Reflections on Sunbelt XXX

by johnmccreery

I spent last week in Riva del Garda, a spectacularly beautiful Italian resort town that lies in what used to be part of the southern Tyrol, i.e., part of the Austro-Hungarian Empire, until it was ceded to Italy in 1918, at the end of WWI. I was there for the 30th annual Sunbelt conference, the annual meeting of the International Network for Social Network Analysis (INSNA), to present the latest findings from my ongoing research on the network connecting winners of an advertising contest in Japan and to search for new approaches to analyzing the sort of data with which I am working. What is striking to me, returning to my favorite online forums is how similar the debates I encountered there are to those found all over the Net.

At a session on philosophical roots I heard about precursors to network analysis and balance theory in Spinoza.

At the mixed-methods session in which I participated, most of my colleagues were grappling with projects that begin by selecting a sample of egos, asking them about people with whom they interact, and using the results to generalize about the networks to which they belong. I was starting with whole networks combining several thousand events (winning ads) and twice as many individuals (members of the creative teams). We all were concerned with how best to blend quantitative and qualitative analysis to understand events that clearly require both.

Closest to my own immediate interests was the session by the “Vizards,” experts in visual representation showing off their latest ideas by developing their own representations using the same data set: scraped from a site that provides information on music groups and links to “similar groups.” It was astonishing to see a self-organizing map produce a hugely complex and detailed picture of musical genres based only on this information.

The major difference between Sunbelt and other forums in which I participate  is, of course, that most of the participants in the former are comfortable with quantitative analysis and involved in a subfield of social science to which mathematicians and physicists have and continue to make profound contributions.

Advertisements

2 Comments to “Reflections on Sunbelt XXX”

  1. Great stuff. Was the Vizards’ source Pandora? Their ‘music genome’ approach seems like it might have some overlap with network analysis.

  2. The Vizards’ source was last.fm. And, yes, the mapping begins with network analysis. The way I understand the map is supposed to work [Warning: I am nothing like an expert at this level of the game], the algorithm begins with triads discovered in a network of similarity relationships (artist A is similar to artists B and C). These are collapsed into lines (A-B-C) with a score proportional to the distance between B and A and C respectively. The process is repeated until proportions are determined for every triad. The proportions are then converted into distances and the layered clusters mapped onto a 2-dimensional space.

    This is a very rough description since part of the problem is accounting for the fact that the original network is directed. A may point to a list (B, C, D…) said to be similar to A, but one or more of the members of the list may not point back to A.

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: