Communications of the ACM
How Social Networking Sites May Discriminate Against Women
Columbia University researchers have shown how two common recommendation algorithms intensify a network effect known as homophily to effectively minimize the influence of women.
Credit: W. Ryan Holliday
Columbia University researchers have shown how two common recommendation algorithms intensify a network effect known as homophily, in which similar or like-minded people cluster together.
The research also illustrates how algorithms on a network with homophily effectively minimize the influence of women.
With the introduction of recommendation algorithms, women whose photos on Instagram were slightly less likely to be "liked" or commented on grew even less popular. Men's photos tended to be more popular, with 52% of men receiving at least 10 "likes" or comments compared to 48% of women.
Men were 1.2 times more likely to interact with other men's photos rather than women's, while women were just 1.1 times more likely to engage with other women.
When they added two widely used recommendation algorithms, the percentage of women connected to, or predicted to be recommended to, at least 10 other Instagram users fell from 48% in the original dataset, to 36% and 30% respectively. The disparity was greatest among Instagram's super-influencers, and when algorithms were used with this group, women's visibility dropped sharply.
From Columbia News
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found