DARPA launches Semantic Forensics project to identify fake news and online disinformation

Algorithms developed under DARPA's SemaFor project will be able to scan more than 500,000 stories, videos, images and audio files to identify fakes

US research agency DARPA has launched a new project intended to identify 'fake news' on internet.

The project, called Semantic Forensics (SemaFor), will try to create custom software capable of scanning more than 500,000 stories, videos, images and audio files to identify fake news stories.

SemaFor will have three trial phases over a period of next four years, according to Bloomberg. Based on trial results, the project may be expanded to include many more features in the new system, such as identifying malicious intent in the content.

In the first phase, the project will cover online news and social media. The second and third phases will focus on analyses of technical propaganda and conducting week-long 'hackathons', respectively.

During the training and testing phase, the algorithms will scan a mix of approximately 250,000 news stories and 250,000 social media posts with nearly 5,000 fake stories/posts. The algorithms will need to identity those fake posts in order to be considered workable.

DARPA believes that increasing the number of algorithm checks in the system will help it to identify fake news created with malicious intent before they have a chance to go viral.

Program manager Matt Turek discussed the program with potential software designed on Thursday in Arlington, Virginia.

"A comprehensive suite of semantic inconsistency detectors would dramatically increase the burden on media falsifiers, requiring the creators of falsified media to get every semantic detail correct, while defenders only need to find one, or a very few, inconsistencies," according to the supporting documentation for the Semantic Forensics program.

"These SemaFor technologies will help identify, deter, and understand adversary disinformation campaigns," it added.

Presently, DARPA is also working on another research program, dubbed Media Forensics (MediFor), which is intended to create an image authentication system. When completed, the system will not only be able to detect manipulations in images, but will also provide information about how those manipulations were performed.

So-called fake news and disinformation is having a significant influence on politics and society, given the ease with which it can be created and disseminated via social media.

Fake news is false information that is purposely created to deceive people or to harm an individual, institution/organisation or country. Fake news is claimed to have had an impact on the 2016 US presidential elections.

After Donald Trump won in 2016, several news outlets claimed that his election win was aided by dubious content pushed by Russia, over platforms such as Facebook and Twitter.

In recent days, China is being said to have adopted similar tactics to spread disinformation on social media outside of China - social media within China is closely policed and, effectively, state-controlled.

Last month, Twitter and Facebook said that they had suspended hundreds of accounts that they believed were linked to a state-backed misinformation campaign against pro-democracy protesters in Hong Kong.

In March, Czech company Semantic Visions was announced as the winner of 'The Tech Challenge' - a joint initiative of the UK and US governments to fight disinformation online.

As the winner of competition, Semantic Visions received a $250,000 grant to develop innovative technology to address the challenge posed by the spread of misleading and fake information over social media.