It is also possible that Facebook can apply for patents on technology to automatically detect unpleasant content such as pornography and hate speech, and it can be used to eliminate lie news

ByCharis Tsevis

The patent that Facebook applied in June 2015 was released. Published patent technology is a system that automatically detects content that makes users uncomfortable such as pornography and hate speech and pointed out that using this technology it is possible to automatically detect news of lie that is shared on Facebook It is being done.

Patent Images

According to the published patent "Patent 0350675", Facebook applied for "Objectable Material (a system that detects people discomforting content"), using machine learning to streamline an exclusion system for objectionable content, Learning common points of unpleasant content over time, making it possible to guess about contents to be eliminated.

The "Objectable Material" that Facebook repeatedly mentioned in patents is defined in the patent as "porno · hate speech · net bullying / illegal content or similar". Facebook has worked on eliminating content that makes readers feel uncomfortable for a long time and has established community regulations and is striving to secure users.

Community regulations | Facebook

Facebook patents found this time are mentioned as technologies for pornography, hate speech, cyber bullying, etc, but on the IT news siteThe VergeYaEngadgetArgues that the patented technology can be a tool to prevent the spread of lie's news which has become a problem in recent years.

Unlike pornography and hate speech including specific images and words, it is a fact that it is difficult to automatically detect lie's news including a wide variety of information. However, According to The Verge, the records of people who reported unpleasant content, "confirmation of profile", "number of people who reported offensive content", "account age of the reported users (Period from account establishment to report report) "that elements that become indicators when studying common points of machine learning unpleasant contents may be useful for automatically distinguishing lie news .

Following the report that Facebook's Mark O. Zuckerberg reported that many hoax information was shared on Facebook during the presidential election and affected the election results, "The ability to report and eliminate lie news is working , It is unlikely that 99% of the news displayed on the service is credible and it is unlikely that it had a major influence on the result of the election, but excluding the remaining 1% of the lie news and correcting it to the user It is our goal to deliver information "Published on November 12Furthermore, as a countermeasure to the lie newsProjects Facebook is working onWe revealed. Among them, he said, "In order to distinguish lie's news, a better system is needed to detect lie news before reporting."

A lot of you have asked what we are doing about misinformation, so I wanted to give an update.

The bottom line is: we ...

Mark ZuckerbergPosted byNovember 18, 2016

The Verge predicts that 'This is a better system for detecting lie's news before users report,' which Zuckerberg CEO said about the patented technology that was revealed this time, but Facebook Public relations is said to have responded to The Verge "Our company does not adopt the technology even if it gets a patent, this patent does not indicate our future plan".

in Web Service, Posted by darkhorse_log