The “fake news” phenomenon may have attracted Americans ’imagination during the 2016 presidential election, and subsequently investigated Russia ’s attempts to use false news on Facebook to push the election to Donald Trump.
The fact is that in many years before the 2016 election, fake news or telephone news has been widely spread as a tool to spread propaganda and conspiracy theories. Websites including InfoWars and Brietbart are spreading fake news to support their agenda.
However, since the election, this has become a political and social issue, and poor Facebook has become a typical representative of the site that failed the plan.
Recently, the social media company admitted its mistakes and tried to make subscribers correct. Now, it is marking phone news articles to be sent to Facebook members through its news feed. It is using AI to achieve this goal.
The company is using AI to identify words or phrases that may indicate that a product is actually fake. The data for this task is based on articles that Facebook members have marked as false stories.
The technology currently uses four methods to discover false news. They include:
- Rate the page. The first to use this technology is Google. It uses facts to create scores for websites. Obviously, scoring the website is an improvement. However, as Google has been doing, this technology has made great strides.
- Weigh the facts. This method uses a natural language processing engine to review the theme of the story. AI uses other models to determine whether other sites report the same facts.
- Predict reputation. The technology is based on AI, which uses predictive analytics and machine learning to predict the reputation of a website by considering many functions, including domain names and Alexa website rankings.
- Found sensational words. Supporters of fake news use sensational headlines to attract the interest of potential audiences. This technology uses keyword analysis to discover and mark false news headlines.
It is an arduous effort for AI to actually detect these types of articles. Of course, it involves the analysis of big data, but it also involves the accuracy of the data. Determining it actually involves determining the authenticity of the data. This can be done by balancing facts. What if fake news articles appear on hundreds of websites at the same time? In this case, the use of technology that weighs the facts may cause the AI to determine that the story is legal. It may be helpful to combine the method of predicting reputation and balancing facts, but there may still be problems. For example, a reliable news source website can assume that it is true if it does not take time to verify the news report.
Obviously, using AI to identify these articles requires more development. Many organizations are involved in enhancing AI capabilities. West Virginia University is one of them.
The Reed School of Media and West Bengal University ’s Benjamin Stettler School of Engineering and Mineral Resources collaborated on a course that focused on using AI to recognize voice news.
Senior students participating in computer science electives are developing and implementing their own AI programs in teams, and are also involved in the project.
Another group called fake News Challenge It is also seeking ways for AI to successfully combat fake news. It is a grassroots organization composed of more than 100 volunteers and 71 teams from academia and industry, aiming to solve the problem of fake news. It is developing tools to help people conduct fact checks and identify fake news stories.
When organizations are committed to enhancing AI to find these stories, a variety of tools can be used to combat them. These include:
Spike can identify and predict breakouts and virus stories, and use big data to predict the factors that drive interaction.
Scam, this is a tool that helps users identify fake news sites.
Snoopey, this is a website that helps identify phone news articles.
CrowdTangle, this is a tool that helps monitor social content.
Meedan, this is a tool that can help verify online news.
Google Trends, used to monitor searches.
La Decodes from Le Monde, this is a database of fake news and real news websites.
Pheme, this is a tool to verify the accuracy of user-generated online content.