Hot news

Linguists and AI experts define the language specifications of fake news


Many dictionaries and linguistic institutions launched the term "false news" in 2017, RT reported, and with the development of social media , many of us have clarified that if something seems too good or bad to be true, it is often a lie or a distortion, but what about language itself? Can you give an idea of ​​the authenticity of the text we are reading?
Linguists at Norway's University of Oslo are working with artificial intelligence experts to debunk the language of fake news they call Fakespeak.
"Our goal is to improve existing fact-checking tools," said project manager Celje Suzanne Alvestad.
In 2003, New York Times journalist Jason Blair was convicted of fabricating a number of news articles. Scholars collected these fake texts and compared them with a selection of real news stories written by Blair, and it turned out that the texts were different in terms of the style of their wording.
The researchers found several significant language differences, including:
The pseudo-texts had a more informal style, while the authentic texts contained a greater density of information.
- In authentic texts, the frequent use of nouns and words that replace nouns is noted. Words were longer, on average.
- It was noted in the pseudo-texts the frequent use of verbs, especially in the present tense, in addition, pronouns, adjectives, words and emotionally colored interjections were more common.
"He (the journalist) also used less metaphors in his fake news articles than when he wrote the truth," said Celje Susan Alvestad.
In addition, it is interesting that Blair often uses language elements that describe or attempt to evoke positive emotions, which is unusual for fake news, which is usually prone to scaremongering.
The researcher said that this could be related to the topic. Many of Blair's texts are fake stories about heroic American soldiers during the Iraq war, as the journalist tried to present the American Iraq war in a positive light.
However, Jason Blair's texts are only 80 pages long, and AI experts prefer to work with much larger datasets, so a bunch of texts by different authors from verification services have been added.
The researchers are now working on defining the standards for fake news in other languages, and they are sure that this will be a huge boost to combating fake news online.