Post-Disaster Situation Analysis and Resource Management using Delay-Tolerant Peer-to-Peer Wireless Networks

A Project Proposal

Using online social media for gathering situational updates

Problem description

Several studies have found that these online media play vital roles in the aftermath of natural or man-made calamities (see references below). A recent study [Gupta PSOSM 2012] showed that about 30% of the tweets posted immediately after calamities such as UK riots, Hurricane Irene, Mumbai blasts, contained important situational information. However, only 17% tweets contained credible situational information [Gupta PSOSM 2012]. This shows the necessity of developing methodologies to estimate the trustworthiness of tweets in post-disaster scenarios.

One possible approach is to consider a tweet trustworthy if it is posted or retweeted by a known authoritative user (e.g., a well-known expert or government agency related to tsunamis). However, known authoritative users may not be present in the region where a calamity has occurred, or may not be tweeting actively during the calamity. Hence relying only on authoritative users can cause intolerable delay in obtaining situational information. So, tweets posted by the common users located in the region of the calamity need to be utilized as well.

Research Challenges

Building up a large database of authoritative Twitter users related to different types of calamities.
Prior work: Developed a search system for topical experts in Twitter, named Cognos, which can be used to identify experts on specific topics such as tsunami and hurricane [Ghosh SIGIR 2012].
Future plans: Identifying the expert-networks in Twitter on specific topics related to calamities, which can help in quickly knowing the collective opinion of the community of experts during a calamity, ...
Developing methodologies to estimate the trustworthiness of a tweet posted by a common user.
Prior work: Developed a novel who-is-who system that can be used to know the credentials of individual Twitter users [Sharma WOSN 2012]. The methodology utilizes social annotations, i.e., how the social connections of a user describe her.
Future plans: Utilize the features of a tweet to estimate its usefulness / trustworthiness, e.g., using the geo-location of the tweet to check whether it has been posted from the region where the calamity has occurred, or whether a tweet (posted by a common user) has been retweeted by a known authority, ...

Sample tweets posted after calamities

We are collecting tweets posted immediately after certain calamities in the recent year. Here we give some sample tweets from the data collected by us.

UK Riots (August 2011)

Tweet text	Type of tweet
Riot police called to Tottenham, London - patrol cars attacked. Friend of man shot on Thurs warned of unrest to #c4news: http://t.co/8Ajj6ph	situational update
Uh oh. That's no good. RT @***: BREAKING #London #Riot Hooded teenagers surround BBC crew in Tottenham. Transmission ended	situational update
The sad truth behind London riot: http://t.co/jpnskUe	public opinion
Oh my god! This can't be happening at London Eye! #Londonriots #Londonriot #Prayforlondon http://twitpic.com/6372vo	rumor

Hurricane Irene (August 2011)

Tweet text	Type of tweet
Hurricane Irene to approach coast of Carolinas Saturday morning as major hurricane: National Hurricane Center director - @Reuters	warning
BREAKING:: New York City mayor says those who need to evacuate should do it now. -EF #hurricane #irene	Government declaration
Hurricane #Irene is now Tropical Storm Irene, downgraded as maximum sustained winds fall to 65 mph	situational update
Tiny library has its children section destroyed by Hurricane Irene, could use some help: http://t.co/l5w1oIL	Request for aid
President Obama to deliver live statement on Hurricane Irene from the Rose Garden - NBC News	Government declaration

Earthquake and tsunami in Japan (March 2011)

Tweet text	Type of tweet
8.8 Magnitude Earthquake in Japan with a possible Tsunami soon. Sending my prayers to everyone in Japan. #GodBlessJapan	warning
Ship carrying 100 people swept away by #japan tsunami #quake	situational update
Hawaii orders evacuation of low-lying coastal areas after massive #earthquake off #Japan triggered #tsunami alert	Government declaration
#SavetheChildren standing by to respond to children's needs in #Japan #tsunami http://ow.ly/4civi	post-disaster aid

Relevant publications / services developed

[Ghosh SIGIR 2012] S. Ghosh, N. Sharma, F. Benevenuto, N. Ganguly, K. Gummadi, Cognos: Crowdsourcing Search for Topic Experts in Microblogs, ACM SIGIR International Conference, 2012.

[Sharma WOSN 2012] N. Sharma, S. Ghosh, F. Benevenuto, N. Ganguly, K. Gummadi, Inferring Who-is-Who in the Twitter Social Network, Workshop on Online Social Networks (with ACM SIGCOMM), 2012. Highest rated paper in the workshop, selected for fast-track publication in ACM Computer Communication Review.

Services developed:

Cognos, a search system for topic experts in Twitter.

A who-is-who system that can be used to know the credentials of individual Twitter users.