Network community structure

June 23, 2010

Friendwheel layout of a Facebook networkFriendwheel layout of a Facebook network
Bernie Hogan
bernie.hogan@gmail.com

Advertisements

7 Responses to “Network community structure”

  1. admin Says:

    PEER REVIEW COMMENT No. 1 – In this representation of Facebook, partitions are denoted by position rather than color – which is a clever way to layer more information into the figure. The groups breakdown along foci, or social contexts -high school, camp, college, work etc, and the pinwheel figure nicely illustrates the different foci in ego’s network. However, one wishes the highly interconnected foci were positioned closer together along the wheel in some way that would minimize the edge-crossing in the middle. I wonder if the duplication of degree in size, color and position is necessary, or if one could usefully integrate other information through color (say ratio of ties within and between foci), to increase the information content of the image.

  2. admin Says:

    PEER REVIEW COMMENT No. 2 – The pinwheel layout highlights the organizationally focused nature of friendship ties in the author’s ego network. The diagram captures the relationships that span organizational settings, and more importantly, it does a very good job of showing how densely (or sparsely) the settings are connected with each other. This effect is aided by having the set of ties related to each setting be sized proportional to its contribution to the ego network. The visualization may be improved by imposing some sort of order on where the settings themselves are placed, so that they are chronological or so that settings that share more ties are next to each other.

  3. admin Says:

    PEER REVIEW COMMENT No. 3 – This submission successfully rebels against the convention force-directed layout, reengineering the circular layout in an information-rich way. The degree ordering is effective in allowing one to get a sense of how different partitions connect to each other, both in terms of number and distribution of cross-ties. I’d love to see if this layout would work semi-circle (which might permit degree ordering to be arranged in a roughly up-down fashion while still preserving its radial merits).

  4. Bernie Hogan Says:

    I would like to thank the anonymous reviewers for their comments. I would also like to highlight the fact that the method for producing this network is given an extended treatment in my chapter “Visualizing and Interpreting Facebook Networks” in a forthcoming book, Analyzing Social Media Networks with NodeXL”. I mention this not merely as self-promotion, but because that chapter demonstrates a series of tunable parameters for such a network layout that addresses some of the comments above.

    For example, this network is done using a polar coordinate system, with radius mapped to the log of betweenness. That mapping goes from 0.5 to 1.0 (where 1 is the edge of the circle). However, a practitioner is welcome to use a different scale (such as 0.2 to 1.0) so that the scale goes closer to the center of the circle [reviewer 1].

    Doing this on a semi circle [reviewer 3] is entirely possible and simple. When the network is transformed from x/y coordinates to radians, one need only divide the result by 2 to create a semi-circle.

    To order the groups in some meaningful way [reviewer 2] is more tedious, as it involves finding and replacing the values for the community. There is no easy way to do this, but I will look into it, as I think this is a very useful addition to this technique. If I come up with a useful solution I will happily post it here.


  5. […] Facebook network visualization was published in the Journal of Social Structure Visualization Symposium.  As the author states, this visualization has some nice […]

  6. Vaughn Page Says:

    Like the first three commenters, I’m impressed by how well the pinwheel design depicts the connectedness both within and between the “wings.” It’s interesting to see which clusters mainly connect within the cluster and which reach out to other clusters. However, it’s impossible to see the individual connections within very active clusters, i.e., high school. It’s also frequently difficult to trace where the nodes of connectivity between clusters lie. Perhaps using colored lines rather than relying only on grey would help there. Also, “exploding” each wing to show only the connections within the wing, with interconnection lines leading off into “space,” would allow more detail to show. All in all, though, a great effort in which the author can take a lot of pride.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: