Just half an hour of web browsing is enough time for machine learning mechanisms to uncover a person’s personality and produce identifying digital signatures, researchers say.
Those traits can include conscientiousness and neuroticism, among other characteristics, the scientists from Universiti Teknologi Malaysia say in their media release published by AAAS, the science society.
And it might identify the individual, too.
"Our research suggests a person's personality traits can be deduced by their general internet usage,” says Dr. Ikusan R. Adeyemi, a research scholar at the university.
+ Also on Network World: The shocking truth of how you'll be tracked online and why +
He means web browsing, rather than social network usage. Social networks have been used before to figure out whether people are gregarious or introverted, the scientists explain.
However, this study of network traffic, published in Frontiers in ICT, is different because it “can also reflect their choice, preference and reflexes, which is largely controlled by their unique psychological characteristics.” Those characteristics, such as openness to new experiences and even agreeableness, are discovered through machine learning.
It’s an important development because marketers could adapt their offerings if they can guess how receptive a consumer is going to be to a product. A consumer open to new experiences could be pitched new kinds of products, say. And a known-to-be disagreeable consumer could be pandered to—or not.
Online advertising has its fair share of detractors for a reason. Intrusiveness, being forced to view videos before being allowed to view content, pop-up ads, quantity of ads causing repeated browser slowdowns, and generally irrelevant ads are among causes of consumer griping. The more appropriate the ad, the more lean or effective the ad medium could be.
Behavioral analysis and security
But it isn’t just purveyors of Internet stuff and the intended recipients who might benefit from behavioral analysis.
"It can also be used as a complementary way of increasing security for online identification and authentication. Law enforcement agencies can also apply our findings in the investigation of online crime cases," says Adeyemi in his release.
Behavioral patterns are a holy grail for marketers and others. The internet is providing its fair share. I recently wrote about how smartphone Wi-Fi radios are being used to collect breadcrumb data on pedestrians as they move around a university campus. Building managers can use that electronically gathered intelligence to decide where to place resources such as ATMs.
Most of the data being collected and distributed by wireless providers, for example, is said to be anonymized. That’s done through the simple removal of the identifier, such as a name or phone number, before the data is passed on to the marketer. One further thing that’s interesting about the Universiti Teknologi Malaysia study is that anonymity, as we know it, may not apply any more.
“The notion of online anonymity is based on the assumption that on the internet, the means of identification are limited to network and system identifiers, which may not directly relate to the identity of the user,” the paper says.
But what about the “digital fingerprint of personality trait”? We’re all unique. If the machine learning gets good, are we not going to be identified by our behavioral signature?
Stripping the names or phone numbers out of a database becomes pointless. We’re identified anyway.
This article is published as part of the IDG Contributor Network. Want to Join?