Twitter’s Latest Patent Filing Seeks to Identify “Fake News” Propagation
On February 18th, The USPTO published a patent application from Twitter, titled “Method of News Evaluation in Social Media Networks.” The method describes aims at identifying and evaluating news messages and their propagating behavior.
Twitter’s patent application describes, “frequently [it] is in the interest of a person, enterprise, public authority, political party asf, to distribute certain information such as news, advertisement, bulletins or warnings fast to a large number of people. This may help promote ones viewpoint, business or social status. On the other hand, people actually distributing the information may have an interest to ensure that information is distributed only when correct and/or in a manner not doing harm to them.”
The description of the patent application suggests, “social media are even suspected to have become one of the main sources of information for people around the world.” Further noting, “using social media for news consumption is a double-edged sword: on the one hand, it offers low cost, easy access, and rapid dissemination. On the other hand, it comes with the danger of exposure to “fake news” with intentionally false information.”
The filing acknowledges the “significant social and economical impact” from the spread of false information and relates to the “US 2016 presidential elections or the Brexit vote.”
“It is alleged that the outcome of these votes resulted from the public opinion manipulation by a massive injection of fake news, possibly produced by influence agents or even sponsored by hostile foreign governments…”
The patent application addresses recent public scrutiny of social media apps, stating:
The public opinion is therefore rethinking the responsibility of social networks such as Facebook or Twitter, which have thus far positioned themselves as mere media distribution platforms, essentially shaking off any liability for the published content. This stance is in clear contrast to the stance of conventional news distributors such as newspaper publishers, radio stations and so forth, and this stance is very likely to change in the near future, with eminent legislation in the USA that would hold social network companies accountable for the content published on their platforms.
Such regulation, if approved, should be expected to have a tremendous impact on Internet giants like Facebook, Google, or Twitter, as well as smaller media and advertisement companies relying on social network platforms. These companies are now in extreme need of technological solutions capable of combatting the fake news plague, and admit that existing technology is insufficient. Attempting to combat the fake news phenomenon and responding to the mounting public and political pressure, in March 2017 Facebook rolled out a content alert feature relying on users fact checking and flagging disputed content, which turned unsatisfactory and was taken down later that year. Applicant currently is not aware of any fully automatic commercial solution capable of reliably detecting fake news.
While many big tech companies have turned to algorithms and methods of filtering messaging for users, the patent acknowledges the resulting echo-chamber effects contributing to deeper problems. Furthermore, the application points to a series of studies that suggest the number of engagements to an article and the spaced time between engagements are a potentially powerful indicator of “fake news” versus “non-fake news.”
The proposed method intends to evaluate the spread of news messaging by:
- determining a social graph at least with respect to users and their social relations;
- determining a news message to be evaluated;
- determining a propagation behavior of the news message in the social graph
- evaluating the news message in view of its determined propagation behavior in the social graph
Through these evaluations, the invention intends to “exploit spreading patterns for automatic fake news detection.”
Initial testing data of the method “indicated that highly accurate fake news detection is possible in view of the propagation of a news message.”
“Fake news messages were found to be reliably detectable at an early stage of propagation within the network, typically after just a few hours of their “injection” into the network.”
The method also aims at flagging culprits that continually “inputs or likes or shares fake news” as unreliable sources of information.
Finally, the new method could also offer the potential as a marketing tool for major companies by identifying viral messages that contain “disadvantageous facts or allegations related to a brand product” and notifying a predefined recipient for damage control.
Additional details from the Twitter patent filing can be found at the USPTO Patent Search site.
Interested in more information about the patent process?
Click here to meet our Intellectual Property team and request a free consultation.