Finweek English Edition - - INVESTMENT -

Re­searchers have made the scary rev­e­la­tion that what Face­book users choose to ‘like’ on the so­cial net­work cre­ates pat­terns that pro­vide ac­cu­rate snap­shots of the user’s race, age, sex­u­al­ity and other de­tails. The study looked at 8 000 US Face­book users who vol­un­teered their likes, de­mo­graphic pro­files and psy­cho­me­t­ric test­ing re­sults. By ap­ply­ing an al­go­rithm that uses Face­book ‘likes’, the study proved al­most 90% ac­cu­rate for dif­fer­en­ti­at­ing males from fe­males and 95% ac­cu­rate in dis­tin­guish­ing African-Amer­i­cans from whites. Chris­tians and Mus­lims were clas­si­fied in over 80% of cases, and ac­cu­racy of be­tween 65% and 73% was achieved for re­la­tion­ship sta­tus and sub­stance abuse. Th­ese re­sults will also in­ter­est ad­ver­tis­ers who can make good use of such in­for­ma­tion.

Newspapers in English

Newspapers from South Africa

© PressReader. All rights reserved.