London based startup Fabula AI has developed a deep learning-based system that can help identify fake news across online platforms.
“Automatically detecting fake news poses challenges that defy existing approaches based on linguistic content analysis,” the company stated in a blog post. “News is often highly nuanced and their interpretation requires the knowledge of political or social context or common sense, which current natural language processing algorithms lack.”
The company says they have developed a new class of machine learning algorithms called Geometric Deep Learning, which are comprised of convolutional neural networks that are capable of learning patterns on complex and distributed data sets such as social networks. (Fabula’s Chief Scientist Michael Bronstein has written a paper describing the new algorithm in conjunction with Yann Lecun, Director of AI Research at Facebook.)
The geometric deep learning algorithm looks at how stories are shared as opposed to focusing on the content of the news, learning patterns that are unique to the spread of fake news.
During training, the company used a subset of Twitter data which included around 250,000 user profiles and 2.5 million social connections. The training data was paired with third-party fact-checking information from PolitiFact and Snopes.
“Among the key advantages of Fabula’s technology compared to content-based methods is that it is agnostic to the content and language of news, and much harder to beat by adversarial techniques, as it relies on the collective behavior of the social platform users,” the company said.
The company’s algorithm is still in development and has not been commercially tested on different social media platforms. However, the company intends to do so this year by offering an API for platforms and publishers.
Right now Fabula AI says their algorithm can recognize fake news with 93% accuracy within 2-20 hours of a post publishing. It can take a manual-fact-checker much longer, the company says.