Applications of Artificial Intelligence to Detect Fake News

Fake news is defined as false stories that appear to be news, spread on the internet or other media, usually created to influence political views or for satire. It has been argued that fake news is one of the greatest threats to democracy today and a study by Pew Research Centre found 64% of US adults believe fake news has caused a 'great deal of confusion' about current events. Social media platforms often act as a catalyst for these kinds of articles to spread around the world quickly, where around 1 in 4 US adults have shared fake news either knowingly or unknowingly (Pew Research, 2016). More worryingly, around 59% of articles shared on social media have never been clicked by the sharer - often as a result of misleading (or 'clickbait') headlines. In some cases, companies such as Cambridge Analytica, have leveraged these platforms and news articles to target particular social groups thus influencing people's views around elections and referendums.
It is therefore evident that there are a number of issues surrounding the media we consume and a number of questions emerge when assessing the validity of news articles:
Is the article factually accurate?
Does the headline correspond to the message outlined in the body of the article?
Does the publisher have an agenda and if so, what is it?
Is it an older article that has been reposted? (IFLA.org, n.d.)
To address the stated problem and questions, the proposed PhD work aims to facilitate the verification of online articles and mitigate the propagation of fake news. To meet this aim, the project will involve the:
Investigation and development of state-of-the-art ML algorithms and NLP techniques that will help determine: the accuracy of an online article, whether the article title matches its content, whether the publishing source is trustworthy and the date of the original article.
Investigation and application of interaction design principles and HCI practices to ensure that the proposed system is transparent, trustworthy and usable to non-expert users.
This work draws insight and experience from preliminary work conducted as part of my dissertation which focused on the development of a ML-enabled mobile app to determine the political bias of news content using ML and then provide alternative articles on the same topic.
In brief, the proposed work will seek to leverage modern advancements in ML and Human-Computer Interaction to create a platform to prevent the spread of fake news. In particular, the project aligns exceptionally well with the research carried out by three research groups in the Department of Computer Science: i) the Intelligent Data Analysis group; ii) the Human-Computer Interaction group; and iii) the Interactive Multimedia Systems group. In addition, the project is well-situated within the Brunel Digital Science and Technology Hub of CEDPS and, in particular, the work conducted under the Industrial and Applied AI research theme. Moreover, it is congruent with the agenda of the newly formed Institute for Digital Futures of the University. More widely, it aligns with the Government's Industrial Strategy, namely "AI and data". Finally, 'Artificial Intelligence Technologies' is an EPSRC research area of growing interest, while 'Human-Computer Interaction' is an area of maintained focus.