Algorithm Can Sniff Out True Tweets From False Ones
Because Twitter is by its very nature a fast way of spreading information, it is also very fast at spreading misinformation. Many people have fallen prey to Twitter death hoaxes, for example - everyone from Bill Cosby to Paris Hilton has been rumored to be dead on Twitter.
However, Twitter's problems as a truth machine are highlighted every time there is a tragedy. It provided information about the shooting in Newtown, Connecticut, many of it false. During Hurricane Sandy and its aftermath, users provided information, some helpful and some false. It is in those times that Twitter's misdeeds can prove to be ultimately extremely harmful. For example, during Hurricane Sandy, tweets emerged not to call 911 because the lines were jammed. Though Twitter's saving grace is that it can quickly rectify the misinformation that it spreads, by then, the damage can already be done.
New academic research is being undertaken in an effort to establish which tweets are truthful and which ones are lying. The hope is that an algorithm can be created so that false tweets can be automatically flagged.
For an experienced and knowledgeable Twitter users, the findings may seem obvious. False rumors were more likely to be spread with a question mark or another expression of doubt. Credible tweets tend to be longer and include links. They also are tweeted by people who have higher Twitter followers, and are generally negative rather than positive in tone. Another recent study found that true tweets are more likely to have a frowning emoticon rather than a happy one.
The new algorithm, which will be published in Internet Research next month, had an area under the curve of 0.86 if shown one random true tweet and one random false tweet. With 1 being perfect and 0.5 being no better than chance, that means that the algorithm was generally successful.
If the algorithm is used generally on Twitter, it may not be significantly better than a well-informed Twitter user. However, it may help stem the tide of misinformation. With increasing numbers of people using the microblogging site, it is becoming more and more harmful to post misinformation.