Browse Prior Art Database

A Method to Predict Contextually Viral Social Media Disclosure Number: IPCOM000246816D
Publication Date: 2016-Jul-04
Document File: 4 page(s) / 61K

Publishing Venue

The Prior Art Database


This disclosure proposes a method of predicting viral social media within the context of a given user. There are many systems that attempt to perform this task that concentrate on either (a) the content of the message, which may include multimedia analysis; and (b) social network analysis performed upon the user posting the content. This system differs itself by not assessing either of those two things and instead proposes an "Engagement Rate" metric be employed to measure a user's normal level of engagement and then assessing a recent post against this metric.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 32% of the total text.

Page 01 of 4

A Method to Predict Contextually Viral Social Media

A given piece of social media is contextually popular if the level of engagement around that post is higher than normal. For example in the context of the Twitter system, a user might have a few hundred followers and if their tweeted content gets tens of retweets this is likely to be popular content within their context i.e. they've probably achieved higher engagement than they normally do. For a user with millions of followers the expectation would be that their content would always be popular but tweets with a higher level of engagement than normal would be popular

within their context. The system can be translated to other social networks by measuring whatever form of engagement the network provides, such as Likes and Comments in the Facebook system.

    This disclosure proposes a way to measure contextual popularity, termed the Engagement Rate. This Engagement Rate can then be further employed to assess materials posted by a user to determine if a piece of content is potentially more interesting than usual. The key elements here are that the content of the post or social network of the user are not taken into account and the metric can be quickly calculated and used to highlight popular content for any user no matter how many social links they have in their network.

    Essentially, this disclosure is summarised as: measure the normal levels of interaction a user gets with their posts then assess whether a new post they've made is above that level and if so mark the post as having potential to become contextually popular.

    This disclosure concentrates on an example implementation for the Twitter system. However, the system could be converted to other social networks.

    The rate of retweets for historic tweets of a given user can be used to calculate their Engagement Rate. This metric can be used to assess new content against their usual rate of interaction. Content that is being interacted with more quickly than normal is flagged as having the potential to become viral within the context of the given user. That is, for user's with millions of followers the content

would probably become genuinely viral but for user's with smaller number of followers the content would be viral within their own, much smaller, sphere of influence and connectivity.


    In summary English pseudo-style code, the process by which Engagement Rate can be calculate and used is as follows:

  1. Search for content in the required domain e.g. the US Masters gold tournament

2. For each new root Tweet (ignore retweets and modified tweets), either
2.1. Retrieve the user's Engagement Rate from cache

      2.2. Calculate and cache the Engagement Rate for the user 2.2.1. Set a "null" rate for users with few followers or users with little history from which to calculate an Engagement Rate and skip 2.2.3.

      2.2.2. Search the user's history to calculate their rate, for each root tweet Note the date and time of the root t...