OpenChristianDataOrg/open-christian-data

---
license:
- cc0-1.0
- mit
language:
- en
task_categories:
- text-retrieval
- question-answering
- feature-extraction
size_categories:
- 100K<n<1M
configs:
- config_name: bible_text
data_files:
- split: train
path: data/bible_text.jsonl
- config_name: catechism_qa
data_files:
- split: train
path: data/catechism_qa.jsonl
- config_name: church_fathers
data_files:
- split: train
path: data/church_fathers.jsonl
- config_name: commentary
data_files:
- split: train
path: data/commentary.jsonl
- config_name: devotional
data_files:
- split: train
path: data/devotional.jsonl
- config_name: doctrinal_document
data_files:
- split: train
path: data/doctrinal_document.jsonl
- config_name: prayer
data_files:
- split: train
path: data/prayer.jsonl
- config_name: reference_entry
data_files:
- split: train
path: data/reference_entry.jsonl
- config_name: sermon
data_files:
- split: train
path: data/sermon.jsonl
- config_name: structured_text
data_files:
- split: train
path: data/structured_text.jsonl
- config_name: topical_reference
data_files:
- split: train
path: data/topical_reference.jsonl
---

# Open Christian Data

Public domain Christian literature as structured, machine-readable data — for developers and AI training.

Commentary data is trapped in HTML and PDFs. No structured commentary dataset exists on HuggingFace. No per-chapter commentary JSON exists on GitHub.

This dataset processes public domain Christian literature — commentaries, church fathers, confessions, catechisms, devotionals, prayers, sermons — into clean, schema-validated records with full provenance metadata.

**GitHub:** [OpenChristianData/open-christian-data](https://github.com/OpenChristianData/open-christian-data)

## Schema types

| Config | Records | Description |
|---|---|---|
| `commentary` | 109,774 | Verse-level commentary (Matthew Henry, Barnes, Calvin, Wesley, Adam Clarke, Gill, JFB, KD, Treasury of David) |
| `church_fathers` | 70,191 | Patristic quotes indexed by scripture reference — 325 authors (Augustine, Chrysostom, Jerome, Origen, Aquinas...) |
| `bible_text` | 31,086 | Berean Standard Bible — all 66 books, 31,086 verses, CC0 |
| `structured_text` | 13,207 | Paragraph-level blocks from 11 works (Calvin's Institutes, Augustine's Confessions, Pilgrim's Progress, City of God, Chesterton, Thomas à Kempis, MacDonald, Underhill, Milton, Luther's Large Catechism) |
| `topical_reference` | 5,945 | Nave's Topical Bible — 5,322 topics, 76,957 scripture references |
| `reference_entry` | 11,145 | Easton's, Smith's, Hitchcock's Bible Dictionaries + Torrey's New Topical Textbook |
| `catechism_qa` | 3,279 | Question-and-answer catechisms (Westminster Shorter/Larger, Heidelberg, Baltimore #1–3, Luther's Small, Keach's, and more) |
| `doctrinal_document` | 1,331 | Confessions and creeds at clause level (Westminster, Nicene, Chalcedonian, Belgic, Dort, London Baptist 1689, Savoy, and 30+ more) |
| `devotional` | 1,464 | Spurgeon's Morning and Evening (732 entries) + Daily Light on the Daily Path (732 entries) |
| `sermon` | 36 | George MacDonald's Unspoken Sermons — 3 series, 36 sermons, 171k words |
| `prayer` | 191 | BCP 1662 Collects (85), BCP 1928 Collects (102), Didache Prayers (4) |
| **Total** | **247,649** | |

## Data format

Every record is a flat JSON object. Six fields are inlined from the source file's metadata:

| Field | Description |
|---|---|
| `_source_id` | Unique identifier for the source work |
| `_source_title` | Title of the source work |
| `_author` | Author name |
| `_contributors` | Translators, editors, and digitizers (array, may be empty) |
| `_schema_type` | Schema type (matches the config name) |
| `_license` | License identifier (`cc0-1.0` or `public-domain`) |
| `_source_url` | Canonical URL for the source |

The remaining fields are schema-specific. All verse references use OSIS format (`Gen.1.1`, `Rom.9.1-Rom.9.5`).

### Commentary record example

```json
{
"_source_id": "matthew-henry-complete",
"_source_title": "Matthew Henry's Complete Commentary",
"_author": "Matthew Henry",
"_schema_type": "commentary",
"_license": "public-domain",
"_source_url": "https://bible.helloao.org",
"entry_id": "matthew-henry-complete.Ezek.1.1-3",
"book": "Ezekiel",
"book_osis": "Ezek",
"chapter": 1,
"verse_range": "1-3",
"verse_range_osis": "Ezek.1.1-Ezek.1.3",
"verse_text": "In the thirtieth year...",
"commentary_text": "The circumstances of the vision which Ezekiel saw...",
"word_count": 2042
}
```

### Structured text record example

```json
{
"_source_id": "pilgrims-progress",
"_source_title": "The Pilgrim's Progress",
"_author": "John Bunyan",
"_schema_type": "structured_text",
"_license": "public-domain",
"_source_url": "https://github.com/standardebooks/john-bunyan_the-pilgrims-progress",
"work_id": "pilgrims-progress",
"work_kind": "allegory",
"section_type": "part",
"section_label": "The First Part",
"section_title": "",
"section_path": ["The First Part"],
"block_index": 0,
"text": "As I walked through the wilderness of this world..."
}
```

## Usage

```python
from datasets import load_dataset

# Load a specific schema type
commentary = load_dataset("OpenChristianDataOrg/open-christian-data", "commentary")
church_fathers = load_dataset("OpenChristianDataOrg/open-christian-data", "church_fathers")
catechisms = load_dataset("OpenChristianDataOrg/open-christian-data", "catechism_qa")

# Filter by source
matthew_henry = [r for r in commentary["train"] if r["_source_id"] == "matthew-henry-complete"]

# All commentary on a specific verse
rom_8_28 = [
r for r in commentary["train"]
if "Rom.8.28" in (r.get("verse_range_osis") or "")
]
```

## Sources

- **Bible text**: [Berean Standard Bible](https://berean.bible) — CC0 since April 2023
- **Commentary**: [HelloAO Bible API](https://bible.helloao.org) — Matthew Henry, JFB, Gill, Adam Clarke, KD; [CrossWire SWORD](https://www.crosswire.org/sword/) — Barnes, Calvin, Wesley, Treasury of David; all public domain
- **Church Fathers**: [HistoricalChristianFaith/Commentaries-Database](https://github.com/HistoricalChristianFaith/Commentaries-Database) — 325 authors, public domain
- **Devotionals**: Christian Classics Ethereal Library (ccel.org) — Spurgeon's Morning and Evening (ThML XML); CrossWire SWORD — Daily Light on the Daily Path; public domain
- **Structured texts**: [Standard Ebooks](https://standardebooks.org) — 9 titles in CC0-annotated XHTML; Project Gutenberg — Luther's Large Catechism, Calvin's Institutes, Augustine's Confessions; public domain
- **Catechisms**: Project Gutenberg — Luther's Small Catechism, Baltimore Catechisms #1–3; additional catechisms (Westminster, Heidelberg, Keach's, and more); public domain
- **Confessions & creeds**: Westminster Confession, Nicene Creed, Chalcedonian Definition, and 30+ more historic documents; public domain
- **Prayers**: [eskimo.com BCP 1662](https://eskimo.com/~lhowell/bcp1662/) (Lynda M. Howell); [episcopalnet.org BCP 1928](https://www.episcopalnet.org/1928bcp/); [Wikisource Didache](https://en.wikisource.org/wiki/Didache_(Lake_translation)) (Kirsopp Lake 1912 translation); public domain
- **Bible dictionaries**: Easton's (1893), Smith's (1863), Hitchcock's (1874), Torrey's (1897); public domain
- **Topical reference**: [CrossWire SWORD](https://www.crosswire.org/sword/) — Nave's Topical Bible (1896); public domain
- All authors died before 1928; texts are unambiguously public domain.

## Attribution

Devotional text sourced from the Christian Classics Ethereal Library (ccel.org). ThML parsing permitted per correspondence with CCEL (April 2026).

## License

- **Data** (JSONL datasets): [CC0 1.0 Universal](https://creativecommons.org/publicdomain/zero/1.0/) — dedicated to the public domain
- **Code** (build scripts, schemas, tooling): [MIT](https://opensource.org/licenses/MIT)

The underlying texts are public domain. Our value-add is the structuring and provenance tracking, which we also dedicate to the public domain.