Managing Identites at Scale
- TYPE: Combined Session DATE: Thursday, May 17, 2018 TIME: 12:00-13:00 LOCATION: AMMERSEE II
Inferring User Identity
The outdated myth that a single human being controls a single account on your Internet service is no longer adequate to understand who is using your system for which purpose. We model identity in three tiers. The "group" tier includes households, families, companies, and other social circles of people who are connected in some way. The "human being" tier is our best-guess, frequently-changing approximation of the human being that may be using our system at the moment. The "account" tier is the user account that was used to authenticate this user. In this talk we explain how some of these approximations are currently inferred, and which behaviors give insights into probabilities associated with the probabilities / weights of each inferred link.
- Offering several users per household-account (e.g. Amazon, Netflix, GMX) will never guarantee that several people do not share accounts. One human or group can control several accounts on your system, possibly thousands.
- Behaviors can infer demographics and psychographics more accurately than self-identified information in a profile.
- Enormous cultural differences among different regions of the world prevent the use of a single trust / risk model for the users of your system
- Registration fee:
- Contact person:
Mr. Levent Kara
+49 211 23707710
- May 15 - 18, 2018 Munich, Germany