Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
7
7
language
stringclasses
4 values
topic
stringclasses
13 values
era
stringclasses
7 values
pd_basis
stringclasses
4 values
author_death_bucket
stringclasses
8 values
first_published_bucket
stringclasses
5 values
edition_type
stringclasses
5 values
ocr
bool
2 classes
characters
int64
2.15k
2.88M
text
stringlengths
2.15k
2.88M
sc-0002
zh
bat-tu-menh-ly
pre-modern, undated
premodern-dated
1800-1900
1600-1800
manuscript-copy
false
21,763
珞琭子三命消息賦註   欽定四庫全書     子部七   提要   珞琭子三命消息賦注 術數類五〈命書相書之屬〉臣等謹案珞琭子三命消息賦注二卷宋徐子平撰珞琭子書為言禄命者所自出其法専以人生年月日時八字推衍吉凶禍福李淑邯鄲書目謂其取琭琭如玉珞珞如石之意而不知撰者為何人朱弁曲洧舊聞云世傳珞琭子三命賦不知何人所作序而釋之者以為周世子晉所為然考其賦所引有秦河上公又如懸壺化杖之事皆後漢末壺公費長房之徒則非周子晉明矣是書前有楚頤序又謂珞琭子者陶宏景所自稱然禄命之說至唐李虚中尚僅以年月日起算未有所謂八字者宏景之時又安有是説乎考其書始見於宋藝文志而晁公武讀書志亦云宣和建炎之間是書始行則當為北宋人所作舊稱某某皆依托也自宋以來注此賦者王廷光李同釋...
sc-0003
zh
bat-tu-menh-ly
pre-modern, undated
ancient-anonymous
anonymous
1600-1800
manuscript-copy
false
23,477
李虛中命書   欽定四庫全書     子部七   提要   李虛中命書三巻   術數類五〈命書相書之屬臣〉等謹案李虛中命書三卷舊本題鬼谷子撰唐李虛中註虛中字常容魏侍中李冲八世孫進士及第元和中官至殿中侍御史韓愈為作墓誌銘見於昌黎文集後世傳星命之學者皆以虚中為祖愈墓誌中所云最深五行書以人之始生年月日所值日辰支干相生勝衰死王相斟酌推人夀夭貴賤利不利輙先處其年時百不失一二者是也然愈但極稱其説之汪洋奧羙萬端千緒而不言有所著書唐書藝文志亦無是書之名至宋志始有李虛中命書格局二卷鄭樵藝文畧則作李虚中命術一卷命書補遺一卷晁公武讀書志又作李虚中命書三卷焦氏經籍志又於命書三卷外别出命書補遺一卷名目卷數皆參錯不合世間傳本久絶無以考正其異同惟永樂大典所...
sc-0004
zh
bat-tu-menh-ly
1368-1643
premodern-dated
1800-1900
1600-1800
manuscript-copy
false
446,825
"三命通會\n\n  欽定四庫全書     子部七\n\n  三命通會    (...TRUNCATED)
sc-0006
zh
bat-tu-menh-ly
1000-1367
premodern-dated
1800-1900
before 1600
manuscript-copy
false
30,396
"珞琭子賦註\n\n  欽定四庫全書     子部七\n\n  提要\n\n  珞琭(...TRUNCATED)
sc-0007
zh
bat-tu-menh-ly
pre-modern, undated
premodern-dated
pre-modern, undated
pre-modern, undated
retyped-premodern
false
16,794
"←\n\n累害篇第二\n\n論衡\n命祿篇第三\n作者:王充\n\n氣壽篇第四\n\n  凡(...TRUNCATED)
sc-0008
zh
bat-tu-menh-ly
pre-modern, undated
premodern-dated
1800-1900
1600-1800
manuscript-copy
false
15,949
"玉照定真經\n\n  欽定四庫全書     子部七\n\n  提要\n\n  玉(...TRUNCATED)
sc-0009
zh
bat-tu-menh-ly
1000-1367
premodern-dated
before 1600
before 1600
retyped-premodern
false
193,954
"## 《五行精紀》\n\n宋-廖中\n\n序一\n\n天錫丸疇,一曰五行。凡天、地、人(...TRUNCATED)
sc-0010
zh
bat-tu-menh-ly
pre-modern, undated
ancient-anonymous
anonymous
1600-1800
manuscript-copy
false
17,665
"三命指迷賦\n\n  欽定四庫全書     子部七\n\n  提要\n\n  三命(...TRUNCATED)
sc-0011
zh
bat-tu-menh-ly
1644-1839
premodern-dated
1800-1900
1600-1800
manuscript-copy
false
41,083
"目錄\n\n四庫全書總目提要\n\n◀上一卷\n\n卷一百九•子部十九\n\n下一卷▶\n(...TRUNCATED)
sc-0012
zh
bat-tu-menh-ly
pre-modern, undated
premodern-dated
pre-modern, undated
pre-modern, undated
retyped-premodern
false
3,864
"殿中侍御史李君墓誌銘\n作者:韓愈 唐\n\n\n殿中侍御史李君,名虛中,字(...TRUNCATED)
End of preview. Expand in Data Studio

Public domain classical texts

373 full texts, 72.7 million characters, in the public domain, on the traditions Nhat Nguyet works with: the I Ching and its commentaries, Zi Wei Dou Shu (Tử Vi), BaZi, the Three Arts, feng shui, physiognomy, the calendar and date selection, numerology, Buddhism, Vietnamese folk belief, rites and history. Classical Chinese, English, French, Vietnamese.

Data: Nhat Nguyet (nhatnguyet.org). Project page: https://nhatnguyet.org/en/ma-nguon-mo

What is in it

Each text is one file, named by a neutral id (sc-0001 to sc-0433, with gaps). Ids are stable across releases: a text keeps its id, and the id of a text withdrawn from the set is never reused. Titles, authors and source sites are not part of this release. Instead, every text carries the facts its public domain status rests on, in a form that does not identify the book: the legal basis (pd_basis), the period in which the last of its creators died (author_death_bucket), the period by which it existed as a book (first_published_bucket) and the kind of edition (edition_type). A long work can be split across several files (one file per group of volumes), and a few works appear in more than one edition, so the number of distinct works is a little lower than the number of files.

Language Texts Characters
Classical Chinese (Hán văn) (zh) 306 37,080,318
English (en) 36 15,856,892
French (fr) 20 15,928,152
Vietnamese (vi) 11 3,785,607

The French texts were printed during the colonial period. They are kept as historical sources and may carry the views of their time.

Subject group Texts Characters
I Ching and its commentaries, divination (kinh-dich-boc-phe) 192 27,960,230
Folk belief, family rites, festivals (tin-nguong-phong-tuc) 24 7,068,723
Vietnamese history and culture (lich-su-van-hoa-viet-nam) 22 6,984,246
Numerology and number philosophy (than-so-hoc) 21 7,247,953
Three Arts (Qi Men Dun Jia, Da Liu Ren, Tai Yi) (tam-thuc) 20 832,957
Calendar, stems and branches, date selection (lich-phap-chon-ngay) 19 6,452,812
BaZi and Chinese fate calculation (bat-tu-menh-ly) 17 1,147,893
Feng shui (phong-thuy) 16 1,833,541
Buddhism and Zen (phat-giao) 14 4,088,623
Physiognomy and palmistry (nhan-tuong) 13 824,552
Astrology and Tarot (chiem-tinh-tarot) 7 5,562,200
Zi Wei Dou Shu (Tử Vi) (tu-vi) 6 810,740
Other (khac) 2 1,836,499
Period of composition or first printing (era) Texts Characters
before 1000 4 400,100
1000-1367 10 1,402,388
1368-1643 13 1,269,999
1644-1839 25 8,232,218
1840-1911 54 23,698,189
1912-1930 21 10,924,394
pre-modern, undated 246 26,723,681

38 texts (26.8 million characters) come from optical character recognition of old printed books (ocr: true). They keep OCR errors, old typography (long s, broken hyphenation) and some library stamps; use them with care.

Why each text is in the public domain

Every text was checked against an internal record (title, every author, commentator, editor and translator, their death years, the year of the edition used) under one rule set that holds for both Vietnamese law (50 years after the author's death) and United States law, where this hub is hosted (works published before 1931). When a fact needed for the rule was missing from the record, the text was left out rather than kept on a guess.

pd_basis Texts Characters
author-died-before-1955 88 35,776,639
premodern-dated 223 30,336,929
ancient-anonymous 47 1,508,027
government-work 15 5,029,374
  • author-died-before-1955: every named author, commentator, editor and translator has a recorded death year, the latest in or before 1955, and the edition used (for a translation, the translation) took book form before 1931.
  • premodern-dated: named pre-modern authors dated only by dynasty or by a dated preface, colophon or copy (latest dated activity no later than 1875, so every creator died long before 1955); the text circulated before 1931.
  • ancient-anonymous: anonymous text, or one only traditionally attributed, recorded as an ancient or pre-modern text; no identifiable modern creator.
  • government-work: official work of a former state (court histories, law code) compiled by a government office and printed before 1931; no individual author.
pd_basis by language zh en fr vi
author-died-before-1955 23 35 19 11
premodern-dated 221 1 1 0
ancient-anonymous 47 0 0 0
government-work 15 0 0 0
Latest creator death (author_death_bucket) Texts Characters
before 1600 22 2,476,674
1600-1800 23 5,151,186
1800-1900 189 32,700,021
1900-1930 38 16,146,670
1930-1955 22 8,983,417
pre-modern, undated 17 655,600
anonymous 47 1,508,027
institutional 15 5,029,374
In book form by (first_published_bucket) Texts Characters
before 1600 88 10,182,161
1600-1800 156 24,879,219
1800-1900 58 20,780,482
1900-1930 34 15,702,368
pre-modern, undated 37 1,106,739

author_death_bucket is the period of the latest death among all named creators. Where the record only gives a dynasty or a dated preface, it holds the latest year that bound allows (dynasty end, or the dated activity plus 80 years). anonymous means no named creator; institutional means a government office. first_published_bucket is the period by which the record shows the text existed as a book: the year of the edition used, the imperial library copy (1773 to 1782), a dated preface or colophon, or, for a pre-modern author, the author's lifetime. No text was first printed in 1931 or later; no creator died after 1955.

edition_type Texts Characters
manuscript-copy 186 25,514,587
woodblock 3 205,988
retyped-premodern 117 11,359,743
typeset 50 25,271,703
translation 17 10,298,948
  • manuscript-copy: hand-copied edition (imperial library copy or manuscript compilation), retyped.
  • woodblock: woodblock edition, retyped.
  • retyped-premodern: retyped digital text of a pre-modern edition (edition not stated).
  • typeset: book printed in movable type before 1931.
  • translation: translation or transliteration printed before 1931 (translator also out of copyright).

Left out by the same review: modern editions and translations (printed from 1931 on, or whose translator or editor died after 1955, or with no printing year), works whose creators' death years could not be established, texts released only under a CC0 statement by a digitiser, digital copies whose licence does not allow republication, critical editions and transcriptions made by modern scholars, and digital copies that mix in notes, prefaces or explanations written by present-day editors.

Files

Path Contents Size
data/train-00000-of-00001.parquet one row per text: id, language, topic, era, pd_basis, author_death_bucket, first_published_bucket, edition_type, ocr, characters, text 70 MB
text/<language>/<id>.txt the same texts as plain UTF-8 files 147 MB
metadata.jsonl one line per text: the columns above without text, plus file, bytes, sha256
from datasets import load_dataset

ds = load_dataset("nhatnguyet/sach-co", split="train")
print(ds)
row = ds[0]
print(row["id"], row["language"], row["topic"], row["era"], row["pd_basis"], row["characters"])
print(row["text"][:200])

# only the Vietnamese texts on folk belief and rites
vi = ds.filter(lambda r: r["language"] == "vi" and r["topic"] == "tin-nguong-phong-tuc")
print(len(vi))

# only texts whose every creator has a recorded death year
dated = ds.filter(lambda r: r["pd_basis"] == "author-died-before-1955")
print(len(dated))

Cleaning

Cleaning removed download headers, file markers, web addresses, source-site templates and notes, and scanner notices. Nothing else in the text was edited: no punctuation was added, no characters were converted between traditional and simplified forms.

License and citation

The texts are in the public domain. The selection, cleaning and metadata are released under CC BY 4.0. Credit line: Data: Nhat Nguyet (nhatnguyet.org).

These books describe divination, fate calculation and other traditional practices. Their rules are the rules of the schools that wrote them; none of them has been empirically validated, and nothing here is a prediction.

Tiếng Việt

Bộ gồm 373 văn bản toàn văn, 72,7 triệu ký tự, thuộc phạm vi công cộng, về các môn Nhật Nguyệt làm: Kinh Dịch và các bản chú, Tử Vi, Bát Tự, Tam thức, phong thủy, nhân tướng, lịch pháp và chọn ngày, thần số học, Phật giáo, tín ngưỡng dân gian, lễ tục và sử Việt. Mỗi văn bản một tệp, mang mã trung tính, không ghi tên sách, tác giả, nơi lấy. Thay vào đó mỗi văn bản mang căn cứ phạm vi công cộng không định danh: mã căn cứ (pd_basis), khoảng năm mất muộn nhất của người làm ra văn bản, khoảng năm văn bản đã thành sách, loại bản. Mọi văn bản được rà theo một luật đúng cho cả Việt Nam (50 năm sau khi tác giả mất) lẫn Mỹ (in trước 1931): không người làm nào mất sau 1955, không bản nào in từ 1931; thiếu dữ kiện thì bỏ, không giữ theo phỏng đoán. Phần chọn lọc, làm sạch và siêu dữ liệu theo giấy phép CC BY 4.0; ghi nguồn "Dữ liệu: Nhật Nguyệt (nhatnguyet.org)". Quy tắc trong sách là quy tắc của phái đã viết ra nó, chưa được kiểm chứng. Các văn bản tiếng Pháp in thời thuộc địa được giữ như tư liệu lịch sử, có thể mang cách nhìn của thời ấy. Trang dự án: https://nhatnguyet.org/en/ma-nguon-mo.

Downloads last month
87