions or contributions, please open an issue on the Hugging Face repository or contact the ArabicNLPWorld team.

---

Automatically generated from cleaned and deduplicated JSONL source. Last update: June 2026.