NameScrub
NameScrub removes personal names and other identifying details from text. Processing happens in the browser, so the text stays on your device.
NameScrub removes personal details from text. You paste in a text, and the tool marks what it has found. You check the marks and copy out the result. Everything runs in the browser, and the text stays on your device.
Purpose
You want to file a letter, pass on a report or have an AI assistant analyse the minutes of a meeting. First, the names and contact details have to go. Doing that by hand is laborious, and it is easy to miss something.
There are cloud services for this, but you would have to entrust them with exactly the data you want to remove. And there are specialist applications that need installing and setting up. Neither suits the one document that needs passing on now.
Detection
Names, email addresses, telephone numbers and IBANs are always detected, and for IBANs the tool also validates the checksum. Dates and addresses can be switched on.
NameScrub overlaps with Datenlotse on the classics: names, addresses, contact details, dates and IBANs. Even so, the two are cut differently. Datenlotse follows an administrative taxonomy with file references, payment references and insurance numbers. NameScrub is built for general correspondence, with interactive review. For official notices, use the administrative tool; for a quick clean-up before passing something on, use this one.
Use
You paste the text, run the detection, work through the findings and copy out the result. The review is not a formality: you see every mark and decide whether it stays.
If you need more thorough detection, you can use NameScrub+, a desktop application with a trained model for German named entity recognition. It also picks up places and organisations. The price for that is an installation.
Limits
Detection in the browser works with a dictionary and context, not with a model. An unfamiliar surname is found when something gives it away: a form of address or a first name before it, a fitting verb, a pairing or a repetition in the text. If it stands alone at the start of a sentence, it stays in.
Surnames that are also ordinary words are the hard case. “Frau Stark” is recognised, “die Aussage Stark” not reliably. Conversely, with names such as Richter, Koch or Bauer, the tool marks too much rather than too little. That is a deliberate choice, not a flaw in the implementation: a person who is missed weighs more heavily than a word blacked out needlessly.
File references, tax identification numbers and vehicle registrations are not recognised. Dates are recognised only in numeric form, addresses only with a street suffix. The web version does not recognise places or company names at all. That would take a lexicon of proper nouns, and NameScrub+ provides one.
Nor does it recognise indirect descriptions: “the claimant's older brother” stays in, although the phrase identifies a person. That is why the review remains obligatory. The tool takes the mechanical work off your hands, not the responsibility.
Origin
There is a gap between cloud services and installed specialist applications. The services would require you to entrust them with precisely the data you want to protect. The applications are not worth it for a single document. NameScrub closes that gap: you open it, paste, check and copy, and everything stays on your own device. The desktop version came later, for higher demands on detection.
The source code is public on GitHub, under an MIT licence.