Communications of the ACM
Fake News Detector Algorithm Works Better Than a Human
An algorithm-based system developed at the University of Michigan identifies linguistic cues in fake news stories could provide news aggregators and social media sites with a new weapon against misinformation.
Credit: umich.edu
Researchers at the University of Michigan have developed an algorithm-based system that identifies linguistic cues in fake news stories, which could provide news aggregators and social media sites with a new weapon against misinformation.
The researchers found the system successfully identified fake stories up to 76% of the time, surpassing the human success rate of 70%.
In addition, the linguistic analysis approach could be used to identify fake news articles that are too recent to be disproved by cross-referencing their facts with other stories.
The researchers created their own data by crowdsourcing volunteers recruited with the help of Amazon Mechanical Turk to turn short actual news stories into fake news items, mimicking the journalistic style of the articles. The team then fed the labeled pairs of real and fake news stories into an algorithm that performed a linguistic analysis, teaching itself to distinguish between real and fake news.
From University of Michigan News
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found