Communications of the ACM
Stanford Researchers Find That Friends of Friends Reveal Hidden Online Traits
Researchers at Stanford University have demonstrated more ways than previously established to infer demographic traits that people might be trying to hide online. They tapped datasets reflecting the type of information that websites make available to advertisers or share with outside groups when people permit third parties to access their social profiles. "We set out to study the relationship between friend networks and predictability, and ended up uncovering an inference mechanism that hadn't been noticed before," says Stanford professor Johan Ugander.
The researchers note concealed traits, such as gender, can be deduced from studying the friends of friends by using a social structure called monophily, in which people have extreme preferences for traits that are not necessarily their own. They tested the method on academic datasets of friendship networks with complete information on individual traits, erasing gender data for specific individuals and applying their "friends of friends" analysis to correctly predict the missing data.
From Stanford University
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found