Project to detect fake news using artificial intelligence "Fake News Challenge"

With the cooperation of more than 100 volunteers and 71 teams from academic societies and industry all over the world, the grassroots movement being done to eradicate fake news is "Fake News Challenge"is. Promoting the project isDean PomerleauWith Mr.Delip RaoMr. Mr. and two of us, the goal of the project is to explore the possibility of identifying nebulings or false alarms hidden in news articles using techniques such as machine learning, natural language processing, and artificial intelligence (AI), " Thing.

Fake News Challenge
http://www.fakenewschallenge.org/


The fake news is said by the New York Times "a fictional story made with the intention of deceiving the reader"DefinitionIt is being done. The move to correct this fake news is clearly becoming the most serious challenge in the current press circle and posting blacklist which already posted more than 200 sites that Washington Post delivers fake news ... ....

Criticism is gathered in blacklist articles that Washington Post enumerates over 200 as 'a lie spreading news site' - GIGAZINE


Google is expelling 200 sites that deliver fake news.

Google banished 200 sites that spoof false news - GIGAZINE


American think tank · Pew Research Center published in December 2016InvestigationAs criticism against false news is increasing day by day, about 64% of adults in the United States are answering that "news of freshly made" is "creating confusion" about the latest events.

Meanwhile, a project that started with the goal of "seeking the possibility of identifying forged or false alarms hidden in news articles using techniques such as machine learning, natural language processing, and artificial intelligence (AI)" "Fake News Challenge" was born. Project organizers Pomerleau and Rao think that it is possible to automate the fact confirmation process to confirm whether the content of human news is true or false with AI related technology.

Evaluating the accuracy of news is a complicated and awkward task for trained persons but fortunately the fact validation process can be broken down into several stages and its first step It is easy to automate. The first step is to understand exactly what "other press reports." This is called "Stance Detection" in "Fake News Challenge", and it is said that several teams and individuals are involved in development so that stance can be detected more accurately.


According to the timeline, on December 1, 2016, we began recruiting participation in the project on February 1, 2017, when the training data set is released, how much performance will be the baseline The announcement is announced on March 1, 2017, the deadline of application for participation in the project is May 1, 2017, the dataset for testing is released on June 1, 2017, the result of the test The submission deadline is June 2, 2017, and the result announcement is June 15, 2017.