We all remember the time, in 2012, when the angry father of a high-school daughter showed the Target manager the mail sent to his daughter with coupons for baby clothes and cribs. And a few days later, it turned out, as the father said, that "there's been some activities in my house I haven't been completely aware of. She's due in August. I owe you an apology." If Google with words like "Target store detects pregnant high school girl," you can find millions of results relating to that story. What makes that event so hot in the search engine is that those behaviour changes that may be unnoticeable to others, even our closest ones, may wait to be seen, hiding openly in the data.
Though you may feel gooey about what Target was doing on the data it collected from its customers, on the other hand, we are probably in desperate need of the information those data fed. For example, imagine you are just out of luck understanding what your children are doing recently with the Internet. They already sit in front of the computer for a whole day but cannot finish the school homework sooner than midnight each day. As we talk to more and more parents about lockdown to their school-age children, the way they express is that the isolation impacts their kids' mental health profoundly. And those impacts, in turn, cause parents to worry about if their young adults are addicted to the Internet. Is there a way to answer those worried parents about their little ones' online mental health status? Probably yes. Ikuesan Richard Adeyemi, Shukor Abd Razak and Mazleena Salleh conclude that "a high probability that the signature of conscientiousness personality trait exists in online communication." You can find their report - "Understanding Online Behavior: Exploring the Probability of Online Personality Trait Using Supervised Machine-Learning Approach" (Front. ICT, May 31, 2016). That's a compelling message and offers hope that MangoProtection can make progress in finding an accurate algorithm that can predict online mental health disorders from only platform-independent features such as Web page visit characteristics, Web request characteristics, etc.