VietAlphaLabs/zh-vi-1955-dictionary

VietAlpha Việt Hán Tân Từ Điển

越漢新辭典, 1955

VietAlpha Lab · Website

The VietAlpha Việt Hán Tân Từ Điển is a machine-readable, verified digital edition of Việt Hán Tân Từ Điển (越漢新辭典), compiled by Phó Căn Thâm and published in 1955.

This Vietnamese–Chinese character dictionary maps Vietnamese headwords to their Hán (漢字) equivalents, with optional glosses and domain labels written in Classical Chinese. This release contains 21,204 clean entries (4,958 lemmas and 16,246 sub-entries).

---

Domain Labels

Entries may carry one or more domain labels (in Chinese), printed in parentheses. The most frequent labels are:

Label	Meaning	Count
地	Geography	217
醫	Medicine	133
植	Botany	99
俗	Colloquial / Folk	98
引	Extended / Figurative	98
法	Law	73
解	Anatomy	65
律	Legal	57
宗	Religion	43
理	Physics / Science	40
動	Zoology	39
數	Mathematics	39
佛	Buddhism	30
化	Chemistry	28
軍	Military	25

Labels appear in the parsed.labels array. Entries with no domain marking have an empty labels array.

---

Record Schema

{
  "record_id": "p0001-L-001",
  "text": "A 阿·啊·丫·鴉 — 闖進",
  "headword": "A",
  "han": "阿·啊·丫·鴉 — 闖進",
  "parsed": {
    "han_equivalents": ["阿", "啊", "丫", "鴉"],
    "gloss": "闖進",
    "labels": []
  },
  "is_lemma": true,
  "lemma_id": null
}

Field	Type	Description
record_id	str	Stable record identifier
text	str	Full entry text as printed
headword	str	Vietnamese headword
han	str	Chinese character equivalent(s) and optional gloss, as printed
parsed.han_equivalents	list[str]	Individual Hán character forms, split from han
parsed.gloss	str | null	Chinese-language gloss or definition, if present
parsed.labels	list[str]	Domain labels in Chinese (see table above)
is_lemma	bool | null	true for main headword entries; null for sub-entries
lemma_id	str | null	record_id of the parent lemma (for sub-entries)

---

Files

File	Description
dictionary.parquet	21,204 entries in Apache Parquet; backs the dictionary config and the Hub viewer
dictionary.jsonl	The same 21,204 entries as JSON Lines (UTF-8, one object per line)
stats.json	Release statistics: entry counts and label distributions
NOTICE	Attribution and historical source statement
LICENSE	Apache License 2.0

---

Loading the Data

Hugging Face datasets

from datasets import load_dataset

dataset = load_dataset(
    "VietAlphaLabs/zh-vi-1955-dictionary",
    "dictionary",
    split="train",
)

print(f"Loaded {len(dataset)} entries")
print(dataset[0])

Pandas

import pandas as pd

df = pd.read_parquet("dictionary.parquet")

# Find all entries labelled Geography (地)
geo = df[df["parsed"].apply(lambda p: "地" in p.get("labels", []))]
print(f"Found {len(geo)} geography entries")

# All lemma entries
lemmas = df[df["is_lemma"] == True]
print(f"Found {len(lemmas)} lemma entries")

---

Licensing & Attribution

• Transcriptions & dataset code: Licensed under the Apache License, Version 2.0 © 2026 VietAlpha Lab.
• Historical work: The underlying lexical work Việt Hán Tân Từ Điển (1955) by Phó Căn Thâm is in the public domain.
• When citing or redistributing this dataset, please retain the NOTICE file and cite as follows:

@dataset{vietalpha2026viethantan,
  author       = {VietAlpha Lab},
  title        = {VietAlpha Việt Hán Tân Từ Điển (1955)},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/VietAlphaLabs/zh-vi-1955-dictionary}},
  note         = {Digitized and reconstructed from Việt Hán Tân Từ Điển by Phó Căn Thâm (1955)}
}