Published on June 10th, 2014 | by EJC0
Featured Tool: Dataminr
In a nutshell:
Initially targeted towards financial professionals, for whom up-to-the-minute knowledge of market developments is crucial, Dataminr has quickly proved itself useful for journalists and newsrooms.
Created in 2009 by Ted Bailey, Jeff Kinsey and Sam Hendel, the app uses an algorithm to mine Twitter for information tailored to the needs of users. It then forwards this customised information in the form of an alert stream to email addresses, in-house warning systems, instant messaging or as pop-ups – in real time. These live notifications of events have the potential to save the lives of journalists or citizens in volatile situations, and to provide newsrooms with a head start in investigations.
Dataminr users in the media create a unique profile and portfolio, indicating their specific information needs. For example, newsrooms may specify their topics of interest while field reporters may also determine their current regions. Dataminr then uses a proprietary pattern-recognition algorithm that analyses tweets and “delivers the earliest warning for breaking news, real-world events, off-the-radar content and emerging trends”. To break it down: the algorithm scores, tags and classifies millions of daily tweets into proper categories (like User Reputation, Information Value, Topic Density or Linguistic Signatures). Tweets are then clustered before the algorithm filters information through detection and evaluation criteria in order to produce a customised alert for the user’s distinct portfolio (Dataminr.com).
Breaking news coverage and reporters in the field:
In light of the app’s tremendous success for the financial sector, Dataminr stated that in late 2014 the algorithm will be used by the News sector in a collaboration effort with CNN. Again, the idea here is that millions of tweets will be filtered and narrowed down to those useful for journalists requesting specific topics or regions of focus.
Dataminr also offers an analysis of the significance and relevance of tweets for each individual portfolio, without any human analysis. When disasters and catastrophes strike, they usually result in hundreds or thousands of immediate tweets. For example, when Dataminr notified of an explosion in Mansoura, Egypt, in December 2013 at 6:11 PM, the incident was reported by major outlets at around 7:15 PM, with a remarkable time difference of a little over an hour. Dataminr thereby becomes a kind of fast, digital newswire, as newsrooms receive automatic alerts of events in real-time through the Dataminr enterprise application, email, instant messaging, pop-ups or their integrated internal systems.
These features come with great potential to enable journalists to begin investigating a story while it is practically still in its birth phase. Journalists reporting in crisis situations or in an unstable location who have entered their geographically specialised portfolio, would receive analytical updates of on-the-ground events as they develop. In the long run, this could have live-saving potential.
The crux of Dataminr’s contribution to crisis reporting, and what sets it apart from apps with similar filtering potential, is its contribution to verification. Identifying the source behind an event is imperative, given the enormous amount of tweets that could be sent about it and the countless examples of errors made in the past.
The app not only delivers information on the spot, but includes a thorough analysis based on a user’s preferences. Through its analytical module, Dataminr can determine the source behind an event, enabling journalists to “replay how that event broke to identify potential sources that were on the scene”, explains Dataminr’s Dan Bailey. Here, Dataminr can be particularly useful for the human aspect of the verification process.
What’s the catch?
On a more worrisome note, Subbaraman notes that a reaction on Twitter to any event naturally requires the local presence of a tweeter, yet digital divides and penetrations of Twitter are not consistent throughout the world. Other technologies, for instance, may provide more effective warning alerts than Dataminr in terms of regionally available technology. Although this might pose restrictions for the potential of Dataminr to inform journalists and newsrooms, increasing technological access in remote areas suggests that this may only be a minor problem for the future of the app.
About the Author: