paint-brush
The Outbreak: How to Detect The Real Viral Posts Compared to The One Hour Share Spikeby@baditaflorin

The Outbreak: How to Detect The Real Viral Posts Compared to The One Hour Share Spike

by Florin BaditaJanuary 13th, 2020
Read on Terminal Reader
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

To get more information's about what is the Outbreak tool, read this post.

People Mentioned

Mention Thumbnail

Companies Mentioned

Mention Thumbnail
Mention Thumbnail
featured image - The Outbreak: How to Detect The Real Viral Posts Compared to The One Hour Share Spike
Florin Badita HackerNoon profile picture

To get more information's about what is the Outbreak tool, read this post.

Here i will discuss just how we can improve the algorithm, by analyzing the detection of an article as viral multiple times, for the same permanent_link in multiple Fb Pages or within the same page.

As a study-case i will select this image post.

The algorithm first detected this post when it had 1400 shares, and was growing at as a speed of 760 shares per hour.

This is not a fake article. But is not true also.
Welcome to Logical Fallacies. This specific post is a logical fallacy, called Argument from ignorance.


I will let Wikipedia do the explanation:

Argument from ignorance (from Latinargumentum ad ignorantiam), also known as appeal to ignorance (in which ignorance represents “a lack of contrary evidence”), is a fallacy in informal logic. It asserts that a proposition is true because it has not yet been proven false (or vice versa).

This represents a type of false dichotomy in that it excludes a third option, which is that: there may have been an insufficient investigation, and therefore there is insufficient information to prove the proposition be either true or false. Nor does it allow the admission that the choices may in fact not be two (true or false), but may be as many as four,

  • true
  • false
  • unknown between true or false
  • being unknowable (among the first three).[1]

In debates, appeals to ignorance are sometimes used in an attempt to shift the burden of proof.

Also, if we look at the emotional fingerprint of the post, we can see a big difference compared with a fake news article.

OK, now that we had finished explaining why this is a shady post, let`s look at the data. The algorithm detected the post 4 times in the last 24 hours, the first time when it had 1432 shares, then next hour when it was at 2142 shares, 2 hours after when it rose to 2875 shares and final time when it was at 3592 shares.

This allows us to analyze recurring viral identified articles, and elevate them to a higher viral threat configuration. I`m doing a 5 levels classification of virality, but for the sake of simplicity, let`s explain it using the count method.

We can add a column that does the count, if this article is declared viral, increase the count to 1.

If we detect the article again, in the future hours, we increase the count to 2.

Journalists can set a threshold and get in their feed only the articles that were detected 2 times as viral, so that they have less articles to look into.

I don`t have a front End to the app, so the only way to output the results is via this google spreadsheet. The first results are the oldest, scroll to the end for the most recent future possible future fake news.

In the list i have a list of grey listed and legitimate websites.

I have added some white listed Fb pages, and you will see posts from that pages because i want to learn the difference between viral news that are legitimate compared with the ones that are fake or misleading.

Also, i`m scanning some politicians fb pages, because they post news that can be related, and they can be the first to break a news.

Help support this effort: https://www.youcaring.com/newspapersandjournalists-711969

Sorry for the language mistakes in the article, English is not my first language.

About Me

In the last 3 years i`m a collaborator with the Organised Crime and Corruption Reporting Projects (OCCRP), were i do data analysis and pattern recognition to uncover patterns of corruption in unstructured datasets.

In September 2016 i have moved to San Francisco, to start a new life here.

You can find me online on Hackernoon, Medium Florin Badita, AngelListTwitter , Linkedin, Openstreetmap, Github, Quora, Facebook