Skip to main content
Version: Next

Personalization data flow

Azure / Machine Learning hosting provider​

The following colors in this dataflow represent three asynchroneous timelines :

  1. (blue): Recording user interest through their navigation in the app and by their answer in the habbits questionnaire.
  2. (green): Generating machine learning model using keywords logged from each pseudonomized users
  3. (red): Providing recommendation to the user for motivation feature.

Azure is only an hosting provider in this scenario. Data is still supposed to be owned by AOK.

Important considerations :

  • User ID is never stored in Azure Database. The link with their pseudonym is stored on ITSCare.
  • Azure components are not allowed to reach any data hosted on ITSCare. The data flow is mono-directional only : from ITSCare to Azure.
  • User age and user location could be used as keywords, and thus saved in Azure. It is also possible to use age ranges and location range instead of exact information.
  • The database used by the recommendation engine only stores keywords. It does not know the user data that have generated these keywords.

Adobe Target / Kameleoon / 3rd-party SaaS​

The following colors in this dataflow represent three asynchroneous timelines :

  1. (blue): Recording user interest through their navigation in the app and by their answer in the habbits questionnaire.
  2. (red): Providing recommendation to the user for motivation feature.

Important considerations :

  • AOK has no direct control on algorithms used by Target/SaaS.
  • AOK has no direct control on data management and use on Target/SaaS.
  • Data ownership on Target/SaaS depends on their terms and conditions.
  • User age and user location could be used as keywords, and thus saved in Azure. It is also possible to use age ranges and location range instead of exact information.