VietAlphaLabs/vi-en-mathematics-dictionary

VietAlpha English–Vietnamese Mathematics Dictionary

Research page · VietAlpha Lab · Source scan

The VietAlpha English–Vietnamese Mathematics Dictionary turns a 709-page printed reference work into a machine-readable bilingual lexicon. It contains 26,205 English and Vietnamese mathematics entries digitized from Cung Kim Tiến's Từ Điển Toán Học Anh – Việt, Việt – Anh and organized as JSON Lines.

What is in the dataset

Direction	Entries	Source pages
English to Vietnamese	14,888	1–408
Vietnamese to English	11,317	410–709
Total	26,205	709 pages

The entries cover analysis, algebra, geometry, topology, probability, statistics, logic, numerical methods, and adjacent terminology from computing, physics, and astronomy.

Each record contains the original headword and translation, a stable identifier, the lookup direction, the source page, and convenience fields for the English and Vietnamese sides.

{
  "id": "envi-00011",
  "direction": "en-vi",
  "entry_no": 11,
  "headword": "Abelian group",
  "translation": "nhóm Aben, nhóm giao hoán",
  "senses": ["nhóm Aben, nhóm giao hoán"],
  "en": "Abelian group",
  "vi": "nhóm Aben, nhóm giao hoán",
  "page": 1
}

Use cases

• English–Vietnamese terminology lookup for mathematics
• Glossary injection and lexical constraints for translation systems
• Retrieval and RAG over bilingual mathematical terms
• Vocabulary coverage checks for OCR, language models, and tokenizers
• Research on the development of Vietnamese mathematical terminology
• Search and indexing across digitized textbooks, archives, and teaching material

The dataset provides lexical alignment, not sentence-level translation or mathematical reasoning examples. Its value is precision at the term level: a shared vocabulary that other systems can search, compare, and reuse.

Methodology

The source is the scanned copy of Cung Kim Tiến's dictionary available through the Internet Archive.

Data files

File	Contents
data/en_vi.jsonl	14,888 English-to-Vietnamese records
data/vi_en.jsonl	11,317 Vietnamese-to-English records
data/dictionary.json	All 26,205 records in one JSON array
data/stats.json	Counts and build provenance
dictionary/structured.md	Segmented OCR source table
dictionary/raw_lines.md	Unsegmented OCR output
build_dataset.py	Deterministic dataset builder

Load with Hugging Face Datasets

from datasets import load_dataset

# Both directions as separate splits
dataset = load_dataset("VietAlphaLabs/vi-en-mathematics-dictionary")

# One direction
en_vi = load_dataset(
    "VietAlphaLabs/vi-en-mathematics-dictionary",
    "en-vi",
    split="train",
)

License and attribution

Two license layers govern the release:

• VietAlpha's OCR transcription, segmentation, schema, build tooling, and documentation are released under the Apache License 2.0.
• Cung Kim Tiến's underlying dictionary content is listed by the Internet Archive under CC BY-NC-ND 4.0.

Use involving the dictionary entries must follow the underlying license: attribution to Cung Kim Tiến, non-commercial use, and no distribution of modified dictionary content. The repository's NOTICE file records the full attribution and layer boundaries.

Citation

@misc{vietalphalab2026mathdictionary,
  title     = {Từ Điển Toán Học Anh–Việt, Việt–Anh: An English–Vietnamese Mathematics Dictionary Dataset},
  author    = {VietAlpha Lab},
  note      = {Digitized from the printed dictionary by Cung Kim Tiến, CC BY-NC-ND 4.0},
  year      = {2026},
  publisher = {Hugging Face},
  url       = {https://huggingface.co/datasets/VietAlphaLabs/vi-en-mathematics-dictionary}
}