Creative · Community report
Turn historical records into an interactive map
After years of trying to map historical Vienna address books, one person on X says they finally did it with Muse, combining open city data with OCR on Wienbibliothek scans from 1938 to 1942. It is one dataset, so this guide covers adapting the approach to your own historical records and checking the OCR against the originals.
What you can get
What it is
Interactive map
An interactive map built from your historical scanned records, checked against the originals before you publish it.
It came back right when
- A sample of mapped addresses matches both the original scanned record and a real, period-appropriate location.
- Names and addresses in the map match what OCR actually extracted, not a guessed correction.
- The map's data sources (scans and geographic data) are both correctly attributed.
- Any uncorrected OCR errors are flagged rather than silently plotted.
The checks are editorial, not from the source.
- Confirm sources and rightsStops for your approval
Check the scans and geographic data are openly licensed or rights-cleared.
- OCR the records
Extract text from scanned pages.
- Match to geography
Place each record using period-appropriate open data.
- Audit a sampleStops for your approval
Check entries by hand before publishing the map.
Editorial, not from the source
Works when
- You have scanned historical records with text that can be OCR'd, such as printed address books, directories, or registers.
- Open or openly licensed geographic data exists for the place and period you're mapping.
- You're prepared to check extracted text against the original scans rather than trusting OCR blindly.
Does not work when
- Your source is handwritten rather than printed; OCR accuracy on handwriting is not addressed by this report.
- You don't have rights or an open license to the scans or data you want to use.
- You need the map finished quickly; the author describes this as something they'd been trying to do for years.
Turn historical records into a map: review template
| Record/entry | Source and OCR check | Status |
|---|---|---|
| [Sample address entries] | [Original scan vs. OCR text] | [Matches] |
| [Mapped location] | [Period geographic data] | [Plausible] |
| [Licensing] | [Scan and data source] | [Confirmed open/rights-cleared] |
The source case
On X, @bermaniastudios reports finally mapping historical Vienna address books after years of trying, built with Muse. I've been trying to put historical Vienna address books on a map for years. Built with @muse: They name the specific source, Lehmann's Wohnungsanzeiger from 1938 to 1942, and say it combines open city data from Vienna's city administration with OCR performed on open-licensed scans from the Wienbibliothek. This is one report, checked on 2026-10-09, of one specific historical dataset; it does not establish how Muse handles OCR accuracy, other archives, or other languages.
A public report by someone outside this library. Not independently reproduced here.
From the source
Source excerpt
I've been trying to put historical Vienna address books on a map for years. Built with @muse:
Quoted from the original posts. Not written by this guide, and not a result produced here.
Suggested prompt
Bring Scanned historical records (the report uses address-book pages), Open or licensed geographic data for the area and period. Acts Publishes a public map or dataset once you're ready to share it. Source community report, checked 2026-10-09.
Written by this guide as a starting point, not quoted from the source. Your edits stay in this browser.
Other ways to ask
Editorial rewrites of the brief above, for a different emphasis. Copy them separately.
What you need
Editorial, not from the source
You supply
- Scanned historical records (the report uses address-book pages)
- Open or licensed geographic data for the area and period
Connected
- The scan source (the report names a specific library's open-licensed scans)
- The open geographic data source (the report names the Vienna city administration)
Approved by you
- Confirmation that your source material is open-licensed or that you otherwise have rights to use and publish it
How to try it
Editorial, not from the source
- Identify your historical source (the report uses scanned address-book pages) and confirm you have rights or an open license to use it.
- Have the scans OCR'd into text, matching addresses and names to what's on the page.
- Match the extracted addresses against open geographic data for the area and period, to place each record on a map.
- Check a sample of mapped points against the original scans before publishing or sharing the map.
"For years" is part of the story
The author explicitly says they'd been trying to put these records on a map for years before Muse helped them finish it. That detail matters: it suggests the hard part of a project like this isn't necessarily the idea, it's the grinding work of OCR, matching, and correction across a large historical dataset. If your own project has stalled for the same reason, this report is evidence that an agent can help carry that load, but it doesn't say how much of the years-long effort Muse replaced versus how much manual checking still happened at the end. Go in expecting to do real verification work, not expecting the years of friction to disappear entirely.
The specific sources matter as much as the task
This report names its sources precisely: a specific address-book series, Lehmann's Wohnungsanzeiger, 1938 to 1942, open data from Vienna's city administration, and OCR performed on open-licensed scans from a named library. That specificity is a strength of the report and a limit on what it proves. It confirms the task works for printed records with available open geographic data and open-licensed scans; it says nothing about handwritten records, archives without an open license, or places lacking comparable open geographic data. Before starting your own project, check whether your source material and the geographic data you'd need are actually available on similarly open terms, since a missing license or an unmapped region can stall a project as easily as a skill gap can.
Treat the first map as a draft to audit, not a finished answer
Historical OCR is rarely perfect, old typefaces, faded ink, and inconsistent spelling of place names all cause trouble, and the report gives no number for how accurate the extraction was or how much correction happened before publishing. Before you call a historical map finished, pull a sample of entries, maybe a few dozen, and check each one by hand against the original scan: does the name match, does the address match, does the point land in a place that could plausibly correspond to that address in that period? A map built from imperfect OCR and plotted with confidence is more misleading than one that visibly flags the entries nobody has checked yet.
Open Muse to try it. That is the product's own site; this guide is not part of it.
Limits and confirmation points
Limits to confirm
- One report, about one specific archive (Vienna address books, 1938–1942); it doesn't show how OCR or mapping performs on other archives, handwriting, or languages.
- OCR accuracy on historical scans is not quantified in the report; expect to check and correct some entries.
- The report doesn't say how long the project took or how much manual correction was needed before the map was considered finished.
- Open data and open-licensed scans were specifically used; a project on material without an open license or open data may need separate rights clearance.
Stop and check
Editorial, not from the source
- Before OCR: confirm license/rights status of the source scans.
- After OCR: spot-check a sample against the original pages for accuracy.
- Before publishing the map: verify a sample of plotted points against both the OCR'd text and the geographic data source.
If it goes wrong
Editorial, not from the source
- OCR misreads historical type, old spellings, or faded scans
- Check a sample manually against the original scans; the report doesn't quantify OCR accuracy, and historical printed text often needs manual correction.
- Addresses from the period no longer match a usable modern location
- Use period-appropriate geographic data where available (the report uses historical open city data), and flag addresses that can't be confidently placed rather than guessing.
- You're unsure if your source material can legally be mapped and published
- Confirm the license or rights status of both the scans and the geographic data before publishing; the reported project specifically used open-licensed sources.
Common questions
Can Muse really build a historical map from scanned records?
One user reports doing exactly this: mapping a Vienna address-book series from 1938 to 1942 using OCR on open-licensed scans and open city data. This is one report about one dataset, not a guarantee for other archives.
Does it work on handwritten records?
The report describes printed address-book pages, not handwriting. OCR accuracy on handwritten historical documents isn't addressed.
How accurate is the OCR?
The report doesn't give a number. Historical OCR commonly needs manual correction, so plan to check a sample of entries against the original scans.
What licensing do I need?
The reported project specifically used open-licensed scans and open city data. If your own source material isn't openly licensed, you'd need to clear rights separately before publishing.
How long does a project like this take?
The author says they'd been trying for years before finishing it with Muse. The report doesn't break down how much time the Muse-assisted portion took versus earlier manual attempts.
Next
Related to this task
More in Creative
Get recurring task recommendations — ask your own agent to check for relevant updates.