9 thoughts on “Visualization Demos”

1. Pingback: US election zoom-in | Interactive Visualizations

2. Pingback: Updated with IE-compatibility, conference appearance, wizard coming soon | Interactive Visualizations

3. Looks great! Several comments:

   1. Is it necessary to display uid, name, first name and last name? Name is already on the top.
   2. Locale: it’s better to convert the country code to country name, for user’s convenience.
   3. big_pic: since the URL is available, why don’t you just show the picture? Or at least make the link linkable.
   4. Modularity Class and Eccentricity are something that make sense in Gephi, but maybe users do not need to know. Sometimes people will get confused by the numbers.
   5. what’s the clustering method for different color display? not sure why some people bundle together.
   6. “Return to Full Network” is a little bit hard to find. Maybe by clicking on the blank area can guide users back to the home screen?
   7. “Directed” vs “Undirected” – since the follower and followee relationships are known, maybe the graph can be directed. Some people may have many connections, but it’s not because he/she is popular. Maybe he/she just followed a lot of people. So the pointers will make more sense.. I guess..

   Reply ↓

   • Shujian, thanks for the comments. You looked at the visualizations just after some changes were pushed and within them we changed the data set. So, the dataset you saw was actually a Facebook-ego net, not the Twitter data. The ultimate goal of the project is to create a wizard to allow other individuals to easily create visualizations like these. Most of the very good ideas you raise are decisions we’ll leave up to the user, e.g.

     1. The creator of the visualization will be able to choose what attributes to display in the right hand panel. So, yes, probably most creators will not want to include all these
     2. We show attribute text exactly as in the dataset; so if a creator of a visualization will have to expand these in their network if they want to display them
     3. It is possible to indicate which attribute is an image file and load this and the restored Twitter data does this. The creator of a visualization will have to indicate what attribute has a url to do this
     4. We’ll leave displaying these or not up to the creator of the visualization. I agree that they can be confusing to viewers without proper explanation.
     5. Clustering in the Facebook network was a force-directed layout on who linked together most. The Twitter network now in use is much more dense so it doesn’t break apart as much.
     6. Good point on the “Return to Full Network”. I have to look into how easily we can capture the clicks onto empty areas of the canvas
     7. Directed networks are indicated with (small, probably too small) arrows and directed can be indicated in the right-hand panel (as in the government visualization). The Facebook data is undirected and the Twitter data has been narrowed to only mutual relationships, which is why direction is not shown in these graphs.

     Thanks again for the comments! We really appreciate them.

     Reply ↓

4. Great work! It reminds me of Wolfram Alpha’s (https://www.wolframalpha.com/) facebook report – do you have a stake in that also?

   Reply ↓

   • Hi Dominik! Thanks for your kind words.

     We don’t have a stake in the Wolfram Alpha personal analytics project, but we have certainly taken note of those developments. In collaboration with my supervisor (Dr. Bernie Hogan) we have used the InteractiveVis project to visualise Facebook networks and allow users to export their data. If you are interested in taking a look, you can find the app at https://apps.facebook.com/namegenweb/

     Reply ↓

     • Thanks – this looks even better than WolframAlpha, because one can actually export the data! Can wait to play around with them in Gephi ^^
       I was wondering lately – how do you get access to FB data? Or is it only the personal user data they are supposed to give to you by law? In the end, it’s probably the biggest sociological database on the planet… in private hands -.-

       Reply ↓

       • Yes Dominik this is something of a personal peeve of mine too – there is no way to access raw Facebook data unless you are a member of the data team or you are one of the privileged few researchers who are given access. The possibilities afforded by open access to researchers would be very great indeed.

         Our app simply accesses the data of the user who has authorised it, using the Facebook API. There is no way to access any data that would not otherwise be available to the user (for example two degree “friends of friends” networks are not possible because users cannot download their friend’s friend list, so we must instead capture a 1.5 degree mutual friends network). If you are interested in how to capture Facebook data, the sourcecode for NameGen is available on GitHub for free. We welcome contributions.

         Reply ↓

         • Uh, nice! I’ll have a look at it! t-1.5y to finish my MPP at Hertie School, the so applying to OII!