For a given tweet, I need to determine the probability of it going viral. I have given this a thought and have considered various attributes like the sentiment, number of followers, time of post, length of the post, number of hashtags. But I am not sure on how to proceed.

Is there a given framework, already prepared models that could be used to determine the virality of the given feed? I would also like to extend this idea to Facebook feeds and other blog feeds.

  • 2
    $\begingroup$ There are indeed many frameworks for virality and diffusion that include Twitter but are not limited to that. Among the best papers analyzing tweets is Watts and Hofman The Structural Virality of Online Diffusion, unfortunately it doesn't appear to be ungated. Then there's Watts and Goldstein's The Structure of Online Diffusion Networks which is ungated (dangoldstein.com/papers/…). Finally, Lana Adamic's video discusses a framework for thinking about FB cascades that is generalizable ... youtube.com/watch?v=bA0Gin1t0m0 $\endgroup$ Jun 6, 2018 at 12:17
  • $\begingroup$ @DJohnson Could you point me to any framework? Like Google's tensorflow already has a pre-trained model for image recognition. Is there one for virality too? Or is there any framework using which I could build on? $\endgroup$
    – Den Hall
    Jun 6, 2018 at 13:02
  • 1
    $\begingroup$ Framework? By that are you referring to something like point and click software or an R module? Or are you referring to behavioral, mechanical, scientific and/or statistical assumptions underlying/driving viral diffusion processes? $\endgroup$ Jun 6, 2018 at 19:50
  • $\begingroup$ @DJohnson Some R module $\endgroup$
    – Den Hall
    Jun 7, 2018 at 5:13

1 Answer 1


There is some interesting meta research on this topic in regards to high impact academic papers. Schilling and Green (2011) came up with a few models describing what makes a paper high impact. For a direct look at tweets, there is the Jenders et al. (2013) paper that uses machine learning to predict viral tweets. I believe they used a naive bayes model.

Those two papers, and looking at what work has cited them, should give you a really good start on the modeling techniques used to look at high impact pieces of literature/media.

Jenders, M., Kasneci, G., & Naumann, F. (2013, May). Analyzing and predicting viral tweets. In Proceedings of the 22nd international conference on World Wide Web (pp. 657-664). ACM.

Schilling, M. A., & Green, E. (2011). Recombinant search and breakthrough idea generation: An analysis of high impact papers in the social sciences. Research Policy, 40(10), 1321-1331.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.