Email spam and fake news are really not that different. Unsolicited bulk email with the goal to promote services or products mostly relies on “baiting” techniques to get the recipient “hooked” and covert it to a paying customer. Just like fake news. But email spam rarely has any success nowadays due to the efficient automated spam filter put in place by antivirus software and email server providers.
So why are we still relying on human fact checkers to clear the mess that fake news is creating nowadays?
Well, it’s mostly because adopting an automated tool for filtering out untrustworthy content – aka “fake news” – is still regarded as something that journalist need to be doing. Producing high quality trusted content and debunking misinformation is on every honest journalist’s job description, but constant firefighting shouldn’t be. That’s because the volume of fake news and click-bait content produced on a daily basis is getting growing rapidly beyond the verification capabilities of any human, and is now becoming a task for 100% software automation tools, like TrustServista. Just like spam fighting.
Still, the fear of false negatives (true content labeled as untrustworthy or the other way around, false positives) is preventing media professionals and readers to rely entirely on automated verification tools. The same way, people tend to be much less bothered by spam slipping through filters into their mail box (false negatives), than having desired email (“ham”) blocked. The success of any spam filtering system depends on combining its capabilities to block unwanted content with allowing desired email go through.
To date, statistical content filtering approaches have proved to be the most effective. And that’s exactly how we handle news content with TrustServista.
By looking at content like if it was email content, we can implement statistical algorithms to filter our “spammy” looking content from high quality content. Continuous processing of content producers allows us to create rules and block sources of constant untrusted posts and find patterns into how fake news content is being created and distributed. And all with the sole purpose on giving the consumer a highly intelligent, effective and also transparent filtering system for news. And as even the most trusted spam fighting systems will same the filtered out content in a “Spam” folder in case the user needs double-checking for false negatives, the same way a trusted news verification system needs to allow users to access and verify the content that as flagged “untrustworthy”.
TrustServista currently processes 200,000 English-language articles per week. Out of those, 10% are automatically flagged by our statistical algorithms as being highly untrustworthy. That’s 20,000 posts, per week. 60% of all email traffic in 2017 was spam. All filtered out automatically. By comparison, the popular manual fact checking websites like FactCheck.ro only debunks 30 to 40 stories per month!