aqoHISTORY
A Question of History
From forgotten artefacts, we unlock hidden insights, use AI to uncover new connections, and tell the stories they reveal — differently.
Historic documents and wartime correspondence
Hosted on aqo.ai

aqoHISTORY Experiment

View the underlying data extracted

This is the data the aqoHISTORY digitisation process extracts from each original item, stored and indexed ready for stories to be created. Every image is divided into its separate parts — addresses, stamps, postmarks, passages, signatures — and the text in every part is tied back to the exact spot where it appears. Most documents show the raw automated reading; some have been checked by a person. Choose a document, then tap any outlined area to see what was recorded there.

  1. 1ImageEvery side and page of an original item is captured as a high-quality image.
  2. 2DivideEach image is split into its meaningful parts, and each part is identified: what it is and how it was made.
  3. 3RecoverThe text in every part is read, and it stays linked to its position on the page.

Experiment ID: aqoh-exp6

Status: proof of concept

Feature: interactive view of digitised documents — each page divided into its parts, with the text recovered from every one

Choose a document

From an image to a readable record

An image of a letter preserves how it looks. It does not let you search it, compare it with other documents or follow the people and places it mentions. The first step is to turn every image into a structured record, while keeping each piece of text anchored to where it sits on the page.

What the outlines show

Every outlined area is one part of the page: an address, a postmark, a stamp, a censor’s mark, a paragraph of a letter, a signature. For each part the record holds what kind of part it is, how it was made — handwritten, typed, printed or stamped — where it is on the page, and the words it contains. The colours group the parts by how they were made.

Honest about what cannot be read

Old documents are faded, stained, creased and overstamped. Where a word or a whole part cannot be read with confidence, the record says so rather than guessing, and it is shown here as unclear. Some documents have been checked and corrected by a person. Others show the automated reading exactly as it was produced, with nothing tidied up afterwards. Each document tells you which it is.

Why it matters

Once every part of every page is recorded like this, a collection can be searched by name, place or date. Postmarks can be lined up into journeys, and people can be traced from one document to the next. Individual pieces of evidence can be cited precisely. The stories on this site are built on exactly this kind of record.