smsociety15 has ended
Back To Schedule
Wednesday, July 29 • 15:16 - 16:45
"Detecting Measures for Community Wellbeing on Social Media"

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

Author: Ziad Matni

Information about our local environment is everywhere on the Internet and other information and communication technologies (ICT), especially in certain forms of social media. Our engagement and interaction with this information in our daily social lives, as individuals and groups, has grown tremendously in the past decade and continues to do so. Social awareness streams (SAS) encompass “real-time” information streams generated by Internet services like Twitter, Facebook, FourSquare, and Flickr, and are at the forefront of changing the framework of how our society utilizes information. SAS are used by many millions of people on a daily basis to help them stay connected by communicating via brief messages, typically in public or semi-public forums. These real-time technologies also yield very large amounts of data from, and about, the people communicating and their communities, both virtual (Gruzd et al., 2011) and geographically-localized (Schwartz et al., 2013). But looking beyond events being detected, how can we gauge from the data how well individuals and groups of community-based individuals are doing when something disruptive occurs in their lives? 

I want to find reliable measures of community wellbeing by looking at data in user-generated informational technologies, like SAS (or in other instances, like in personally-carried monitoring devices like smart phones, health monitors or other wearable computing devices)? There are many aspects to what might be termed “community wellbeing” – its measure of civic unrest, the health of its aggregate population, the financial stability or wealth of its aggregate population, the level of education of its population, etc… I propose to detect some of these aspects in SAS data and show that when these wellbeing measures are low, we begin to see signs of stress and anxiety on individual and community bases. Likewise, when these measures are high, we see what I term as “tranquility” of individuals and community groups. The tranquility of a community is important for community leaders looking to assuage possible concerns from the citizenry about abnormal health and/or safety events concerns that can arise from wide-spread anxiety. This anxiety can be due to health epidemics, a rise in police or fire fighter activity, natural disaster occurrences and their aftermaths, or extreme weather occurrences – in short, anything that might disrupt the lives of people’s daily lives. 
Based on these initial findings, I would like to extend this research to include looking at ways to see if we can influence individuals and groups to change or modify their behavior on the basis of this data and information we might gleam from their social networks. Once data is aggregated and analyzed, what manner of feedback via the social network might prove to be the most effective and why? The motivation is to better understand how influences through social networks can help ease or moderate the level of anxiety that people and groups of people are feeling vis-à-vis their community. Likewise, this might help us better understand if this feedback mechanism can be used effectively to agitate communities in order to give rise to community activism (e.g. get people more motivated about community-level causes/perceived threats or even global ones, like climate change)? 

I propose to study this through content analysis of collected data from SAS. Collecting geo-located data from individuals in localized communities is relatively straight-forward, as established in various other studies and demonstrations (Schwartz et al., 2013; Matni et al., 2014). The research would then necessitate some content analysis and coding of the data – or at least a randomized portion of the data – to reveal keyword use in messages that mention something about community wellbeing or aspects of community wellbeing (such as mentions of civic unrest, increased police or fire-fighter activity, health contagions like colds, announcements of new jobs, or those lost, and so on). The content analysis should additionally look at how these detected messages talk about events that bring about either an anxious or tranquil state-of-mind of the messenger. Part of this includes an investigation of how often these keywords are used during community wellbeing-related messages and if their use has predictive value. Finally, the analysis should show if certain proposed aspects of community wellbeing are more influential and more commonly mentioned than others, giving us a sense of how to best construct a definition of “community wellbeing” as reflected in SAS. 
For the second part of my study (which seems contingent on the results of the first part), I would use methodologies that could include examining the extracted social networks of the SAS users, if made available, or alternatively the exercise of a simulation of different social network structures, as inspired by the study by Suri and Watts (2011) that looked at cooperation and contagion for public goods in social networks. 

avatar for Ziad Matni

Ziad Matni

PhD Candidate, Rutgers University
My research on information behavior in social media. And how to make a killer margarita!

Wednesday July 29, 2015 15:16 - 16:45
(9th Floor) TRS 3-176 (Ted Rogers School of Management) 55 Dundas Street West, Toronto, ON M5G 2C3

Attendees (0)