Education and training program in the hospital APR.pptx
Mining Online Social Spaces in Support of Policy Decision-Making
1. Mining Online Social Spaces
in Support of Policy Decision-Making
Dana Rotman, PhD candidate
February 3, 2010
2. What do online social spaces offer us?
Ever-growing in popularity, online social spaces
capture rich data provided by their users
Types of available data:
Dimensional data – demographics, location, patterns, personal
features
Structural data – personal ties, interests, commonalities,
ratings, favoring, group-memberships
Navigational data – movement (clicks) among topics
Content – user-generated content, remixes, content reuse,
commentary
3. What tools can we use?
Log analysis
Social network analysis tools (Hadoop, NodeXL)
NLP
Manual extraction and analysis of data
4. YouTube and the Healthcare reform debate
An example of the YouTube healthcare-reform network representing
rating and comments
6. Twitter #Haiti mentions
Mentions and retwitts of the hashtag #Haiti, January 27, 2010 (14 days
after the earthquake). Several hubs of discussion emerge.
7. Content of tweets created multiple hubs of topical mentions and
information sharing
I’d like to present a couple of examples of much discussed issues and the ways they are reflected in online social spaces. These are visualizations of the networks of content that are available on two major social spaces – YouTube and Twitter. The visualizations were created using NodeXL, a free add on to Excel. The YouTube healthcare reform network is a good example how through mining of online social spaces we can learn about attitudes, interests and points of friction. In this visualization we can see the network with an emphasis on the number of views and comments that each video received. Comments are mapped to size and colors to ratings, and we can see that videos that generated a lively discussion (more comments) are not necessarily those that were rated higher, hinting that controversial content may be the crux of many discussions, but the heated debate they create causes lower ratings.In the next visualization we can see the different clusters of videos that discuss the same topic, healthcare. We can then dig deeper into the content and see on what mutual ground these clusters were created. Similar political opinions? (the red cluster – supporting the plan, yellow – opposing, others – parodies, commentaries, random).
Another example is the Haiti earthquake and its reflection in the Twitter network. These graphs were taken off the Twitterverse about 2 weeks after the initial shock, and we can see that despite the hype that surrounded the role of Twitter in the efforts following the earthquake.When we look at the content of the tweets, and here I brought just a couple of examples to such tweets, we can see the clusters, or hubs, of mentions and information sharing that is based on the specific topics: one hub is all about prayer and religious efforts, the other is much more practical, calling for users to share their ff miles to help others reach Haiti and help in rebuilding it. If we want to understand more about information sharing, interests of users, intervene, formulate or implement a policy based on the zietgeist that’s reflected online, there is a multitude of tools that are readily available, could and should be used not just for academic probing but for as a very practical tool.