[SOLVED] Inter and Intra attribute links

[Ali_Abbas]

Hi folks,
This is my first post here. Before I ask my question, I'd like to thank Gephi team for developing such a nice and handy graphical tool. The active support you give here is also quite commendable. I have been using Gephi for a few days. Have accustomed myself with its basis functionalities.
Here is my question: I have a network with 0.1 Million vertices and 0.5 Million edges. Each vertex has religious and geographical location as its attributes. All attributes are in String format. My aim is to identify attribute level communities. For instance, I would like to know (and visualize as well), how much links exist between each religious and geographical attribute. For example, there is a 'Christian' value for religious attribute. Now I would like to know how many (Christian)-(Christian) and (Christian)-(Other religious values) link exist. In Social Network Analysis, this is known as Silo Index. I have played with the filtering mechanism, but so far, I couldn't figure out how to achieve this. Kindly suggest.

Thanks a lot!

Cheers

[admin]

Hi, welcome here!

Use the Partition panel, select the religion attribute, then use the "Group" button.

The weight of the edges between the religious groups are equal to the sum of the weights of the edges between the nodes of different religion. So if all the edge weights are equal to 1, the weight of the hyper-edges is the sum of the edges between the groups.

I'm not sure if it's clear...

[Ali_Abbas]

wow! that was really quick. Thanks a bunch!
I can't give this solution a try at the moment since the network that large requires all my computer's resources; and I currently am busy with R.

There is one more question I'd like to ask. Is there any mechanism to do this analysis (both visual and statistical) on a number of graphs? I have many on which the same analysis has to be done. If somehow I can automate the whole procedure, it will expedite the whole exercise.

Thanks again!

[admin]

So you may need this or this.

[Ali_Abbas]

Thanks! So nice of you. I will look at it.

[Ali_Abbas]

After sometime, I have resumed using Gephi. Your suggestion did work. Thanks again! There is one little thing I could not get it though. When you group nodes by their attributes, and then see your graph, it is always shown in a circular manner (all group nodes fall into one another according to their sizes and no edges are shown). I don't know if I have clearly stated the problem. If you like, I can share you a snapshot of it?
I would instead like to have a minimized graph where node size is represented by the number of nodes with that attribute, and edge weight as the number of links between them (which is already being done).
As a workaround for this, I tried to export the nodes and the edges table as a csv file, and then import it in another workspace. Although I could see the graph then but the node size information is missing in this case. Isn't there an easy way to achieve it?
Thanks!

[admin]

Hi,

Unfortunately not, but this could be an interesting feature. I filled the Ideas List.

[shan_333]

The weight of the edges between the religious groups are equal to the sum of the weights of the edges between the nodes of different religion. So if all the edge weights are equal to 1, the weight of the hyper-edges is the sum of the edges between the groups.

I'm not sure if it's clear...

What about the size of the node ? How is it computed ?