Publish data/CC-MAIN-2026-08/00000.parquet

LICENSE

@@ -1,15 +1,21 @@
-Common Crawl License Notice
+Open Data Commons Attribution License (ODC-By) v1.0
 
-This repository contains data derived from Common Crawl.
+This dataset is made available under the Open Data Commons Attribution License:
+https://opendatacommons.org/licenses/by/1-0/
 
-Common Crawl makes its datasets publicly available subject to its Terms of Use:
-https://commoncrawl.org/terms-of-use
+You are free to share, create, and adapt this data — even for commercial purposes —
+as long as you attribute the source.
 
-Important:
+Attribution requirements:
+- Cite "Open Index, derived from Common Crawl (https://commoncrawl.org)"
+- Include a link to this dataset when used in publications or products
 
-1. Common Crawl is an archive of third-party web content.
-2. The original content remains subject to the rights and terms of its respective publishers.
-3. You are responsible for complying with applicable law, downstream licensing obligations,
-   robots restrictions, privacy requirements, and content removal requests.
+Additional notices:
 
-Refer to the Common Crawl Terms of Use for the governing terms for the crawl data itself.
+1. This dataset contains data derived from Common Crawl, which archives third-party
+   web content. The original content remains subject to the rights of its respective
+   publishers and the Common Crawl Terms of Use: https://commoncrawl.org/terms-of-use
+
+2. You are responsible for complying with applicable law including downstream licensing
+   obligations, robots.txt restrictions, privacy requirements, and content removal
+   requests from original publishers.

README.md

@@ -1,26 +1,125 @@
 ---
-license: other
-pretty_name: Open Index Draft
+license: odc-by
+pretty_name: Open Index
+language:
+- en
+tags:
+- common-crawl
+- web-crawl
+- markdown
+- text
+size_categories:
+- 10B<n<100B
 ---
 
-# Open Index Draft
+# Open Index
 
-This dataset contains markdown exports derived from Common Crawl shard CC-MAIN-2026-08.
+**Open Index** is a large-scale web text dataset derived from [Common Crawl](https://commoncrawl.org) with HTML converted to clean Markdown. Designed for language model training, information retrieval research, and web-scale NLP.
 
-Layout:
+This snapshot is built from crawl **CC-MAIN-2026-08**.
 
-- data/*.parquet: one parquet file per packed markdown WARC shard
-- README.md: dataset description
-- LICENSE: Common Crawl licensing and usage notice
+---
+
+## Dataset Summary
+
+| Property | Value |
+|---|---|
+| Source | Common Crawl (CC-MAIN-2026-08) |
+| Format | Apache Parquet (Zstd compressed) |
+| Content | Markdown-converted web pages |
+| License | [ODC-By 1.0](https://opendatacommons.org/licenses/by/1-0/) |
+
+---
+
+## Dataset Structure
+
+Parquet files are organised by crawl ID:
+
+`
+data/
+└── CC-MAIN-2026-08/
+    ├── 00000.parquet
+    ├── 00001.parquet
+    └── ...
+`
+
+Each file corresponds to one packed WARC shard (~1 GB source WARC).
+
+### Data Fields
+
+| Field | Type | Description |
+|---|---|---|
+| `doc_id` | string | UUID derived from the WARC-Record-ID |
+| `url` | string | Original URL of the crawled page |
+| `host` | string | Lowercase hostname extracted from the URL |
+| `crawl_date` | string | RFC3339 timestamp from the WARC record |
+| `warc_type` | string | WARC record type (conversion, response, …) |
+| `warc_record_id` | string | Original `<urn:uuid:…>` WARC record identifier |
+| `warc_refers_to` | string | Record ID of the source response record |
+| `content_type` | string | HTTP Content-Type of the original response |
+| `html_length` | int64 | Byte length of the original HTML body |
+| `markdown_length` | int64 | Byte length of the converted Markdown body |
+| `warc_headers_json` | string | All WARC headers as stable-key JSON |
+| `markdown_body` | string | Clean Markdown text converted from HTML |
+| `source_warc_file` | string | Source packed .md.warc.gz shard filename |
+| `source_file_index` | int32 | Index of the source file in the crawl manifest |
+
+---
+
+## Usage
 
-Parquet schema:
+### Hugging Face datasets
 
-- doc_id, url, host, crawl_date
-- warc_type, warc_record_id, warc_refers_to
-- content_type, content_length, markdown_length
-- warc_headers_json: all WARC header metadata serialized as JSON
-- markdown_body: markdown body extracted from the packed WARC record
+`python
+from datasets import load_dataset
+
+# Stream the full snapshot
+ds = load_dataset("open-index/draft", split="train", streaming=True)
+for doc in ds:
+    print(doc["url"], doc["markdown_body"][:200])
+
+# Load a single shard
+ds = load_dataset(
+    "open-index/draft",
+    data_files="data/CC-MAIN-2026-08/00000.parquet",
+    split="train",
+)
+`
+
+### DuckDB
+
+`sql
+SELECT url, host, markdown_length
+FROM read_parquet('hf://datasets/open-index/draft/data/CC-MAIN-2026-08/*.parquet')
+WHERE host LIKE '%wikipedia.org'
+LIMIT 10;
+`
+
+### pandas
+
+`python
+import pandas as pd
+
+df = pd.read_parquet(
+    "hf://datasets/open-index/draft/data/CC-MAIN-2026-08/00000.parquet",
+    columns=["url", "host", "crawl_date", "markdown_body"],
+)
+`
+
+---
+
+## Data Processing Pipeline
+
+1. **Download** — Raw .warc.gz files from Common Crawl S3.
+2. **Filter** — HTTP 200 responses with text/html content only.
+3. **Convert** — HTML → Markdown via [trafilatura](https://github.com/adbar/trafilatura) (removes boilerplate, navigation, ads).
+4. **Pack** — Seekable .md.warc.gz files (one gzip member per record, CC-compatible format).
+5. **Export** — Parquet with Zstd compression, 100K rows per row group.
+
+---
 
-Source:
+## Source & License
 
 - Common Crawl: [https://commoncrawl.org](https://commoncrawl.org)
+- Terms of Use: [https://commoncrawl.org/terms-of-use](https://commoncrawl.org/terms-of-use)
+- This dataset is released under the [Open Data Commons Attribution License (ODC-By) v1.0](https://opendatacommons.org/licenses/by/1-0/)

data/CC-MAIN-2026-08/00000.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4b1f55ef10025539e07eb92ae726a39073a80ac09c7d25d7ae9c6ec8abfaaa8
+size 33308582