Tag Archives: Foursquare

Analyzing Foursquare Check-Ins During Hurricane Sandy

In this new study “Extracting Diurnal Patterns of Real World Activity from Social Media” (PDF), authors Nir Grinberg, Mor Naaman, Blake Shaw and Gild Lotan analyze Fousquare check-in’s and tweets to capture real-world activities related to coffee, food, nightlife and shopping. Here’s what an average week looks like on Foursquare, for example (click to enlarge):

Foursquare Week

“When rare events at the scale of Hurricane Sandy happen, we expect them to leave an unquestionable mark on Social Media activity.” So the authors applied the same methods used to produce the above graph to visualize and understand changes in behavior during Hurricane Sandy as reflected on Foursquare and Twitter. The results are displayed below (click to enlarge).

Sandy Analysis

“Prior to the storm, activity is relatively normal with the exception of iMac release on 10/25. The big spikes in divergent activity in the two days right before the storm correspond with emergency preparations and the spike in nightlife activity follows the ‘celebrations’ pattern afterwards. In the category of Grocery shopping (top panel) the deviations on Foursqaure and Twitter overlap closely, while on Nightlife the Twitter activity lags after Foursquare. On October 29 and 30 shops were mostly closed in NYC and we observe fewer checkins than usual, but interestingly more tweets about shopping. This finding suggests that opposing patterns of deviations may indicate of severe distress or abnormality, with the two platforms corroborating an alert.”

In sum, “the deviations in the case study of Hurricane Sandy clearly separate normal and abnormal times. In some cases the deviations on both platforms closely overlap, while in others some time lag (or even opposite trend) is evident. Moreover, during the height of the storm Foursquare activity diminishes significantly, while Twitter activity is on the rise. These findings have immediate implications for event detection systems, both in combining multiple sources of information and in using them to improving overall accuracy.”

Now if only this applied research could be transfered to operational use via a real-time dashboard, then this could actually make a difference for emergency responders and humanitarian organizations. See my recent post on the cognitive mismatch between computing research and social good needs.

bio

MatchApp: Next Generation Disaster Response App?

Disaster response apps have multiplied in recent years. I’ve been  reviewing the most promising ones and have found that many cater to  professional responders and organizations. While empowering paid professionals is a must, there has been little focus on empowering the real first responders, i.e., the disaster-affected communities themselves. To this end, there is always a dramatic mismatch in demand for responder services versus supply, which is why crises are brutal audits for humanitarian organizations. Take this Red Cross survey, which found that 74% of people who post a need on social media during a disaster expect a response within an hour. But paid responders cannot be everywhere at the same time during a disaster. The response needs to be decentralized and crowdsourced.

Screen Shot 2013-02-27 at 4.08.03 PM

In contrast to paid responders, the crowd is always there. And most survivals following a disaster are thanks to local volunteers and resources, not external aid or relief. This explains why FEMA Administrator Craig Fugate has called on the public to become a member of the team. Decentralization is probably the only way for emergency response organizations to improve their disaster audits. As many seasoned humanitarian colleagues of mine have noted over the years, the majority of needs that materialize during (and after) a disaster do not require the attention of paid disaster responders with an advanced degree in humanitarian relief and 10 years of experience in Haiti. We are not all affected in the same way when disaster strikes, and those less affected are often very motivated and capable at responding to the basic needs of those around them. After all, the real first responders are—and have always been—the local communities themselves, not the Search and Rescue Teams that parachutes in 36 hours later.

In other words, local self-organized action is a natural response to disasters. Facilitated by social capital, self-organized action can accelerate both response & recovery. A resilient community is therefore one with ample capacity for self-organization. To be sure, if a neighborhood can rapidly identify local needs and quickly match these with available resources, they’ll rebound more quickly than those areas with less capacity for self-organized action. The process is a bit like building a large jigsaw puzzle, with some pieces standing for needs and others for resources. Unlike an actual jigsaw puzzle, however, there can be hundreds of thousands of pieces and very limited time to put them together correctly.

This explains why I’ve long been calling for a check-in & match.com smartphone app for local collective disaster response. The talk I gave (above) at Where 2.0 in 2011 highlights this further as do the blog posts below.

Check-In’s with a Purpose: Applications for Disaster Response
http://iRevolution.net/2011/02/16/checkins-for-disaster-response

Maps, Activism & Technology: Check-In’s with a Purpose
http://iRevolution.net/2011/02/05/check-ins-with-a-purpose

Why Geo-Fencing Will Revolutionize Crisis Mapping
http://iRevolution.net/2011/08/21/geo-fencing-crisis-mapping

How to Crowdsource Crisis Response
http://iRevolution.net/2011/09/14/crowdsource-crisis-response

The Crowd is Always There
http://iRevolution.net/2010/08/14/crowd-is-always-there

Why Crowdsourcing and Crowdfeeding may be the Answer
http://iRevolution.net/2010/12/29/crowdsourcing-crowdfeeding

Towards a Match.com for Economic Resilience
http://iRevolution.net/2012/07/04/match-com-for-economic-resilience

This “MatchApp” could rapidly match hyper local needs with resources (material & informational) available locally or regionally. Check-in’s (think Foursquare) can provide an invaluable function during disasters. We’re all familiar with the command “In case of emergency break glass,” but what if: “In case of emergency, then check-in”? Checking-in is space- and time-dependent. By checking in, I announce that I am at a given location at a specific time with a certain need (red button). This means that information relevant to my location, time, user-profile (and even vital statistics) can be customized and automatically pushed to my MatchApp in real-time. After tapping on red, MatchApp prompts the user to select what specific need s/he has. (Yes, the icons I’m using are from the MDGs and just placeholders). Note that the App we’re building is for Androids, not iPhones, so the below is for demonstration purposes only.

Screen Shot 2013-02-27 at 3.32.29 PM

But MatchApp will also enable users who are less (or not) affected by a disaster to check-in and offer help (by tapping the green button). This is where the match-making algorithm comes to play. There are various (compatible options) in this respect. The first, and simplest, is to use a greedy algorithm. This  algorithm select the very first match available (which may not be the most optimal one in terms of location). A more sophisticated approach is to optimize for the best possible match (which is a non-trivial challenge in advanced computing). As I’m a big fan of Means of Exchange, which I have blogged about here, MatchApp would also enable the exchange of goods via bartering–a mobile eBay for mutual-help during disasters.

Screen Shot 2013-02-27 at 3.34.17 PM

Once a match is made, the two individuals in question receive an automated alert notifying them about the match. By default, both users’ identities and exact locations are kept confidential while they initiate contact via the app’s instant messaging (IM) feature. Each user can decide to reveal their identity/location at any time. The IM feature thus enables  users to confirm that the match is indeed correct and/or still current. It is then up to the user requesting help to share her or his location if they feel comfortable doing so. Once the match has been responded to, the user who received help is invited to rate the individual who offered help (and vice versa, just like the Uber app, depicted on the left below).

Screen Shot 2013-02-27 at 3.49.04 PM

As a next generation disaster response app, MatchApp would include a number of additional data entry features. For example, users could upload geo-tagged pictures and video footage (often useful for damage assessments).  In terms of data consumption and user-interface design,  MatchApp would be modeled along the lines of the Waze crowdsourcing app (depicted on the right above) and thus designed to work mostly “hands-free” thanks to a voice-based interface. (It would also automatically sync up with Google Glasses).

In terms of verifying check-in’s and content submitted via MatchApp, I’m a big fan of InformaCam and would thus integrate the latter’s meta-data verification features into MatchApp: “the user’s current GPS coordinates, altitude, compass bearing, light meter readings, the signatures of neighboring devices, cell towers, and wifi networks; and serves to shed light on the exact circumstances and contexts under which the digital image was taken.” I’ve also long been interested in peer-to-peer meshed mobile communication solutions and would thus want to see an integration with the Splinternet app, perhaps. This would do away with the need for using cell phone towers should these be damaged following a disaster. Finally, MatchApp would include an agile dispatch-and-coordination feature to allow “Super Users” to connect and coordinate multiple volunteers at one time in response to one or more needs.

In conclusion, privacy and security are a central issue for all smartphone apps that share the features described above. This explains why reviewing the security solutions implemented by multiple dating websites (especially those dating services with a strong mobile component like the actual Match.com app) is paramount. In addition, reviewing  security measures taken by Couchsurfing, AirBnB and online classified adds such as Craig’s List is a must. There is also an important role for policy to play here: users who submit false misinformation to MatchApp could be held accountable and prosecuted. Finally, MatchApp would be free and open source, with a hyper-customizable, drag-and-drop front- and back-end.

bio

Crowdsourcing Humanitarian Convoys in Libya

Many activists in Egypt donated food and medical supplies to support the Libyan revolution in early 2011. As a result, volunteers set up and coordinated humanitarian convoys from major Egyptian cities to Tripoli. But these convoys faced two major problems. First, volunteers needed to know where the convoys were in order to communicate this to Libyan revolutionists so they could wait for the fleet at the border and escort them to Tripoli. Second, because these volunteers were headed into a war zone, their friends and family wanted to keep track of them to make sure they were safe. The solution? IntaFeen.com.

Inta feen? means “where are you?” in Arabic and IntaFeen.com is a mobile check-in service like Foursquare but localized for the Arab World. Convoy drivers used IntaFeen to check-in at different stops along the way to Tripoli to provide regular updates on the situation. This is how volunteers back in Egypt who coordinated the convoy kept track of their progress and communicated updates in real-time to their Libyan counterparts. Volunteers who went along with the convoys also used IntaFeen and their check-in’s would also get posted on Twitter and Facebook, allowing families and friends in Egypt to track their whereabouts.

Al Amain Road is a highway between Alexandria and Tripoli. These tweets and check-in’s acted as a DIY fleet management system for volunteers and activists.

The use of IntaFeen combined with Facebook and Twitter also created an interesting side-effect in terms of social media marketing to promote activism. The sharing of these updates within and across various social networks galvanized more Egyptians to volunteer their time and resulted in more convoys.

I wonder whether these activists knew about another crowdsourced volunteer project taking place at exactly the same time in support of the UN’s humanitarian relief operations: Libya Crisis Map. Much of the content added to the map was sourced from social media. Could the #LibyaConvoy project have benefited from the real-time situational awareness provided by the Libya Crisis Map?

Will we see more convergence between volunteer-run crisis maps and volunteer-run humanitarian response in the near future?

Big thanks to Adel Youssef from IntaFeen.com who spoke about this fascinating project (and Ushahidi) at Where 2.0 this week. More information on #Libya Convoy is available here. See also my earlier blog posts on the use of check-in’s for activism and disaster response.