Solving the Fake News Problem through Network Theory
Link: https://www.theguardian.com/technology/2016/nov/29/facebook-fake-news-problem-experts-pitch-ideas-algorithms
In 1996, Stanford Computer Science students Sergey Brin and Larry Page famously published their renowned paper: “The Anatomy of a Large-Scale Hypertextual Web Search Engine.” Since then, this paper was been academically cited 16839 times and the engine that they have proposed has turned into the famous company Google, which now has a valuation of almost half a trillion dollars. Their paper proposed the PageRank algorithm, in which the relevance of a website is determined based on an iteratively computed algorithm of the websites that are linked to it and their “authority” score.
Nicky Woolf in “How to solve Facebook’s Fake News Problem” references this idea of using a node-based approach to solve the fake news crisis. In recent years, with the prevalence of social networks and social media, fake news has become a much bigger problem. According to the article, fake news was in circulation far more than regular news prior to the 2016 election. And with the recent political polarization, fake news is far more a threat than most people think. The article claims that individuals are now posting on forums such as Y Incubator’s HackerNews and coming together to come up with solutions to much of this crisis.
Some of the proposed solutions include a cross-partisan index score in which the people who share an article represent a node in a network and the score of an article is determined by the balance of liberal leaning and conservative leaning nodes. If an article is shared far more by one political group than another then it can be deemed to be a partisan article that most likely contains misleading or faulty information. Another proposed solution is to develop a sharer ranking score for each node that gives a news article a lower score if many people who have shared fake news in the past share the article. Lastly, a proposed solution is to build off of Page and Brin’s PageRank Algorithm in which an article’s authenticity can be determined by the number of authentic nodes in a news network that reference the article or contain similar information, adding to a node’s authority score.
These ideas directly build off our in-class discussions as we have spent significant time discussing the implications of shape within social networks. If we represent nodes as people and edges as article shares or similar information, then we can get information about which articles are far more credible than others and we can get some information about whether an article is partisan or has a fair distribution of sharers. Either way, algorithms are current being developed to tackle this fake-news crisis that rely on the social network structure. That in itself is significant.