Social media is increasingly used to communicate during major crises. But what about small-scale incidents such as a car crash or fire? These “micro-crises” typically generate a far smaller volume of social media activity during a much shorter period and more bounded geographical area. Detecting these small-scale events thus poses an important challenge for the field of Crisis Computing.
Axel Schulz just published co-authored a paper on this exact challenge. In this study, he and co-authors Petar Ristoski & Heiko Paulheim “present a solution for a real-time identification of small scale incidents using microblogs,” which uses machine learning—combining text classication and semantic enrichment of microblogs—to increase situational awareness. The study draws on 7.5 million tweets posted in the city centers of Seattle and Memphis during November & December 2012 and February 2013. The authors used the “Seattle Real Time Fire 911 Calls” dataset to identify relevant keywords in the collected tweets. They also used WordNet to “extend this set by adding the direct hyponyms. For instance, the keyword “accident” was extended with ‘collision’, ‘crash’, ‘wreck’, ‘injury’, ‘fatal accident’, and ‘casualty’.”
An evaluation of this combined “text classication” and “semantic enrichment” approach shows that small scale incidents can be identified with an accuracy 89%. A copy of Axel et al.‘s paper is available here (PDF). This is a remarkable level of accuracy given the rare and micro-level nature of the incidents studied.
Twitter seems to be here to stay and is proving to have an increasingly valuable role in emergency management, at whatever scale. It would therefore seem worth considering the creation of an equivalent to Channel 16 (the emergency maritime VHF calling frequency) for Twitter or a similar service. It would need technical, organisational and legal effort and agreements on a global level, but has the potential to save lives. A UNISDR led initiative perhaps?
Hi John, thanks for reading and for your comment, very interesting suggestion. Maybe FEMA would be a good start, ie, focus first on local/national level as test-bed, then go international?
Hi Patrick, Agreed, FEMA would be a fine starting point. They have the right reach as an organisation and potentially the technical and legal framework to get started. http://www.emergencyaus.info is also in the right direction, but I was thinking something even simpler.
Thanks John, what exactly do you have in mind re “something even simpler”?
Hi Patrick, I’ve looked apps such the Red Panic Button and One Touch Location and emergencyaus as mentioned above. However, your post got me thinking about the monitoring and responding side of the equation. Maritime Channel 16 and the Digital Selective Calling capabilities came to mind. The Wikipedia entries for Ch16 and DSC will give you a flavour of where I’m thinking (e.g. designators = common hashtags, IMMI = IMEI, GPS loc = GPS loc, etc). The post got me thinking about trying to bring together the best concepts from both approaches, with a bit of content classification thrown in. The trick would be to keep it a simple as possible for the distressed user but as rich as possible, with system derived detail, for the emergency responders.
Rather than fill up this post with early ideas please drop me an e-mail contact via my e-mail or a DM on Twitter (@johnbmccubbin) and I’ll try to lay out my thoughts in a more structured document. Perhaps there is a joint blog post in it.
Will do, thanks John.
Pingback: Accident tweets | CiFa