UNDP Project Manager
Assistant Professor at the Faculty of Computer Science and Engineering
18 Jun 2014
As the recent devastation wrought by flooding in Serbia and Bosnia and Herzegovina has shown, social media has a vital role to play in spreading information about natural and human-made disasters.
Twitter is a great example of a platform that can deliver information to a vast number of people at rapid speed. So Igor Miskovski and I decided to run an initial analysis of tweets in the country from 2011–2013 to find out what percentage of tweets were related to Disaster Risk Reduction (or what we call DRR in UN-speak).
There are estimated to be some 20,000 – 25,000 Twitter users in the country, of which approximately 15,000 can be identified through a number of online applications.
Analysis of DRR-related tweets provided a very interesting insight: Macedonian Twitter users tweet most on Mondays and least on Saturdays, suggesting that tweeting is an activity that people consider part of their working week.
Twitter activity increases during work hours from 08:00 till 16:00, reaching saturation from 16:00 till 18:00. After 18:00 it starts to grow again, reaching a peak at 23:00. But even Twitter users need some rest: from 00:00 to 08:00 there is an exponential drop.
Most Twitter users are in the big city areas, with Skopje on the top. It is interesting that there are active users in Turkey (Istanbul, Bursa, Izmir, Marmaris areas), as well as in the USA (Cleveland/Detroit areas).
The latter points to an interesting correlation, according to US Census figures, a large cluster of Macedonian Americans live in the Midwest, particularly Detroit, which has a population of roughly 10,000 Macedonian Americans.
The hazardous events most mentioned on Twitter during the period of analysis were as follows:
- Earthquakes – 495 tweets
- Fires – 132
- Air pollution/fog – 122
- Floods – 104
- Traffic accidents – 73
- Diseases/virus infections – 59
As well as identifying the weekly/hourly distribution of DRR-related tweets, we also identified the most used hashtags, the most active users, the most active DRR users, as well as users that tweeted about different kinds of hazardous events.
A number of challenges still remain: geo-tagging of tweets (or lack of it), the use of different languages (e.g. Macedonian, Albanian, Turkish, English) and alphabets (e.g. “fire” could be tweeted as: #fire, #pozzar, #пожар or another word in the languages of the ethnic minorities living in the country), the low awareness of the Twitter community about DRR-related issues and hazards, as well as the low level of use of crowdsourcing.
We are currently developing a very simple web-based application that will enable easier and more precise analysis of tweets which will include the number of DRR-related tweets during a certain time period, the first DRR-related tweets relating to hazardous events, and the mapping of DRR tweets and related events.
It’s an important time to raise awareness through Twitter about DRR activities. Developing this technology and its accessibility could have great implications for the people who are most affected when disaster does strike By empowering them through social media, they can have the opportunity to report a disaster and get real-time information on how to stay safe.
Special thanks go out to Patrick Meier and Ingmar Weber for their inspiration, support, and guidance during this exercise.
We’d love to hear from you, so join the Twitter 4 #DRR community & #UNDP4Future
Tweet us at:
@UNDPMK @PopovskiVasko and @IgorMiskovski