Developing a Framework for Validating Crowdsourced Data

By Angela Okune
iHub Research
  Published 05 Apr 2013
Share this Article
Developing a Framework for Validating Crowdsourced Data

Blog post by: Nanjira Sambuli

Is crowdsourced information (that is, information collected from citizens through online platforms such as Twitter, Facebook, and text messaging) more representative of on-the-ground reality than traditional media reporting like television, radio and newspapers? If so, can crowdsourced information be organized and used to find and capture news as generated? How can such information be confirmed to match events reported to have happened in a particular location?

These are some of the guiding questions to a ground-breaking research project we are conducting, using the recently concluded Kenyan General Elections as a case study. The study, funded by the International Development Research Centre, runs to July 2013.

Political events in Kenya have been noted to spark many online conversations, especially with the continued uptake of social media. Opinions, facts, rumors, and events have been shared and reported online with increased frequency. We have been tracking such social media activity, with a particular focus on Twitter (due to ease of analysis), and assessing what information has been generated in the build up to, during, and after the March 4th voting process. With approximately 2.1 million tweets aggregated thus far, suffice to say there’s a wealth of data/information to be explored!

Some of the things we are exploring using this dataset includes:

  • whether there are events that were first reported on social media by ‘ordinary users’ before being reported on traditional media, or before being shared by news outlets’ official social media channels;

  • how many unique reports were generated per event, how such reports might be confirmed to be true/false (whether accompanied by photo or video evidence);

  • how many reports were generated online from different parts of the country (is social media activity limited to major towns and cities?);

  • is possible to create an automated way of verifying information shared on social media.

Stay posted for more information on our findings and upcoming activities soon!

comments powered by Disqus