{"id":14164,"date":"2019-02-06T23:46:52","date_gmt":"2019-02-06T23:46:52","guid":{"rendered":"https:\/\/rsoftsolution.com\/blog\/fabula-ai-is-using-social-spread-to-spot-fake-news.html"},"modified":"2019-02-06T23:46:52","modified_gmt":"2019-02-06T23:46:52","slug":"fabula-ai-is-using-social-spread-to-spot-fake-news","status":"publish","type":"post","link":"https:\/\/www.rbsitsoftwaresolution.in\/blog\/fabula-ai-is-using-social-spread-to-spot-fake-news.html","title":{"rendered":"Fabula AI is using social spread to spot \u2018fake news\u2019"},"content":{"rendered":"

UK startup Fabula AI<\/a> reckons it\u2019s devised a way for artificial intelligence to help user generated content platforms get on top of the disinformation crisis<\/a> that keeps rocking the world of social media with antisocial scandals<\/a>.<\/p>\n

Even Facebook\u2019s <\/a> Mark Zuckerberg <\/a> has sounded a cautious note about AI technology\u2019s capability to meet the complex, contextual, messy and inherently human challenge of correctly understanding every missive a social media user might send, well-intentioned or its nasty flip-side.<\/p>\n

\u201cIt will take many years to fully develop these systems,\u201d the Facebook founder wrote two years ago<\/a>, in an open letter discussing the scale of the challenge of moderating content on platforms thick with billions of users. \u201cThis is technically difficult as it requires building AI that can read and understand news.\u201d<\/p>\n

But what if AI doesn\u2019t need to read and understand news in order to detect whether it\u2019s true or false?<\/p>\n

Step forward Fabula, which has patented<\/a> what it dubs a \u201cnew class\u201d of machine learning algorithms to detect \u201cfake news\u201d \u2014 in the emergent field of \u201cGeometric Deep Learning\u201d; where the datasets to be studied are so large and complex that traditional machine learning techniques struggle to find purchase on this \u2018non-Euclidean\u2019 space.<\/p>\n

The startup says its deep learning algorithms are, by contrast, capable of learning patterns on complex, distributed data sets like social networks. So it\u2019s billing its technology as a breakthrough. (Its written a paper on the approach which can be downloaded here<\/a>.)<\/p>\n

It is, rather unfortunately, using the populist and now frowned upon badge \u201cfake news\u201d in its PR. But it says it\u2019s intending this fuzzy umbrella to refer to both disinformation and misinformation. Which means maliciously minded and<\/em> unintentional fakes. Or, to put it another way, a photoshopped fake photo or a genuine image spread in the wrong context.<\/p>\n

The approach it\u2019s taking to detecting disinformation relies not on algorithms parsing news content to try to identify malicious nonsense but instead looks at how such stuff spreads on social networks \u2014 and also therefore who is spreading it.<\/p>\n

There are characteristic patterns to how \u2018fake news\u2019 spreads vs the genuine article,\u00a0says Fabula co-founder and chief scientist, Michael Bronstein.<\/p>\n

\u201cWe look at the way that the news spreads on the social network. And there is \u2014 I would say \u2014 a mounting amount of evidence that shows that fake news and real news spread differently,\u201d he tells TechCrunch, pointing to a recent major study by MIT academics<\/a> which found \u2018fake news\u2019 spreads differently vs bona fide content on Twitter.<\/p>\n

\u201cThe essence of geometric deep learning is it can work with network-structured data. So here we can incorporate heterogenous data such as user characteristics; the social network interactions between users; the spread of the news itself; so many features that otherwise would be impossible to deal with under machine learning techniques,\u201d he continues.<\/p>\n

Bronstein, who is also a professor at Imperial College London, <\/a> with a chair in machine learning and pattern recognition, likens the phenomenon Fabula\u2019s machine learning classifier has learnt to spot to the way infectious disease spreads through a population.<\/p>\n

\u201cThis is of course a very simplified model of how a disease spreads on the network. In this case network models relations or interactions between people. So in a sense you can think of news in this way,\u201d he suggests. \u201cThere is evidence of polarization, there is evidence of confirmation bias. So, basically, there are what is called echo chambers that are formed in a social network that favor these behaviours.\u201d<\/p>\n

\u201cWe didn\u2019t really go into \u2014 let\u2019s say \u2014 the sociological or the psychological factors that probably explain why this happens. But there is some research that shows that fake news is akin to epidemics.\u201d<\/p>\n

The tl;dr of the MIT study<\/a>, which examined a decade\u2019s worth of tweets, was that not only does the truth spread slower but also that human beings themselves are implicated in accelerating disinformation. (So, yes, actual human beings are the problem.) Ergo, it\u2019s not all bots doing all the heavy lifting of amplifying junk online.<\/p>\n

The silver lining of what appears to be an unfortunate quirk of human nature is that a penchant for spreading nonsense may ultimately help give the stuff away \u2014 making a scalable AI-based tool for detecting \u2018BS\u2019 potentially not such a crazy pipe-dream.<\/p>\n

Although, to be clear, Fabula\u2019s AI remains in development at this stage, having been tested internally on Twitter data sub-sets at this stage. And the claims it\u2019s making for its prototype model remain to be commercially tested with customers in the wild using the tech across different social platforms.<\/p>\n

It\u2019s hoping to get there this year, though, and intends to offer an API for platforms and publishers towards the end of this year. The AI classifier is intended to run in near real-time on a social network or other content platform, identifying BS.<\/p>\n

Fabula envisages its own role, as the company behind the tech, as that of an open, decentralised \u201ctruth-risk scoring platform\u201d \u2014 akin to a credit referencing agency just related to content, not cash.<\/p>\n

Scoring comes into it because the AI generates a score for classifying content based on how confident it is it\u2019s looking at a piece of fake vs true news.<\/p>\n

\"\"<\/p>\n

A visualisation of a fake vs real news distribution pattern; users who predominantly share fake news are coloured red and users who don\u2019t share fake news at all are coloured blue \u2014 which Fabula says shows the clear separation into distinct groups, and \u201cthe immediately recognisable difference in spread pattern of dissemination\u201d.<\/p>\n

In its own tests Fabula says its algorithms were able to identify 93 percent of \u201cfake news\u201d within hours of dissemination \u2014 which Bronstein claims is \u201csignificantly higher\u201d than any other published method for detecting \u2018fake news\u2019. (Their accuracy figure uses a standard aggregate measurement of machine learning classification model performance, called ROC AUC.)<\/p>\n

The dataset the team used to train their model is a subset of Twitter\u2019s network \u2014 comprised of around 250,000 users and containing around 2.5 million \u201cedges\u201d (aka social connections).<\/p>\n

For their training dataset Fabula relied on true\/fake labels attached to news stories by third party fact checking NGOs, including Snopes and PolitiFact. And, overall, pulling together the dataset was a process of \u201cmany months\u201d, according to Bronstein, He also says that around a thousand different stories were used to train the model, adding that the team is confident the approach works on small social networks, as well as Facebook-sized mega-nets.<\/p>\n

Asked whether he\u2019s sure the model hasn\u2019t been trained to identified patterns caused by bot-based junk news spreaders, he says the training dataset included some registered (and thus verified \u2018true\u2019) users.<\/p>\n

\u201cThere is multiple research that shows that bots didn\u2019t play a significant amount [of a role in spreading fake news] because the amount of it was just a few percent. And bots can be quite easily detected,\u201d he also suggests, adding: \u201cUsually it\u2019s based on some connectivity analysis or content analysis. With our methods we can also detect bots easily.\u201d<\/p>\n

To further check the model, the team tested its performance over time by training it on historical data and then using a different split of test data.<\/p>\n

\u201cWhile we see some drop in performance it is not dramatic. So the model ages well, basically. Up to something like a year the model can still be applied without any re-training,\u201d he notes, while also saying that, when applied in practice, the model would be continually updated as it keeps digesting (ingesting?) new stories and social media content.<\/p>\n

Somewhat terrifyingly, the model could also be used to predict virality, according to Bronstein \u2014 raising the dystopian prospect of the API being used for the opposite purpose to that which it\u2019s intended: i.e. maliciously, by fake news purveyors, to further amp up their (anti)social spread.<\/p>\n

\u201cPotentially putting it into evil hands it might do harm,\u201d Bronstein concedes. Though he takes a philosophical view on the hyper-powerful double-edged sword of AI technology, arguing such technologies will create an imperative for a rethinking of the news ecosystem by all stakeholders, as well as encouraging emphasis on user education and teaching critical thinking.<\/p>\n

Let\u2019s certainly hope so. And, on the educational front, Fabula is hoping its technology can play an important role \u2014 by spotlighting network-based cause and effect.<\/p>\n

\u201cPeople now like or retweet or basically spread information without thinking too much or the potential harm or damage they\u2019re doing to everyone,\u201d says Bronstein, pointing again to the infectious diseases analogy. \u201cIt\u2019s like not vaccinating yourself or your children. If you think a little bit about what you\u2019re spreading on a social network you might prevent an epidemic.\u201d<\/p>\n

So, tl;dr, think before you RT.<\/p>\n

Returning to the accuracy rate of Fabula\u2019s model, while ~93 per cent might sound pretty impressive, if it were applied to content on a massive social network like Facebook \u2014 which has some 2.3BN+ users, uploading what could be trillions of pieces of content daily \u2014 even a seven percent failure rate would still make for an awful lot of fakes slipping undetected through the AI\u2019s net.<\/p>\n

But Bronstein says the technology does not have to be used as a standalone moderation system. Rather he suggests it could be used in conjunction with other approaches such as content analysis, and thus function as another string on a wider \u2018BS detector\u2019s bow.<\/p>\n

It could also, he suggests, further aid human content reviewers \u2014 to point them to potentially problematic content more quickly.<\/p>\n

Depending on how the technology gets used he says it could do away with the need for independent third party fact-checking organizations altogether because the deep learning system can be adapted to different use cases.<\/p>\n

Example use-cases he mentions include an entirely automated filter (i.e. with no human reviewer in the loop); or to power a content credibility ranking system that can down-weight dubious stories or even block them entirely; or for intermediate content screening to flag potential fake news for human attention.<\/p>\n

Each of those scenarios would likely entail a different truth-risk confidence score. Though most \u2014 if not all \u2014 would still require some human back-up. If only to manage overarching ethical and legal considerations related to largely automated decisions. (Europe\u2019s GDPR<\/a> framework has some requirements on that front, for example.)<\/p>\n

Facebook\u2019s grave failures around moderating hate speech in Myanmar<\/a> \u2014 which led to its own platform becoming a megaphone for terrible ethnical violence<\/a> \u2014 were very clearly exacerbated by the fact it did not have enough reviewers who were able to understand (the many) local languages and dialects spoken in the country.<\/p>\n

So if Fabula\u2019s language-agnostic propagation and user focused approach proves to be as culturally universal as its makers hope, it might be able to raise flags faster than human brains which lack the necessary language skills and local knowledge to intelligently parse context.<\/p>\n

\u201cOf course we can incorporate content features but we don\u2019t have to \u2014 we don\u2019t want to,\u201d says Bronstein. \u201cThe method can be made language independent. So it doesn\u2019t matter whether the news are written in French, in English, in Italian. It is based on the way the news propagates on the network.\u201d<\/p>\n

Although he also concedes: \u201cWe have not done any geographic, localized studies.\u201d<\/p>\n

\u201cMost of the news that we take are from PolitiFact<\/a> so they somehow regard mainly the American political life but the Twitter users are global. So not all of them, for example, tweet in English. So we don\u2019t yet take into account tweet content itself or their comments in the tweet \u2014 we are looking at the propagation features and the user features,\u201d he continues.<\/p>\n

\u201cThese will be obviously next steps but we hypothesis that it\u2019s less language dependent. It might be somehow geographically varied. But these will be already second order details that might make the model more accurate. But, overall, currently we are not using any location-specific or geographic targeting for the model.<\/p>\n

\u201cBut it will be an interesting thing to explore. So this is one of the things we\u2019ll be looking into in the future.\u201d<\/p>\n

Fabula\u2019s approach being tied to the spread (and the spreaders) of fake news certainly means there\u2019s a raft of associated ethical considerations that any platform making use of its technology would need to be hyper sensitive to.<\/p>\n

For instance, if platforms could suddenly identify and label a sub-set of users as \u2018junk spreaders\u2019 the next obvious question is how will they treat such people?<\/p>\n

Would they penalize them with limits \u2014 or even a total block \u2014 on their power to socially share on the platform? And would that be ethical or fair given that not every sharer of fake news is maliciously intending to spread lies?<\/p>\n

What if it turns out there\u2019s a link between \u2014 let\u2019s say \u2014 a lack of education and propensity to spread disinformation? As there can be a link between poverty and education\u2026 What then? Aren\u2019t your savvy algorithmic content downweights risking exacerbating existing unfair societal divisions?<\/p>\n

Bronstein agrees there are major ethical questions ahead when it comes to how a \u2018fake news\u2019 classifier gets used.<\/p>\n

\u201cImagine that we find a strong correlation between the political affiliation of a user and this \u2018credibility\u2019 score. So for example we can tell with hyper-ability that if someone is a Trump supporter then he or she will be mainly spreading fake news. Of course such an algorithm would provide great accuracy but at least ethically it might be wrong,\u201d he says when we ask about ethics.<\/p>\n

He confirms Fabula is not using any kind of political affiliation information in its model at this point \u2014 but it\u2019s all too easy to imagine this sort of classifier being used to surface (and even exploit) such links.<\/p>\n

\u201cWhat is very important in these problems is not only to be right \u2014 so it\u2019s great of course that we\u2019re able to quantify fake news with this accuracy of ~90 percent \u2014 but it must also be for the right reasons,\u201d he adds.<\/p>\n

The London-based startup was founded in April last year, though the academic research underpinning the algorithms has been in train for the past four years, according to Bronstein.<\/p>\n

The patent for their method was filed in early 2016 and granted last July.<\/p>\n

They\u2019ve been funded by $500,000 in angel funding and about another $500,000 in total of European Research Council <\/a> grants plus academic grants from tech giants Amazon, <\/a> Google and Facebook, awarded via open research competition awards.<\/p>\n

(Bronstein confirms the three companies have no active involvement in the business. Though doubtless Fabula is hoping to turn them into customers for its API down the line. But he says he can\u2019t discuss any potential discussions it might be having with the platforms about using its tech.)<\/p>\n

Focusing on spotting patterns in how content spreads as a detection mechanism does have one major and obvious drawback \u2014 in that it only works after the fact of (some) fake content spread. So this approach could never entirely stop disinformation in its tracks.<\/p>\n

Though Fabula claims detection is possible within a relatively short time frame \u2014 of between two and 20 hours after content has been seeded onto a network.<\/p>\n

\u201cWhat we show is that this spread can be very short,\u201d he says. \u201cWe looked at up to 24 hours and we\u2019ve seen that just in a few hours\u2026 we can already make an accurate prediction. Basically it increases and slowly saturates. Let\u2019s say after four or five hours we\u2019re already about 90 per cent.\u201d<\/p>\n

\u201cWe never worked with anything that was lower than hours but we could look,\u201d he continues. \u201cIt really depends on the news. Some news does not spread that fast. Even the most groundbreaking news do not spread extremely fast. If you look at the percentage of the spread of the news in the first hours you get maybe just a small fraction. The spreading is usually triggered by some important nodes in the social network. Users with many followers, tweeting or retweeting. So there are some key bottlenecks in the network that make something viral or not.\u201d<\/p>\n

A network-based approach to content moderation could also serve to further enhance the power and dominance of already hugely powerful content platforms \u2014 by making the networks themselves core to social media regulation, i.e. if pattern-spotting algorithms rely on key network components (such as graph structure) to function.<\/p>\n

So you can certainly see why \u2014 even above a pressing business need \u2014 tech giants are at least interested in backing the academic research. Especially with politicians increasingly calling for online content platforms to be regulated like publishers.<\/p>\n

At the same time, there are \u2014 what look like \u2014 some big potential positives to analyzing spread, rather than content, for content moderation purposes.<\/p>\n

As noted above, the approach doesn\u2019t require training the algorithms on different languages and (seemingly) cultural contexts \u2014 setting it apart from content-based disinformation detection systems. So if it proves as robust as claimed it should be more scalable.<\/p>\n

Though, as Bronstein notes, the team have mostly used U.S. political news for training their initial classifier. So some cultural variations in how people spread and react to nonsense online at least remains a possibility.<\/p>\n

A more certain challenge is \u201cinterpretability\u201d \u2014 aka explaining what underlies the patterns the deep learning technology has identified via the spread of fake news.<\/p>\n

While algorithmic accountability is very often a challenge for AI technologies, Bronstein admits it\u2019s \u201cmore complicated\u201d for geometric deep learning.<\/p>\n

\u201cWe can potentially identify some features that are the most characteristic of fake vs true news,\u201d he suggests when asked whether some sort of \u2018formula\u2019 of fake news can be traced via the data, noting that while they haven\u2019t yet tried to do this they did observe \u201csome polarization\u201d.<\/p>\n

\u201cThere are basically two communities in the social network that communicate mainly within the community and rarely across the communities,\u201d he says.\u00a0\u201cBasically it is less likely that somebody who tweets a fake story will be retweeted by somebody who mostly tweets real stories. There is a manifestation of this polarization. It might be related to these theories of echo chambers and various biases that exist. Again we didn\u2019t dive into trying to explain it from a sociological point of view \u2014 but we observed it.\u201d<\/p>\n

So while, in recent years, there have been some academic efforts to debunk the notion that social media users are stuck inside filter bubble bouncing their own opinions back at them, Fabula\u2019s analysis of the landscape of social media opinions suggests they do exist \u2014 albeit, just not encasing every Internet user.<\/p>\n

Bronstein says the next steps for the startup is to scale its prototype to be able to deal with multiple requests so it can get the API to market in 2019 \u2014 and start charging publishers for a truth-risk\/reliability score for each piece of content they host.<\/p>\n

\u201cWe\u2019ll probably be providing some restricted access maybe with some commercial partners to test the API but eventually we would like to make it useable by multiple people from different businesses,\u201d says requests. \u201cPotentially also private users \u2014 journalists or social media platforms or advertisers. Basically we want to be\u2026 a clearing house for news.\u201d<\/p>\n

Article source: http:\/\/feedproxy.google.com\/~r\/techcrunch\/social\/~3\/EkTS6JcAoac\/<\/p>","protected":false},"excerpt":{"rendered":"

UK startup Fabula AI reckons it\u2019s devised a way for artificial intelligence to help user generated content platforms get on top of the disinformation crisis that keeps rocking the world <\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"jetpack_post_was_ever_published":false,"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[3],"tags":[10,7,14,15,6,12,11],"class_list":["post-14164","post","type-post","status-publish","format-standard","hentry","category-latest-update","tag-computer-news","tag-downloads","tag-microsoft-download-news","tag-networking-updates","tag-technology-update","tag-wordpress-plugin-download","tag-wordpress-theme-download"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t