Half a year passed since GDPR regulation has come into force in Europe, but how to live with it is still on a radar for business. Why? The fines are too high (up to 4 % of the total enterprise revenue), while usage of existent compliance tools is limited.

One of the challenges to still cope with is anonymization of personal data. We all know that it must be anonymized per law, but the quality of data once it has gone through an anonymization process is not the same as of the original dataset. This is a paradox and a nice problem to tackle, especially, with regards to Machine Learning technology: unfortunately, machines (just like humans) cannot learn on random unstructured data. They need connections between different fields in a dataset to memorize and interpret the information.

Now you are probably wondering, what has been done to overcome those issues and make Machine Learning run efficiently on an anonymized dataset? And you are right, something has already been developed. But the picture is still not that ideal as it might seem at a first glance.

So, what are major trends currently present on the market and why the problem remains unsolved?

  1. Data is masked but not anonymized. Just hiding the names of people under keys that can still be deciphered by someone is not a good option that makes you compliant with GDPR.

  2. Before anonymizing the data, you spend quite some time to understand which exact parts of dataset need to be anonymized. Nobody tells you what to do in a form of an educational KIT. This makes anonymization a long and costly process.

  3. Almost all relevant anonymization techniques tailored to Machine Learning usage that are currently being explored by the giants in the data processing field (Google etc.) are at the R&D stage and do not represent a ready-to-use solution.

  4. There is no an end-to-end approach meaning consolidation of single technologies in one solution which fits all possible anonymization use cases.

That is why there is still enough room for innovation and that is why us, SAP people within SAP.io Start-up Challenge, are trying to make the one-size-fits-all anonymization possible. How? With a brand-new solution that helps our customers innovate without breaching GDPR regulations and run a truly Intelligent Enterprise.

Are you interested to learn more? Would you like to be one of those who will test the new solution first? Then, please, help us to figure out how beneficial such a solution might be for your company, answering the following set of questions. Please, be sure that we will not share this data with anyone.  You can also leave some of the non-mandatory fields (for example, ‘’Company name’’) blank. Your responses matter a lot for us, so, thank you in advance for participating in our survey!