 Okay. So once you have download installed Gefi and download the zip file, you can unzip the file and it should present you with a couple of different Excel file sheets and CSV files and PDF. You can ignore this one right here. If you don't have that file, just let me know. Raise your hand or tell us you need a minute. But put it somewhere easily findable on your desktop or other location. And once you are done with that, please open up Gefi and it should take you to this intimidating interface. You can in my project, so everybody, if you are on the Gefi interface, you should be in this overview tab. Data laboratory is where all of our statistics will live. We don't have to really worry about that. Preview will come to that later. This is a tab I wanted to go over. Okay. So we have downloaded a couple of CSV files. CSV stands for Comments Separated Values. You can think of it as a very stripped down Excel spreadsheet file. It's like the barest bones plain text version of a spreadsheet file. But what we're going to do is we're going to go to File and we're going to go to Import Spreadsheet. And then point that to the directory where you unzipped all your files, the Gefi workshop data files. And what we're going to do is highlight ASO, IAF, all nodes, the CSV file. And we're going to open that. And this should bring up this other menu where it's going to give you the breakdown of the columns and rows. And everything looks fine here. Separators comma and identifies it as a nodes table automatically create a Unicode format. That's why I'm going to worry about that. And once we're happy with that, we can click Next. I give you Import Settings. This all looks fine. ID, label, intervals, click Finish. We have another menu, an Import Report. It shows you the number of nodes in 796. Now I happen to know that this is an undirected graph type, so we're going to change this to undirected instead of mixer or directed. And once we're happy with that, we can click OK. This should bring up this square with a bunch of dots. Everybody over distance all right? No. I obviously didn't do something. I downloaded Gefi, but I don't have any of that files. Oh, sorry. We're going to wait for a second here. How about everyone else? The nodes. Any issues? Someone posted about the Java installation error. That's what I had. There's a YouTube video I can link that solves that. OK, great. I'm going to have to, like, change something. Oh, I see. We'll say, like, Java or not. Sorry, what was the problem? Someone in the Zoom chat there said they're having a Java installation error. I had the same thing, I'm sure. What's the video called? Sure. I watched it every week. This is on a Windows machine. Yeah. Something not found. You need to have Java installed on your Windows system. Yeah, well, I already had it installed before. It's still on this error, for some reason, or at least it's a fairly common issue. And again, this is being recorded, so if you follow stuff, you can watch the recording later. So we are playing with a prefab data set that someone has already been nice enough to put together. And this data set is actually all the characters and their relationships in the Game of Thrones book. And if you're familiar with the TV series or the book series, you know, that's quite sprawling. So it's a good example of the kind of robust data set that you can tackle. Obviously, we're not going to build our own data sets, but it's about to be a good way to introduce to you the capability and the potential of the app. How are we doing? We're good? Yeah, thank you. And other than installation problems over distance, anybody else have any issues? Sound like it. Okay, well, everyone's okay, I'm just going to move forward. And then during the workshop portion, we're kind of working on how things come back and help out anybody who's got issues. All right, so we have our nodes. And if you're familiar with the network, there are nodes, which are like the hugs of where the interactions occur. And there's edges or vertices, those are the connections, could be on different nodes. So right now, we just have the nodes, but now we've got to import the edges, your facial about edges. So what we're going to do is go back to file, and go back to import data spreadsheet. And you'll see another file here called ISO IAF, all edges. That's the one we want to import. And click open, similar menu, tables and rows here, automatically recognize this is an edges spreadsheet. And we're going to click okay, this all looks fine. Click finish. And here's where it gets a little tricky. Automatically, the text is undirected, it's fine. We need to append this to the existing workspace, not a new workspace. Click on that radio button. Okay. And you see your network should have changed to this hot mess. You guys good? Okay. So now we have all that information is pictographically, but it's very difficult to read, right? Can't really make a whole lot of sense. So what we need to do is play around with the layout down here on the left. And you're going to see if you click on this drop down menu button, you can see a bunch of options for different layout forms. And you can play with these at your leisure future. But right now, we're just going to focus on four satellites. And it should give you all these options or criteria for satellites. What we're interested in is repulsion. It's going to determine how far along how far the separate nodes are. Let's highlight repulsion strength. I can change this to let's say 9,000. You can put in whatever number you want. Let's start with 9. Click Run. And you should see the network begin to expand. Yeah, it's okay. So it's expanded. If you want to adjust your view, you can click on this magnifying glass. That should take you to a broader view where you can use your mouse wheel to zoom in and out. And when you're happy with it, you can just let it run and click Stop. And now this is better. We can kind of discern different areas of concentration, outliers, but it's still difficult to read. So let's play around the interface a little bit more. What we're going to do is we're going to go focus on the color tab. Let's go to Appearance and then Nodes. Make sure I click on the Nodes tab. We're going to make sure we're going to click on the color palette here. And then we're going to go to ranking. Yeah, go to ranking and then choose an attribute and click on Degree. And once you click on Degree, you should give you this gradient. Let's apply that. You can see that that makes it a little bit more readable. This is by using a different color scheme. You can see again, which areas of concentration are of notes in which areas of outliers are faded, right? And you can play with the color scheme if you like. I'm just going to go to the default, if you have all these options here on right click this little button here. Yeah, every time I hit the gradient, I have a new ticker that pops up. So I have like seven tickers now, but I'm too much more. Okay, so I'm going to make sure we're on pure node ranking. And this should be highlighted. It's a color wheel. It's color balance. Does that result? No, but I'm going to apply it. With all those arrows, there was like three elements. Yeah, so I was like, oh, okay, I was like, what is the more connections Yeah, we'll get into that. Cool. All right, is it okay? This one's people are good. Okay. Okay. Again, on parents, I'm going to be at nodes and make sure the color wheels pick the ranking. Okay, so what we also want to do if we want to make it a bit more legible, we're going to run some statistics. And on the far right here, you have all statistics. And what we're interested in, you have all these options again, and we're going to run the network diameter statistic algorithm. So once you find a tab, you find network diameter. And then you use menu or the couple with window, click okay. And it should give you a little report. This is my report. We don't have to really worry about this, just click out of it, say okay, close. And if you're interested in looking at what kinds of statistics are run, what the results of that data laboratory, and I'll show you all this information. But yeah, we're not going to worry too much about that right now. Okay. So let's go back to appearance. Now that we have statistics and we know it's living in the data laboratory, go back to appearance. And we are going to go back to nodes. And you're going to click on sides. So instead of this color palette, we're going to go to the side. And then we're going to go to rankings. This window right here. And then we choose an attribute. I'm going to choose between this centrality. And it should give you options here minimum size or maximum size. And once you're there, you can click apply. So the results should be a graphical representation of the relative importance of different nodes, bigger nodes being represented by literally an extension of the node itself. Anyone have that? And you get the analysis. The analysis. I'm not sure what that global surgery, please. Jonathan Jonathan. Okay. So I think the question is, how did we get those statistics in the first place? And we did was we went to this menu here and click on statistics. And like over at network overview, and we can click on the network diameter algorithm, random algorithm. Once you did that, just add some data that's been internalized, and then you can represent that data by going back to appearance, those sides breaking. Yes, have you told us what this data is? Or is that like a reveal at the end? Oh, yeah, it's the Game of Thrones novel. Okay. So it's characters in Game of Thrones? And we'll show you how that works. I mean, if you're familiar with the TV show or the books, you can kind of guess who these different nodes are. Take a guess and see if you're going to be right. How do you center the map in the workspace? So there's a little button here on the bottom left. It's got a magnified glass, click that and it should center it. And if you want to zoom in, you can use your mouse wheel to zoom Okay. We get to look over. This at least the representation is not mistaken since that our photographs are usually just because the orientation of the nodes is like there's like there is now or and slash could we reproduce like your graph exactly, see um, no, I don't think so. I mean, every time I've done it, the orientation is a little bit random. Right. You can tweak it if you'd like. And that goes according to the layout representation. Right. So I mean, I could be wrong, I'm not an expert at this, but that's my understanding. Okay. I'm what's very similar. You're looking fine. Maybe need the way it's probably the same, but I told someone else. I mean, it depends on when you, I guess you stop running the rules out all the time. If we all let it run, so it's just inclusion, if I think so. Oh, yeah, I guess maybe that's what's going on here. When I let it continue to run. So at that some new space at the point. Yeah, yeah. And you can, if you're interested in there is yeah, you don't have a mouse. There is nodes around. They're at. Okay. Are we good to move on? Yes. Okay. Okay. So where was I got the sizes, right? Okay, so it's better, right? And that we can kind of turn the various nodes and they're all different ones. But they're, we also want to distinguish between, you know, the nodes and the relationship to each other, right? So all this is kind of on the same color scale. But you know, this node here is not necessarily connected to this node here. So let us run some more stats in terms of modularity. And we can represent that in our color screen. So if you go back to statistics and go to modularity or modularity right here. And we're going to run more stats here. Run, worry too much about these problems. Okay, give you modular reports. Yeah, it's not worry too much about that right now. But close. And you have more stats in our day of laboratory. We're going to play with that again by going back to appearance, going back to nodes, this half here, and making sure we have our color palette fixed. And then now we're going to go click on the partition tab, and choose an attribute and choose modularity class. As you give you a list here of different colors. It's automatically assigned you different colors and tweak this if you like. I'm not going to worry about it. Click apply. And then you should see relative modularity representative colors. How did you get colorful like that? So in modularity class, what did you wish? All right, go back to appearance, nodes, make sure the color palette is fixed here. Partition, put that to an attribute and modularity class. Then you click apply. This looks good. Okay, so we still don't have a good idea of who is what and all that information is stored in our CSV file. So what we need to do is play with our labels, right? So if you click on this T down here, it should give you all the names of the different characters in their own nodes. That's very difficult to read, right? So what we can do to make it a bit more readable is click on this down arrow and sign mode. And those should give you a submenu to see that. We want to click on node size. Click that, then it should be a bit more representative of the various labels and their relative performance. Everyone good? And you can, you can tweak this by playing with a slider here. I'm sorry, not this side of the other slider. Adjust the size of various labels to your content. You can change the font and default size and color if you like. Okay, so you got this very complicated network gravel character that there are various interactions. For my purposes, it's still all hard for me to read because of a bunch of tiny characters here that have a few connections that have a clear meaning of it, at least in my case. So I'm going to get rid of some of these filters. We can can start to filter out some of the minor characters that we don't really care about. So if you go to filters, apology, get submenu. It's the degree range. And let's use this filter. You can double click it or just drag it down here in the queries. And it should give you this nice little range here. Okay, that's what. So I'm going to say I don't really care about characters about five degrees of connection. So I'm just going to look for 10, nine degrees. Good. I just said if you'd like, but that's like the scale here. And I'm going to get rid of all the characters have pure the 10 degrees. Click on filter. And you should see, okay, this is a bit more readable. So all the different minor characters that only have 10 degrees of connection. And you can play with that however you like. Maybe you're interested in character that was nine degrees, more than 40 degrees. So let's say something's good art. This is folks okay, filter, label, and not know. And then you want to get rid of all the labels. Keep the big ones like from the smaller ones, I guess. Oh, um, that's a good question. I'll pop my hedge back. I'm sorry. I don't know. Okay. Alright, so let's say you're happy with your job. Now, what we're going to do next is export into a file that we can use. To do that, we go on the preview. And you should give you some progress here. So presets, default, all kinds of different outputs, play around with these, however you like. So I'm just going to go with people heard for now. And my preview window, it's not working. So I'm using it. I'll take back your graphics or problem. But you should see some people already have those people do. You can customize these according to your preferences. Make sure the labels are shown. And that's what we're on here. You can like, decide how thick you want the edges to be, how thin you want the edges to be. Maybe you want to reschedule the weight of the edges. Maybe you want to mess with the opacity of the edges. And you're gonna have to go back and forth and experiment with this, depending on what you want to look like. Once you're happy with those parameters, you can click on export down here, bottom left. And it should give you some options, right? You can export the PDF, a PNG, or an SVG file. So PNG is just like the jpeg. I wouldn't recommend using this unless you're just going to use it for like, web representation, post on the website. SVG files are vector files. So these are going to be a bit better for especially complicated, correct network graphs. This, let's go put pf on, name it whatever you want. And once you're ready, let's say, save one, and once you have saved it, you should be able to click file. So I can already see that it's still a little cluttered. Maybe I want to filter some further. Maybe I want to make the force out list, repulsion strings stronger, so it's more separated. But you can see basically the gist of it, right? You can see on an already at a glance, sprawling novel, maybe you already intuited the structure, but, you know, you see John Snow's precentral terranolaster precentral Targaryens up here, but still important. But, you know, maybe this, what's useful about this that you can give you at a glance, a structure to the novel that maybe reveals some things that are interesting, right? So you're interested in rice, carrot, what a reason can see at a glance just how connected he is to the other characters in the form. What is our young Martell way down here? And again, you can kind of customize what this looks like based on the different criteria of the graph and what you're interested in. They're one good so far. So that's us playing around the prefab data set. The question then becomes, okay, well, how do we do this ourselves? Like, let's say you're building your own data set, how can you make your own, then run it through the interface or run it through. Yep. Let's click out of that. And I'm going to go back to my, hey, I have included in the zip file, a file called workshop underscore network, XLSX, as an Excel spreadsheet. If you point your computer to it, don't click it, it should open up a spreadsheet here with three sheet worksheets. And what I've done is I've taken the liberty of filling in some of the sections. So let's say you're working on a historical demographic, or social media demographic graph. This is a good way to kind of build your data set, right? So the two outputs that we're going to be interested in are nodes again and edges. And this character interaction sheet is just a worksheet, we're not going to actually export it or use it for data interpretation. And let's just, I'm just going to, I've just made up a nodes list here with a bunch of characters from fiction and history. And these are the main characters. Obviously, this is a small data set might be playing with a larger data set. But just for the sake of illustration, these are are the major characters that were actually playing. So we have this column here called ID. And so the column called label. And so obviously, the label of the name for the different people, we also need to number them. And Jeffy uses a zero zero as the first number. So it's kind of it's incredible. But what we need to do is number them. And if we start, we can label these in automatically, even through Excel, and just highlight these, pull it down, it should recognize the pattern that needs for you. So we're going from zero to 10, zero being Jack Flosson, 10 in Frontel's mother. Everybody got everybody there? Okay, so we already have our nodes csv file and the principle. So let's save it. Go to file, save as and you can name this whatever you like. And the workshop that underscore notes, we're not saving it yet. We want to make sure it's saved as a csv file, all my separate values. Matter, which it can, I run into trouble with people using max who have used like a different, like you can see down here, the max csv file, the msdOS csv file, you know, the safe side, I use UTF-8. So I would just be able to save side, choose UTF-8. And once that form is selected, like save, it should give you this error or warning, say it doesn't support workbooks. That's fine. All we're interested in is saving this right now. So okay. And let's go and check. So you have a file here. I tell you csv is plain text. So I can do the check the integrity of the file is, is notepad or if you're using a Mac, what's the Mac? You just need like a plain text. Not be the edit, is it? Or plus plus there's some whatever your plain text file of choice is, please use it on Mac. And Windows is not bad. If you click over, if you drag this file over, you can see again, that it's just plain text, it's separated by comments. So for some reason, I'm getting this error where it's giving you an FD row problem here. I'm just going to delete that. I don't need that. You can say it's fine. Not. We'll fix it. Anyway, just make sure that the file is saved as csv file. And that's going to get a little bit complicated. Okay, so if you go to the edges tab here, you can see a source, targets, type, I'll explain what that means. Second here. Again, what we're trying to do is build a spreadsheet that is to the different kinds of relationships that the different characters have had with each other and represent how important those interactions are. And what we're going to do first is go through it manually. This sort of labor intensive part of building a data set is like, let's say Joan of Arc is a character, I'm going to sit there and catalog every single character of importance that I determined before that Joan of Arc interacted with it. And do the same for block, do the same for static blocks or puzzle, we probably will still do it. But Gefi is not going to be able to read this. Gefi works with stats. So it needs to be able to interpret this in a numeric sense. So what we're going to do is going to, again, for purposes of just building our data set, we're going to copy the ID label here, this row, paste it over here to the column to the right of label. We're going to do here, the right of Jack Frost on this column. We're going to use the VLOOKUP function in Excel, type this into the chat just so that you have it. You don't really have to worry about what websites equals VLOOKUP. For parentheses, A1, and I'll explain what this means, or shop on the score network, score nodes, commission points, C, C, two, comma, false. Again, make copy and paste this into the chat. Okay, what this is doing is basically is telling Excel, look at this particular column, particular cell in this worksheet, worksheet, and set it back out here. Everybody got that? I'm just going to see that. I think my tab workshop network knows that if you need this, it'll be different. So once you type that into Excel, it should spit back a number, right? In this case, five. And where's the number coming from? Go back to the node tab. You'll see the number five. And so if you go back to character interaction, five corresponds to one. So we're going to do the same for all these other entries. And just to make it easy for control copy. And then I'm going to highlight all these as control D, control paste. And it should automatically copy the formula and should adjust on the fly. So that is responding. It's pulling the right number for each character. So because Jonah Marcus five, all the other uses Jonah R should be five. Lawden and Taylor is two. Double check by going to nodes sheet and seeing as Lawden and Taylor is two. And see how they correspond to each other, right? Okay, great. So what we're going to do is we do the same thing, right? That's slightly different. Copy. And it goes to the next column. And we're going to do the same for these characters. They're corresponding without. Now it should give you an error by just copy the formula. And it adjusted on the fly. Let's point to an empty cell. So instead of D one, it should be D one. But this should be that should give you the next case again. And it should give you all the corresponding numbers for all these different characters. Any issues with that? Okay. Okay. So now that we have the characters represented numerically, we're going to copy all this, all these two rows is to call which thing. And we're going to go over to our edges. But if we just press cut and paste control D, it's going to give you a bunch of errors. Because what it's done at that point is copy the formula. And the formula doesn't work because it's pointing to the wrong place. Nice. So that we want to do is copy, paste, copy and paste, but I'm going to right click my mouse, a special and paste values. It's going to spit out the output that results for us before. So we have our nodes, we have our edges, our character interaction, we the worksheet that's kind of a scratch paper that we use to remember that numerically. Now we're going to need to fill the type. And for your sake of simplicity, we're going to go with my directed, undirected meaning that the edge is bidirectional, symmetrical. If you want to direct it and then it's like the connection of the relationship on one side. So is there a reason that in the previous in the unshark that it picks, if it's undirected, this is the picks one color over another, you know, from the middle blue that over the red, and like sometimes it's a blue or white. I think that's based mostly on the degree of separation. If you if we had played with a directed data set, you would actually see heroes. But for the sake of simplicity, again, we're just going to play with undirected all that. Okay. So we're getting close. Now this is where the interpretive part of working with the humanity status. So we had these relationships, we determined they're undirected for now. Now, weight is how important those relationships are. So I actually have no idea how the game of Thrones data set was determined in terms of the weight. So and then subject it. So if you read the book inside that terran's relationship connection to the one minor character, it's really important because he tells them a secret. You put 10, you put 100, you just represent that numerically. If it's consistent that you build your own sort of kind of legend for what that means. And, you know, it could mean anything, right? Connection the thought about the other character, or they were in the same room with the other character, that they sent them a message with a Raven. Because I mean, they fought each other, that they killed each other, or damaged each other, that's up to you. Take your respective disciplines that might mean different things, right? So you have to build your own legend for what that means. And consistent with otherwise, the data is not really useful. So if I looked at Jack Frost's relationship, the interaction with Joan of Arc is one time, I thought, okay, it's pretty minor, because they just kind of walked each other as I'm doing it on one. But other interaction with Joan of Arc and Grendel's mother is really important, because I'm going to go with reptile test. They talk about something other than math. And you just go on, right? So I'm just going to put random numbers now. We had a question about this undirected directed, if you did put them all in as directed, how would the table change? Or the graph change? So it would be C source and target. So if it was a director relationship, it would be coming from this source this target. But if you want to represent the other way around, you'd have to put the numbers. So I'm just going to put random numbers, just for the sake of exercise. So I got my numbers. You got my edges, I got my source, my target type of connection weights. Now I need to save this file, the file and save ads. And make sure it's not this is named differently that it is the edges, right? The same warning again, fine. And now I should have saved that file from my two CSV files, notes and edges. So double check my edges, pages file, find out I'm going to shut down Excel if I don't need it anymore. I already have my editing notes file. Let me double check my files. Yeah, that's fine. Nodes. No, it's safe. The nodes were long in it. It's my bad. Yeah, just ignore me when I do this. So now that we have our file, we just want to do is plug it in the same with the game of Thrones. So let's go to the project and say this to like, I click on the project myself, go back to overview. And start by reporting your spreadsheet, right? Sports spreadsheet. And again, you want to start with nodes. That's it. Exit nodes. Yeah, eight issues are found. Make sure it's undirected. In the workspace, click Okay, give you the same thing as these nodes that don't really have relationship. I'm going to add your edges spreadsheet. The same thing for the edges. Open the text as edges table rates shows you the weights, different columns. Next, let's find finish. Again, make sure they're pinned to an existing workspace, a new workspace. So now again, see our graph here and we see the relationships in representation. And then in run through different layouts. Is everybody following the part? Is that you slow down or repeat myself? I'm having it up here. Okay, this table needs a source and target column with nodes like these. But in the preview, I have source and target. Also, I don't have any nodes this that you're already imported your notes. And Jeff has the same error if you want to open up your nodes. Jeff has the same problem. I wasn't sure if you wanted to like say it louder for Jeff. That's and then we could help people. Okay. I just in case we lose people at two. I'll shoot later. But if you remember, now that we have the basic structure, it's really just a matter of aesthetics at this point, your own criteria. So if you go to layout, force atlas, and you can use a different 15,000 for propulsion strength, run that. That's not good. It's too restricted. I change this again to another zero, top this algorithm. Right. And then if anybody remember what was next, terms of representation. Sure. I mean, there's no real order. I guess both labels, right? And it shows you the different characters and their notes, connections, right? Click on side control and note sides. All the same side right now. Let's just change the note colors, right? Under occurrence, notes, color palette, select it, right? Same thing. So either different gradients agree, represents the open report for the node. And then we can run the same statistics again, collide with the or diameter modularity. I mean, this is a pretty simple graph. So I'm not sure. So really, you know, that of course, go on that. Here's the notes, size, right? attribute. See, that's way too big. I think we're running out of time. So I'm just gonna stop it there. But you can see like, keep the same steps over again. And you can represent your graph, receive it, and export it when you're ready. I think so at this point, we can play around further. If you're good, then try tinkering with your graph. If you're having problems, I'll try to troubleshoot with you. Thanks, everybody. Thanks, David. And I said this online, but our next workshop is on October 5. We're going to do Zotero, which is just bibliographic management and a tool for that. And it sounds like David will be around for two minutes. I feel like we should clap. People over distance. How is the fancy new web conference? Yeah, I have all of our all of our fall events here with the next one being Zotero or anyone. Or if you think your grad students hate it. It's hard to see people because we're sharing my screen. Is there a way to make the web? So you know, when you hover like a note, it'll show who that note is connected to. Or is it only exported PDF image vector base? Like I know it's probably super complicated. I just don't know if this is possible. I think I've seen the static.