Corpus Usage Terms
Effective 5 October 2026 · Last updated 5 October 2026
In short
- Academic research and publication: allowed for non-commercial use, with attribution.
- Commercial use: not allowed.
- Training third-party AI models: not allowed without the speakers' written permission.
- Some of this data is not ours to license. Parts come from copyrighted third-party works — see section 3.
This project documents regional languages of North Sulawesi. Most of the text is a parallel corpus: the same content in Indonesian and in the regional language. Because the sources differ in origin, the rules differ by layer.
1. Who runs this and what this covers
The project owner is Peter Kambey. These terms cover the language data (dictionary, corpus, recordings, grammar rules) and the derived work (alignments, embeddings, processed data) served from kms.idlce.ai.id and idlce.ai.id. For personal data and privacy, see the Privacy Policy.
2. What you may and may not do
| Use | Status |
|---|---|
| Academic research, non-commercial | Allowed — with attribution |
| Academic publication, non-commercial | Allowed — cite as in section 5 |
| Analysis or statistics published without reproducing the full source text | Allowed — credit the original source |
| Commercial use (selling, paid services, paid products) | Not allowed |
| Training third-party AI models | Not allowed without the speakers' written permission |
| Redistributing third-party source text (section 3) | Not allowed |
Note on model training: the speakers who contributed recordings gave permission for recording, name, and publication. They were never asked about model training, so no such permission exists to rely on.
3. Not all of this data is ours to license
Parts of the corpus and dictionary come from works owned by others. We name them plainly rather than hiding behind pipeline labels:
- Bible translation (TB) — the largest single source. It is the Alkitab Terjemahan Baru, © 1974 Lembaga Alkitab Indonesia (Indonesian Bible Society), all rights reserved. The regional-language text derived from it is a translation of that work.
- Ponosakan–Indonesian dictionary (2023) — published by the Regional Education Office of Minahasa Tenggara.
- Research dissertation material — the work of its author.
For these parts we grant no redistribution rights, because they are not ours to grant. We use them internally for research and product development. If you need to republish them, ask the rights holder.
4. What the project does hold
- Field data — interview recordings, questionnaire answers, speaker notes, collected with the speakers' recorded consent.
- Work of project collaborators — including the dictionary and materials contributed by researchers who are part of this project.
- Derived work — alignments, embeddings, grammar rules, processed dictionary data, and the software itself.
5. How to cite
Cite using the project's CITATION.cff. That file applies to the data, not only to the software.
When you use a specific source, name it as it is — for example “Alkitab (TB, Lembaga Alkitab Indonesia)” or “Kamus Ponosakan–Indonesia 2023 (Dinas Pendidikan Minahasa Tenggara)” — rather than the internal pipeline label.
6. Third-party permissions we do not hold
We have not obtained permission from the Indonesian Bible Society for the TB translation, from the publisher of the Ponosakan dictionary, or from the author of the dissertation material — and we do not claim any. We therefore grant nothing over those works: the third-party layer stays “no redistribution, no model training” regardless of what we would prefer. If a rights holder grants permission, that changes.
One gap we acknowledge as fact: the name of the person who translated the Bible into the regional language is not recorded in our system.
7. Contact
Questions, permission requests, or corrections: [email protected].