Node Degree of centrality in Visone

74 views
Skip to first unread message

Sayeh

unread,
Apr 6, 2021, 11:13:30 AM4/6/21
to visone-users
Hi guys, 
Can you help me how I could visualize degree of centrality for nodes in a two-mode network in visone?

Volha Kryvets

unread,
Apr 6, 2021, 3:39:18 PM4/6/21
to visone...@googlegroups.com
Hi there!
If you're talking about two-mode degree centrality (how many connections a node has with actors from a different mode), you simply run degree centrality calculations, then in attribute manager select node from one mode first, apply your visualization algorithm specifically to them (let say color them in green). Further, you select the second mode and do the same (color them in yellow). 
If it makes sense in your research, you can transform a two-mode network in R into one-mode, do your calculations, and import these calculations to the initial bipartite network, then visualize modes of nodes separately using the algorithm I described.

Best,
Volha

On Tue, Apr 6, 2021 at 6:13 PM Sayeh <n.ash...@gmail.com> wrote:
Hi guys, 
Can you help me how I could visualize degree of centrality for nodes in a two-mode network in visone?

--
You received this message because you are subscribed to the Google Groups "visone-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to visone-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/visone-users/08a37561-1db4-4404-a124-966f138c5168n%40googlegroups.com.


--
Volha Kryvets, Ph.D. candidate
International Relations
School of International Relations and Public Affairs, Fudan University
220, Handan Road, Shanghai, 200433, China

Philipp Burghardt

unread,
Nov 25, 2021, 4:09:59 AM11/25/21
to visone-users
At this point I have a further question:

Maybe a litte introduction: I made a Two-Mode-Discourse Networt with DNA (Discourse Network Analyser), I exported the Data one the one hand as a .graphML file on the other hand as a .csv (just to be more flexible)

My question is: the exportet matrix consits of more than 0 and 1. DNA automaticly uses "0" for "no tie at all", "1" for "Agree", "2" for "Disagree" and "3" for "neither only 1 nor only 2". Is this a problem for calculation Degree/Indegree and/or Outdegree? I alredy analysed it in Visone, Visualized it but I'm not quite sure if the calculated values are actually right or if they are distorted because its not a binary network. 

I'd be thankful for a little help!

Best regards 
Philipp

Müller Julian

unread,
Nov 25, 2021, 10:55:46 AM11/25/21
to visone...@googlegroups.com
Hi Philipp,

> My question is: the exportet matrix consits of more than 0 and 1. DNA automaticly uses "0" for "no tie at all", "1" for "Agree", "2" for "Disagree" and "3" for "neither only 1 nor only 2". Is this a problem for calculation Degree/Indegree and/or Outdegree? I alredy analysed it in Visone, Visualized it but I'm not quite sure if the calculated values are actually right or if they are distorted because its not a binary network.

If you didn't tell visone to calculate weighted degrees, then what you get is that each incident edge adds "1" to a node's degree. You can check whether you accidentally told visone to calculate weighted degrees by checking the settings for the degree computation in the analysis tab. If link strength is set to "uniform", then the calculated degrees are unweighted, i.e., each incident edge adds "1" to a node's degree. If a particular numerical edge attribute is selected there, then weighted degrees based on this attribute are computed.

I can't tell from your description whether edges with associated value "0" are actually explicitly represented in your network or not. I think it is likely that you don't want these included in your degree calculation, so you will have to check if they are there or not.

Best,
Julian
Message has been deleted

Philipp Burghardt

unread,
Nov 26, 2021, 4:00:47 AM11/26/21
to visone-users
Thank you very much for the quick response! That's true, as i see it Visone does not use the "0 ties" to calculate the Indegree.
But how comes that the summed up standardized Indegree of my nodes sums up to 1,939? I thougt a summed up Indegree has to summ up to 1? (Percentage of my summed Indegree is 99,998 so close enough to 100%)

Best,
Philipp

Müller Julian

unread,
Nov 26, 2021, 4:39:52 AM11/26/21
to visone...@googlegroups.com
Hi Philipp,

The "standardized" option means that the calculated indegree value is standardized at each node, i.e., it is divided at each node by (n-1), where n is the number of nodes in the network. Since a node can have up to n-1 neighbors (as long as there are no parallel edges), the standardized indegree is guaranteed to be somewhere between 0 and 1 at each node.
If you want the computed values to sum up to 1 over all nodes, you rather have to choose the "percentage" option.

Best,
Julian
Reply all
Reply to author
Forward
0 new messages