Facebook Page Like Networks with Netvizz and Gephi

hello this is a quick video on page like networks how to make them what they are page like networks are function provided by Panetta fees and if this is a data extractor for the Facebook platform and in this video I just want to quickly introduce this function and walk you through a couple of steps for making a page like network visualization and maybe show some tricks in the in the process so what a page like networks it's a very very simple principle the idea is that if we take a Facebook page for example here one of the what you know it's the American election season so on Republican National Committee and you can scroll down a little bit and here you get like by this page and then you have a whole bunch of other pages that appear here and page like networks are simply a means to analyze and visualize and well basically work with these these light connections right so if we go back to a net booze module here the whole principle is quite easy you simply start with a with a seed the seed is is is kind of the starting point for a crawl into the network and if you choose a depth of one the crawler that's part of net maze will get all of the pages liked by the seed and also the connections between them if you choose a depth of two well we'll go one step further so likes of likes and well this can quickly lead to quite a quite big network so how to get an idea also very simple you basically just get the URL and then we don't here you have a link that leads to this page here you can basically just you know enter the URL and then you get this number anta and that's the seed right so on you do this and you get a network file to download and the network file is in in the gif format which works very well with defi graph analysis toolkit which is a very easy to learn and I want to show you how to analyze this so I already got the like Network for the RNC on my computer I just used a depth 1 to keep it simple but before moving into Gaffey I just wanted to very quickly talk about the different variables or values that that you get for for each node most of them are really quite straightforward they come directly from the Facebook API so we have a category here that's really a facebook has an internal system for page categorization and then you have stuff like you know face count talking fan counts re talking about count and all of those elements but maybe tools are particularly interesting because one of them is not provided by Facebook itself this one here post activity that's and of an estimate of how often the page posts and what I did here is simply say it took the last hundred posts of that page and then saw our kind of a calculated a post per hour metric right so when you're analyzing this keep in mind this is only the last a hundred posts that this is is based on and you know for some pages this may only cover a day or two and of course there may be a lot of variation here so keep that account in mind and the second aspect that I think is kind of interesting here is the users can post variable which tells you whether the page allows for user posts or not users can always comment of course but can they post right so this is also interesting element but actually all those things come directly from the API with the exception of post activity so that out of the way let's move into our Jeffy - I've already opened it here and I just want to show you how to work with a page network file and it's really not very complicated just opening the file here and I have already am I already have a second one here I'm going to show you what to do with the second file in a moment but I here I have my RNC file I actually add it myself or in C in here um - well you know for memory purposes it's probably generally a good idea to - I use for any kind of research project a good kind of file naming system that allows you to easily you know find files and um you know kind of also don't forget which file what was which so I'm just going to open this here it's a directed file that means there's direction to a connection so one page may like another page that page isn't obliged to like back right so here you get some directionality now you can already see this page our RNC page likes 95 other nodes and there are 824 connections in this network and that's an you know it's quite quite that ok let's open this here ah we get our usual kind of chaotic Network element in the middle here it's um well and you know we have to do something with that to analyze it and when it comes to network analysis well the most important steps are to you know do something with maybe size and color to allow us to to see some variables in inside of the network so it's a one of the two important things and the second is to - layouts the network a little bit to make it more readable and to basically use the Apollo apology of the network to create a readable graph so we're going to use the force Atlas to algorithm here which I'm not just run yeah you can see uh-huh let's now looks a little bit better but it's it's still extremely extremely dense here so I'm gonna gonna scale a little bit maybe maybe hundred why not okay and now things are already a little bit more readable so we already noticed some that there are some kind of denser zones around here a little less density up there but of course we need to do a bit more to be able to to a doable interesting analysis here so first thing I'm going to do in addition is to use a color to add some information to our graph so here we have this appearance window on the left side notes I'm just going to color the the nodes is the color palette icon indicates color and I'm going to choose an attribute and here you have all of those sum all of those variables that you also already saw on on the the net miss page so I mean I can take for example post activity the one I've talked about before so it's post per hour and I can use that to color the nodes personally I prefer a nice heat scale and here you can already see that the mood is this page here seems to be very very active compared to the other ones we'll see in a moment what that is and I'm gonna use node size to visualize a second property of the node so I go back here left and here we have this icon and that leads to a two size and then I can exhume use it in exactly the same way I'm gonna use fan count right so how many people are like this page and here you know it can set a minimum size and maximum size this is all really relative it's it's basically about them making it readable right this is this is something that needs to be interpreted by a researcher so those different element Sahra are there to help you then analyze those things they're not to create this kind of super standardized methodology here already SIA this page is like a lot more people than the other ones let me run this again yeah so it's still quite dense um yeah and here we already have something that we can maybe a look at in more detail you're the left bottom we can add labels and while these labels I help us to and to really make this all a more more readable let's make them a little bit smaller like this huh and here we can see so our friends are Mitt Romney here is is the one with a lot of lot of likes well very interesting no Donald Trump here I let you interpret what that means but yeah very very big lots of likes Herman McCain has or he had Herman Cain very reactive for his page he's posting a lot apparently one of those things we can of course also look at in the data laboratory that very often makes a lot of a lot of sense to understand that things a little bit better and a human could for example you know filter close this and with more space filter on label example here we got our different Keynes John McCain and Herman Cain and we could also go look for oops Trump and indeed no Trump here this could be also interesting for statistical network analysis so for example we could go back here they kind of generate some statistics for example you know PageRank Authority measure created that we could use that individualization it also go back into the data rate laboratory we now have a PageRank measure and well unsurprisingly the RNC page has the highest authority value in this network then Mitt Romney the second highest one and this could then be used you know in another kind of a approach you know using Excel to maybe use you know topology metrics for analysis so that real quick is a a page like network Wow could of course now go into the preview you know make all of this maybe a little bit nicer nicer looking show show labels here and and then have a something that we can also output remember very useful on in Jeffy Jeff he uses those rounded edges to to indicate relationality right and it's um the direction of of the arc that shows us that here Republican Party of Pennsylvania likes Pat Toomey but not not the other way around right so that is can be used for interpretation you'll have a lot of pages that have mutual relationships for example here Herman Cain likes the South Carolina Republican Party and they like him him back and it's why you have this kind of double arc structure but it's it's um something fortunately also use some arrows here but I find the curves if you get used to them kind of very very amazed and other thing I really like to do is to use them you can see here sometimes difficult to read but it was a bit too far the labels can be difficult to read because of the black around the node you can you know here choose border color and then use the parent and then you do not get this um this black circle around the node which makes it a little bit more readable another thing that's nice and also we like to that is maybe not use black here but but white and then you get a little bit of a space you know between the no it's a yeah but I wanted to show you quickly two other things that are kind of interesting with a page like networks well first of all this is relatively small because it has only a depth of one if if we were to take a depth of two we probably already go into the thousands right so if we would not just take into account the first kind of set of pages directly liked by the Republican National Committee but also like by those like by the RNC well very quickly we go into into big spaces but sometimes we want to make may be more purposeful comparisons or analyses and you can very easily actually add a second network to to this one so before I already a you know had kind of here a second page open which is from a Breitbart news yeah kind of an alt right kind of Tea Party publication and I've already had generated the networker for them as well I'm just going to open that here now open add that here and instead of selecting here new graph when I say append graph and what it's going to do it's just going to add the second Network directly in here and if nodes are shared between our two networks they're going to be connected because as you can see here in data laboratory the IDS of all of those nodes are actually their Facebook IDs and if while if the two networks have well the same ID and and well they would if they're the same pages then indeed the nodes are are fused and they come in together so let's let's lay out that again and you can already see there is some connection between those two networks here in the middle and that is unsurprisingly Michelle Bachman a bit of a tea party star those two networks stay in a pretty far away because well this one is very dense and in bright Bart's network here which has been smaller is also very dense can also see that actually it's doesn't have color solute and size so we can you know just reapply those things here we can see okay well Fox News and then they go back to color post activity apply oh and here we see Breitbart getting getting really red and you can probably see on this side now that the Herman McCain's page is no longer red sprite Bart now is by far the most active page let's look here in the data laboratory post activity wow it's a roughly 8.8 post per hour right so well it's election season so they are they're on fire but that means that suddenly a Herman Cain which used to be kind of the top top scorer here's is only at a place while five and yeah there are other pages more and more active so take that into account they're pretty far away I don't like that if we want them to come together more closely we can go here on the left into the fourth atlas to a panel and add more gravity I'm going to just go from 100 run it again and you can see two are now a much more closely related so that can be very interesting if if you have no kind of a number of pages doesn't have to be too long can be more so you can basically just add in the same way I just did it you can just add pages to your network and this will allow you to see how okay so what kind of conversions is there which kind of actors and you find in between of those in between those lightning-like networks one can definitely do more with page like networks also you know if you go even deeper if you do crawls that have a depth to kind of explore a little bit the kind of the cultural actor space behind all of those things and you know if you're very very ambitious you could you could map you know quite quite big networks kind of combining several depth two networks or maybe getting an idea about kind of bigger cultural spaces I also didn't talk at all about kind of further analytical methods that are built in when I quickly ran patron here but there's more stuff than one could use but that this is a this is the basics of beige like networks and I hope this was interesting to you have a and have fun analyzing

Channel: Bernhard Rieder

Views: 23,904

Rating: 4.7818184 out of 5

Keywords: digital methods, facebook, data analysis, netvizz, gephi, social media, digital sociology

Id: mLOSLYNWmBA

Length: 18min 41sec (1121 seconds)

Published: Mon Oct 10 2016