[MUSIC] And that brings us to our last point, how can we analyze the network with a software? And that is then part of our lab, but let me start a little bit. Network analysis software has come a long way. And also network analysis only became possible thanks to the software. I mean, you saw how tedious it is to count these things by hand. It's really impossible. So network analysis has been around since the 60s. And people have been reasoning about it. But once you get beyond 20 nodes or 30 nodes, it's really difficult to still do network analysis. But if it's a computer, yes, you can just rattle it through and the faster computer is, some very large networks, not even the computer can get to them. But then you can make approximations, and since we have these powerful computers, network analysis really got a big boost. And there have been several softwares as well that allowed to analyze actually this network structure. Last time I can't actually have been observing this because I'm using the software as well as many of my colleagues, and trying to see which software to use for myself and to teach you. Last time I looked at it in 2013, and these Wikipedia side of social network analysis software, there were 100 different softwares listed. Now the good news is five years later in 2018, there are only 15 listed. [LAUGH] So there seems to be kind of like a process of consolidation going on where some softwares merge of you find out which ones are useful. Some are more useful than others, or more often used, more user-friendly, have more investment behind. And these softwares have still different advantages, I mean, among these 15. Some are very plug-and-play, they make it easy, they're very many predefined buttons, some are very flexible. So, for example, if you work with R yourself, and you do network analysis with R, it's very flexible, you program it yourself. Some others give you the pretty pictures, for example. So yeah so when we analyze it, I do the same as well. So I might use the plug and play to calculate something quickly, then I use the special one in order to calculate some specific things that are not in the plug and play solution. And then I use another one to make the pretty pictures. So there are complementary benefits of different softwares. One of the more common ones, one of the basic ones is called, Gephi, that's a software we will work with. That's pretty much the starter, it's kind of like the Excel for network analysis, the SPSS if you do statistical analysis. And we can get going with it extremely quickly. Now when you work with different softwares and different tools not only that your output is different, also your input is different. And as a final caveat, I want to talk about that because that connects to something we talked about during the lecture, it's the rogue representation. So we had our network here, which was called Jorge, Maria Juan y Magda, and I represented as an adjacency matrix. So in this adjacency matrix, we have one person that's connected to the other person up here. And for some softwares you can just take this adjacency matrix and feed it in and it spits out the network graph. So you will put in this adjacent matrix and it gives you this graph as a result. Now, but there are other ways to represent the network in a data format, right. So this is an adjacent list here. So you have Jorge who is connected to Maria and Magda and you just simply write in out. So who is Jorge connected to? Well, Maria and Magda. Who is Maria connected to? Jorge and Magda. Who is Juan connected to? Nobody. So we just basically write nobody in this line. So that's an adjacency list. And for some networking software, you just give basically this list. And you upload it and then you get this network graph. And then some very common one is an edge list. So, a third way of representing your network. So you have the source and the target. So basically, here you have one line per connection, per degree. So Jorge connects to Maria, Jorge connects to Magda. Maria connects to Jorge da da da, and Magda connects to Juan. So I have here one line per degree in my networks. If you, for example, go to this very common network software Gephi, that's how you often will represent your network, through an edge list. So you have all these edges, edges are links, that's the mathematical term for it, vertices and edges. And vertices are the nodes and links are the edges. And so you have the edge list, you have one relation per and you basically plug it in like this. And then the software will spit out the network. So there are some caveats of how you have to work with that. And the way I presented it today is not necessarily the only way of how to represent a network.