# 1000_novels

1000 Novels Corpus (CLARIN-PL)

## Dataset description
- **Source (upstream):** https://clarin-pl.eu/dspace/handle/11321/312
- **Domain:** literature
- **Language:** Polish (pl)
- **License:** `CC-BY-4.0`
- **Created (range):** unknown
- **Added:** 2026-07-02

## Licensing — traceable basis
The 1000 Novels Corpus is released under Creative Commons Attribution 4.0 International by CLARIN-PL / Wrocław University of Technology, with attribution to the corpus authors and maintainers.

## Rights note (per-text screen, September 2026)
The CC-BY-4.0 licence covers the corpus as compiled by CLARIN-PL. It cannot license
third-party rights in the individual novels and translations. A per-text screen of
the released file found:

| status | texts |
|---|---:|
| public domain in Poland by term (author and any translator died by 1955) | 830 |
| marked public domain by Wolne Lektury | 10 |
| in copyright; published by Wolne Lektury under CC BY-SA 3.0 | 17 |
| author or translator could not be verified | 143 |

Releases up to v0.2.5 label all 1,000 texts `CC-BY-4.0`. Before the next release, the
17 in-copyright texts should be relabelled `CC-BY-SA-3.0` with Wolne Lektury
attribution or removed, and the 143 unverified texts withheld until their rights are
established.

## Provenance
Pulled from SpeakLeash's public redistribution (`speakleash-ds-pub`, key `1000_novels_corpus_CLARIN-PL`) of the upstream CLARIN-PL source. SpeakLeash credited as intermediate aggregator; upstream license/attribution preserved.

## Statistics
| documents | characters | tokens (tiktoken proxy) |
|---:|---:|---:|
| 1,000 | 151,610,758 | 60,517,363 |


## Per-document license metadata
| license | documents |
|---|---:|
| `CC-BY-4.0` | 1,000 |

Author metadata present for **0** documents. Empty values mean the upstream record did not expose a machine-readable author field.

Statistics were recomputed directly from the released parquet file.

## Filters applied (build_dynaword.py)
Minimal, per Dynaword guidelines (heavy filtering left to downstream use):
- drop documents < 200 chars: **0**
- drop non-Polish (diacritic ratio): **0**
- exact cross-source dedup (sha1): **0**
- OCR alpha-ratio < 0.70 (OCR sources only): **0**
- read 1,000 → kept 1,000

Token counts are a fast tiktoken (cl100k) proxy (~1% off Llama-3); the canonical
Llama-3 count is computed at release.