***** To join INSNA, visit http://www.insna.org *****
Thanks for the pointer. I remain confused.
What I have done so far is (1) transform a two-mode network into a one-mode network; (2) remove loops and simplify the result, summing multiple lines; and (3) used the k-neighbors command to identify the k1 networks for two vertices. The problem, as I see it, is to (a) use the Operations>Extract from network command to extract the two k1 networks, (b) count the number of vertices in their union, (c) count the number of vertices in their intersection, and (d) compute the Jaccard coefficient using the results of (b) and (c). But (b) and (c) are where I get hung up.
The summing up neighbors, max of neighbors and min of neighbors commands described on page 35 of the manual assume a vector, but at this point I have no vector to work with. Where does that vector come from? Also, the description in the manual suggests that these commands generate results for individual vertices. What I want to do is to compare pairs of vertices.
As you can see, I am, indeed, confused. Page 35 in the manual has failed to enlighten me. Can you give me a hand here.
Your help is much appreciated.
SOCNET is a service of INSNA, the professional association for social
network researchers (http://www.insna.org). To unsubscribe, send
an email message to [log in to unmask] containing the line
UNSUBSCRIBE SOCNET in the body of the message.
On Sat, Jun 18, 2011 at 9:05 AM, Vladimir Batagelj <[log in to unmask]>
<<<-------- John McCreery-------->>>
> Dear Colleagues, See page 35 in the Pajek's manual.
> The sociological problem is as follows. Two individuals sometimes work
> together on project teams. At one extreme, they always work together on
> same teams; thus, the k1 neighbors in their ego networks are, except for
> each other, identical. At the other extreme, they work together only once,
> on one particular team; thus, the k1 neighbors in their ego networks
> But how much do they differ. I am proposing to explore this issue using
> Jaccard coefficients to quantify the overlap between ego networks. I find
> myself in a situation I often encounter; I am working out a hack using
> Filemaker Pro, where my data is stored, and Pajek, which is my go-to tool
> for network analysis. I suspect that I am reinventing a wheel that already
> exists or unaware of a better approach to the basic problem—comparing the
> overlaps in pairs of networks.
> I would be very grateful, indeed, for helpful advice.
Vladimir Batagelj, University of Ljubljana, FMF, Department of Mathematics
Jadranska 19, 1000 Ljubljana, Slovenia
The Word Works, Ltd., Yokohama, JAPAN
Tel. +81-45-314-9324[log in to unmask]http://www.wordworks.jp/