Finding Fires with Twitter

Finding Fires with Twitter Robert Power robert.power@csiro.au Bella Robinson bella.robinson@csiro.au CSIRO Computational Informatics G.P.O. Box 664 Canberra, ACT 2601, Australia David Ratcliffe david.ratcliffe@csiro.au Abstract This paper presents a notification system to identify in near-real-time Tweets de- scribing fire events in Australia. The sys- tem identifies fire related ‘alert words’ published on Twitter which are further processed by a classifier to determine if they correspond to an actual fire event. We describe how the classifier has been estab- lished and report preliminary results. The original notification system did not in- clude a classifier and could not discrimi- nate between messages unrelated to ‘real’ fire events. In the first three months of operation, the system generated 42 ‘fire’ email notifications of which 20 related to actual fires and 12 of those contained Tweets that may have been of interest to fire fighting agencies. If the classifier had been used, 21 emails would have been is- sued: an improvement in accuracy from 48% to 78%. However, the recall score reduced from 1 to 0.8 which is not desir- able for this particular task. We propose extensions to address this short coming. 1 Introduction In Australia bushfire management is a state and territory government responsibility and each juris- diction has its own agency which takes the lead in coordinating community preparedness and re- sponding to bushfires when they occur. For ex- ample, the Rural Fire Services (RFS) in NSW, the Country Fire Authority (CFA) in Victoria and so on, are each responsible for firefighting activities, training to prepare communities to protect them- selves, land management hazard reduction as well as situations involving search and rescue. During the Australian disaster season, early Oc- tober through to the end of March, these fire agen- cies continuously monitor weather conditions in preparation for responding to events when they oc- cur. They also inform the community about known incidents, see for example the NSW RFS Current Fires and Incidents page1. These agencies publish incident information on social media sites such as Facebook and Twitter. This provides a new channel of communication to interact with the community to both provide infor- mation about known events and to receive crowd- sourced content from the general public. This engagement of social media is yet to be fully utilised. During crisis events, the emergency services effectively use social media to provide in- formation to the community, but their ability to ob- tain information from the public is limited (Lind- say, 2011). While there are social media suc- cess stories, for example the Queensland Police Force during the Brisbane Floods in 2011 (Charl- ton, 2012), they are not yet widespread. Our aim is to develop an emergency manage- ment tool that sources information from social me- dia in near-real-time. The challenges are many: the most significant being how to reliably extract relevant information about emergency events of interest for crisis coordinators. The test case de- scribed in this paper is to extract current informa- tion published on Twitter about actual fire events. The …

Content truncated to 500 words