Spotting Fake News Intended To Manipulate Financial Markets

While fake news is often associated with the distortion of political and scientific discourse, it also has obvious implications for financial markets, with misinformation potentially distorting share prices around the world.  New research from the Universities of Göttingen and Frankfurt and the Jožef Stefan Institute in Ljubljana explores how such attempts can be detected and stopped.

Financial misinformation is often designed to portray a company in a positive light, and the researchers used machine learning to develop classification models to help identify suspicious messages based upon both their linguistic characteristics and their content.

“Here we look at other aspects of the text that makes up the message, such as the comprehensibility of the language and the mood that the text conveys,” the researchers say.

Suspicious text

It’s an approach that is already commonly used by email companies to detect spam.  Detecting malicious content is a constantly moving challenge, however, as fraudsters adapt their content to avoid words and phrases that trigger fake news alerts.

The researchers aim to overcome this via a combination of models that provide them with high detection rates and a generally robust approach that ensures that even if suspicious words are stripped from the text, the linguistic features alone are enough to trigger a warning.

“This puts scammers into a dilemma. They can only avoid detection if they change the mood of the text so that it is negative, for instance,” the researchers say. “But then they would miss their target of inducing investors to buy certain stocks.”

The researchers believe that their approach could have a range of applications, including, for instance, market surveillance so that trading could be suspended temporarily in affected stocks.  It could also provide support to investors to help them avoid falling for fraudulent schemes, and possibly even for criminal prosecutions.

Facebooktwitterredditpinterestlinkedinmail