Fake News analysis with Deep Learning

Equation explaining fake news — IEEE Computer Society

Problem:

Fake news has been accused of influencing election results and giving rise to populist movements and deaths. Is there anything governments — and citizens — can do to fight back?

Real problem:

What is the real problem our society is facing today? Climate change, epidemic, finding a solution for green power regeneration, waste management, etc. But the fake news always finds a way to distract us from the real problem, which we or our children will be facing.

How others are fighting against it:

Facebook’s “war room”

Facebook has already established a physical “war room” designed to bring staff together to find and destroy attempts to meddle with upcoming elections. This uses deep learning to tag fake news and ban the users and bots that are creating and spreading them.

UK startup Fabula AI reckons it’s devised a way for artificial intelligence to help user-generated content platforms get on top of the disinformation crisis that keeps rocking the world of social media with antisocial scandals.

Creation of fake news can’t be stopped, it is rooted deep into human psychology and nature. But with the help of deep learning, we can stop fake news from spreading.

Analysis of fake vs real news:

A visualization of a fake vs real news distribution pattern; users who predominantly share fake news are colored red and users who don’t share fake news at all are colored blue — which Fabula says shows the clear separation into distinct groups, and “the immediately recognizable difference in spread pattern of dissemination”. — Techcrunch

There are lots of bright minds currently growing awareness against it and working on such projects. But what we need is a full-fledged solution that can be applied globally to fight against this kind of behavior. Natural Language Processing+Sentiment Analysis can help us achieve this.

Real-life consequences:

Importance of different types of news outlets on Twitter. The number of distinct tweets (a) and the number of distinct users having sent tweets (b) with a URL pointing to a website belonging to one of following categories: fake or extremely biased, right, right-leaning, center, left and left-leaning news outlets. While the tweet volume of fake and extremely biased news is comparable to the tweet volumes of center and left volume (a), users posting fake and extremely biased news are around twice more active on average (see Table 1). Consequently, the share of users posting fake and extremely biased news (b) is smaller (12%) than the share of tweets directing toward fake and extremely biased news websites (25%)

Remember 2016 US presidential election? Influence of fake news on Twitter during that election is unforgettable. Here are some highlighted details from that analysis. To see the full analysis, refer to this article.

For those who doesn’t know about India’s 2019 elections, It happened again, and at a scale we never saw before. Government, media, news papers, everything is taking profit from people who fights with each other with misunderstood sociopolitical and religious issues, not the real issues of water crisis of Kerala, Assam flood, etc.

Retweet networks formed by the top 100 news spreaders of different media categories. Retweet networks for fake news (a), extreme bias (right) news (b), right news ©, center news (d), left-leaning news (e), and left news (f) showing only the top 100 news spreaders ranked according to their collective influence. The direction of the links represents the flow of information between users. The size of the nodes is proportional to their Collective Influence score, CIout, and the shade of the nodes’ color represents their out-degree, i.e. the number of different users that have retweeted at least one of her/his tweets with a URL directing to a news outlet, from dark (high out-degree) to light (low out-degree). The network of fake (a) and extreme bias (right) (b) is characterized by connectivity that is larger in average and less heterogeneous than for networks of center and left-leaning news

Conclusion:

We and our children could live in a future where our Government tries to solve real problems, people don’t fight with each other over wrong information and fake news, media will spread unbiased news. But this could be only possible with the help of Deep learning, as most of our news floats in social media, throughout the internet. The understanding of what is the truth, and what makes people respect the decision of other people for the collective good of the world always fascinated me.

I worked on a small project Twitter Sentiment Analysis to find out sentiment of tweets. This is a very small instance of what the world’s brightest minds are trying to achieve today, fighting against the fake news.