The installation of Edgesense using the Edgeryders database lives here.@Alberto and @Matthias can confirm this is the version we will be using in the following months.
I think it would be a good start to familiarize ourselves with the tool and launch some key questions worth answering in this community testing.
We could then follow up on those questions in two events:
An online event later in the month (tbc) to introduce what Edgesense can do to the wider community. Hazem is probably best to lead it, as member and community manager.
@Hazem, would you give it a first shot at compiling these questions and noting your first insights, hypotheses etc? I will then update the wiki with my own.
so am in the process of getting to know edgesense better and I have some questions
1st - waiting for the confirmation on the above version ?
since the only option of selection is the nodes, so I assumed the analysis would be node based.
so is there a way to know more info about the node like the no of edges and nodes interacting with it. ( data not visualization) ?
is there a way to show a selected interval of time ?
the starting point is fixed in the time line and in order to focus on lote4 content to have a matching analysis with assemble so the main focus should be 11-12 months ago ? is that right @noemi
I can’t get the logic of colours. ( assumed that the orange is mostly the lote old content + new lote4, sky blue is future spotters , but what about the red and green ? )
what is the in and out degree… couldn’t find info at github
I tried to install Gephi after some struggle to get the alternative to java on ubuntu and am not sure yet how to get info out of it . also the colours isn’t working.
any help with that would be great … @alberto@mathias @ Luca_Mearelli
I forgot to mention this, try getting all that info by going through the tutorial, which contains a small test with explanations. Also, see that every graph has a “?” icon in the upper right hand corner with the term definitions.
You also might want to visit my initial review of Edgesense here, remember?
Hmm. Good point. We do need not to take anything for granted.
In-degree: the number of edges that connect to the ego. In your example, the ego is Alberto. Edgesense is telling you that Alberto has received comments from 250 different users. That, of course, does not mean he has received 250 comments, because each user is counted only once. If a new user leaves Alberto 10 comments, Alberto’s in-degree will become 250 + 1 = 251; it will NOT become 250 + 10 = 260. This is because the in-degree is the number of edges connecting to a node, and in Edgesense edges represent unique relationships. You can think of comments as being stacked on top of each other to give rise to an edge. As you see from the green pane in the top right, Edgeryders has over 4,000 edges made of over 15,000 comments. The difference between the number of edges and the number of comments is accounted for by the definition of edge (=> unique relationships) that I mentioned.
Out-degree: the number of edges that the ego connects to. In your example, the ego is Alberto. Edgesense is telling you that Alberto has left comments from to 391 different users. That, of course, does not mean he has left 391 comments, because each user is counted only once. For example, I am already connected to you, @Hazem: I am now leaving you a comment (because this comment is left to a parent comment, this, authored by you). However, this does not give rise to a new edge, because we are already connected. You can check this by looking for yourself in the graph and checking that the node representing me is connected to the node representing you by two edges, one for “me talking to you” and another for “you talking to me”.
So, your last question is meaningless. Alberto in-interacted (=> was left comments by) with 250 other users; and out-interacted (=> left comments to) with 391 other users. Some of these will be the same users, of course! This was done through 250 incoming edges and 391 outgoing edges. The degree of Alberto in this network is simply 250+391.
There are many notions of centrality. Betweenness centrality is the quality of one node of being part of the shortest path between many pairs of other nodes. Consider the star network:
Look at the node in the center, and take two other nodes at random: you will see that the shortest path connecting those two other nodes necessarily goes through that central node. The betweenness centrality of a node (call it N) is given by
[number of shortest paths of N]/[total number of shortest paths in the network, excluding those starting or ending at N]
In this network, there are 10 paths that do not start or end at the central node, and all 10 go through it. so, the betwenness centrality of the central node is 10/10 = 1. The betweenness centrality of all other nodes is 0 – you can easily check that by taking one node, for example the one at the top, and verifying that no shortest path between any two of the other nodes goes through the one at the top. The betweenness centrality is always a number between 0 and 1.
In sociological literature, betweenness centrality is often associated with brokerage and power. This is because, if two individuals interact only through you, you can derive advantages by this, through withholding information or outright lying (“the other guy says that…”).
thanks for the explanation … was puzzled by the value and how it was calculated ( all my trials failed to get it right , as you explained it is about the shortest paths not only the no of edges )
The whole point of Edgesense is: no installation required, you just need the browser.
I reiterate @Noemi's recommendations: (1) click on all the ? icons for explanation of what you see; and (2) do the tutorial. There is also a short demo video:
and I have some questions
- asked the same question last time-
