Get involved

Send us your own datasets and archival sources for processing, and watch the Archos sort through your material.

We are looking for ...

Historians, archivists and curators with a research question and a corpus, particularly one large enough to be genuinely unmanageable by hand.

Digital humanists working on retrieval, transcription, source criticism or evaluation, who want to compare approaches or build on the architecture.

Or any one who is interested in seeing how an epistemically grounded LLM system works on their own material.

What helps most

• Transcribed or machine-readable text. Plain text, XML/TEI, CSV, or PDFs with a text layer. We can discuss OCR and HTR for material that has not been transcribed yet.
• Source identifiers and dates. Whatever catalogue reference, shelfmark or archival id you already use.
• A real research question. Archos works from a research question. Your research question can be as long as research proposal, or as short as a single sentence.
• Any known problems. Gaps, misattributions, contested datings, duplicate copies — these are exactly the cases we want to see Archos handle rather than smooth over.

What you get

An Evidence Report on your material: a source-grounded text report with sentence-level citations back to your source documents, ambiguities left standing and contradictory sources shown side by side.

A chance for you to collaborate with us or join the project and improve Archos with your expertise.

What we do with your material

We use it to evaluate and improve the system, and we discuss the terms with you before anything is processed. Tell us about any access restrictions, licence conditions, embargo or third-party rights attached to the material and we will work within them — including declining material we cannot handle appropriately. We will not redistribute your sources, and any use of your collection in a publication or demonstration happens only with your agreement and with whatever attribution you ask for.

If your material carries personal or otherwise sensitive data, say so up front so we can agree how to handle it before you send anything.

Get started

Email sebastian.eck@music.ox.ac.uk with a short description of the collection, its rough size, the format it is in, and the question you would like to put to it. We will reply with what we would need and what we think the system could do with it.