Anonymization
For comprehensive data protection - our AI accurately and quickly removes all personal information from your text data.
How it works in DataDesk
01 - Create a project, then create a study and upload desired data set.
02 - The data will be screened and you will choose which information to remove or replace from the texts.
03 - Our algorithms detect sensitive terms and phrases, depending on what you have determined in advance, and removes or replaces them.
04 - After a few seconds the anonymization is finished. Simply download the anonymized data set and continue processing.
5 languages
German, English, Dutch, French and Spanish can be processed directly, other languages after an automated translation.
10 sec - 10 min
short analysis time even with large amounts of data.
+ 5.000.000
Sentences have already been anonymized with deepsight
Questions and answers
Can I determine what exactly is anonymized?
Yes, you can. That means you can have names, titles, emails, phone numbers, addresses, countries, cities, job titles, dates, companies or any other contextual terms removed or replaced. In addition, you can create your personal blacklist of company-specific terms and anonymize to them, such as your own job titles or department names.
What happens when a project title sounds like a name and should not be anonymized?
No problem. You can whitelist words or phrases. All terms on this list remain in the texts. Optionally, you can also differentiate between upper and lower case here, e.g. the project title "TIM" is retained, while the personal name "Tim" is removed.
Can anonymized words also be replaced with a specific term instead of just removing them?
Yes, this is then called pseudonymization. By default, the words and phrases are replaced by "[anonymized]", but they can also be removed completely or replaced with "Loremipsum". Another option is to output pseudonyms per specifics, so for example "[ano_company]" or "[ano_birthday]".
At what point is it worth working with our AI?
Pretty much whenever you would need more than 10 minutes to anonymize manually. No need to suffer unnecessarily - leave the diligence to the algorithms.
How does our AI handle ambiguous cases like strange nicknames?
Insiders are often very difficult to recognize. For example, they may be nicknames that are only known in a small department. But don't panic. For these cases, we have introduced the so-called "conservative" option. Here, anonymization is guaranteed at the slightest hint of a name. The "preservative" option, on the other hand, leaves particularly dubious cases untouched. This comes into play when the difference between the name and the general term is virtually unrecognizable, e.g. in the case of "Hammer is the hammer" - where the first "hammer" is a real person.
Which languages can be anonymized?
Currently, our AI can process texts in English, German, French, Spanish and Dutch. However, we are constantly teaching it new skills and are also always open to your requests and inquiries.
Can our AI really match the quality of human anonymization work?
Definitely. Because unlike us humans, algorithms are not biased and never suffer from lapses in concentration. Although we may still not achieve 100% accuracy now and then (e.g. when you need to be able to read very fine nuances between the lines), we are already extremely close. Overall, we guarantee performance far superior to humans - not to mention speed.
the areas of application
Privacy
Security
Data processing
Research
Anonymization via the data desk
Our DataDesk uses AI technology to anonymize your data quickly and securely. Moreover, it can do a lot more - Apart from other data preparation options, such as data cleaning, we can also do open text analysis for you. We have designed an AI that can pull precise information from large amounts of data. Variable and self-learning, it adapts exactly to your needs and delivers - quickly, easily and, above all, successfully.
Other Data Services
ADJUST
Compliance Check
The data set is searched for compliance relevant texts potential jobs are marked and evaluated.
ADJUST
Sanity Check
The data is checked for sense and nonsense texts are removed accordingly.
Let's talk! Co-Founder Alex discusses with you how you can put our AI to work for you.
Dr. Alexander Meier
Co-Founder