Metro Stock Index: Performance of Top Companies in Major Metros (Updated every 5 minutes)
Average Metro Daily Change 0
Average Metro Change Since Jan 1, 2024 0
Average Metro Change Since Jan 1, 2023 0
Average Metro Change Since Jan 1, 2019 0

Facebook, YouTube and Twitter aren’t just for you or me. Terrorists use social networks as well — and extremist groups are getting more marketing-savvy by the day.

These groups have branding. They make memes. Their videos have production value with drone shots and 3D graphics.

And these videos are effective. ISIS was able to recruit people from all over the world to fight for their cause, around 25,000 from 100 countries.

Over the last four years, ISIS and its ilk have released about 1,300 videos on 400 different platforms. The increase in in the number of platforms had jumped in 2017 after the tech giants started cracking down on the videos. The need for a freely-available tool for monitoring video uploads was identified.

An Algorithm to ID ISIS Videos — As They’re Uploaded

The UK government funded the creation of such an algorithm, created by ASI Data Science. The cost of development was pegged at 600,000 pounds (roughly $830,000). UK Home Secretary Amber Rudd said that the algorithm, which is available as a free tool, was created to help smaller web publishing firms monitor their own networks. Tech giants like Google and Facebook have enough money to create their own detection algorithms but smaller firms do not have the wherewithal to do so.

In order to work, the algorithm must be installed on the publishers’ networks. The software will then evaluate all the videos being uploaded onto the servers in real-time. Questionable videos can then be scrutinized more closely.

Finding Patterns

The detection process is facilitated by machine learning. Using neural networks, a host of computers were “trained” to identify probable ISIS videos by exposing them to previous videos released by ISIS. After thousands of iterations, the finished algorithm learned to identify patterns and signs unique to ISIS videos. This refined algorithm can then begin to detect probable ISIS videos and flag them for security checks in real time. Human operators then get to have final say whether the video gets published or not. ASI Data Science says the algorithm has achieved 94% accuracy in detecting ISIS propaganda videos.

Founded in 2014, ASI Data Science specializes in applying artificial intelligence (AI) to real-world problems. It claims to have completed over 200 AI projects with some of the UK’s biggest companies like the BBC and EasyJet. One of the more interesting projects ASI has finished is an algorithm for the London Irish Rugby Club. The algorithm helps the club identify and recruit affordable but skilled players—sort of like Moneyball for rugby.