 Good morning everyone, thank you for being here and thank you for the FOSDEM team to organize all of this. Really nice event, really happy to be here. To talk to you about CARTIS, which is a software we just released. It's a web application to simply create thematic maps in three steps. It's actually a tool we designed for students. I'll tell you more about this. I am Paul Girard, I am from Media Lab Research Lab in Paris from Sciences Po. And I'm here presenting a work that actually some colleagues worked more than me on Bajama and Audrey who couldn't come today. So I just represent them. Our lab is Social Sciences Research Lab, in which we actually try to design what we call digital research methods. Which means that we are trying to give researchers in social sciences new ways to do research through the means of different digital tools. And then because we are doing that and because we do science and we believe in open science, everything we do are free and open source software. You can check out our Github accounts in which we have many more projects. So in these projects we worked closely with another team from Sciences Po called Atelier de Cartographie. Who are basically cartographers, who are really specialists into graphic semilogies. So they also are doing information visualizations, not only maps, geographical maps, but also data visualizations. And they are doing that for teaching, to support teaching activities, research activities, editions of books, or even for some museums. You can follow those guys here on the names. If you click on the icon on the website, well actually my slides are online in this URL. So thematic maps are maps which looks like that. Basically the kind of maps you might have stumbled upon during your teaching periods when if you open a geographical history book you will find these kind of maps. So in this map for instance it represents the world's carbon dioxide emissions in 2013. With two different visual forms. The first one is the color of the area of the country representing the perhabitant emissions where the cycle forms represent the total amount of emissions. If you want to see more of their work from Atelier de Cartographie, they have a Cartotech which is a website where they actually publish all their predictions online on a Creative Commons license. So you can find more than that. So the story starts when actually these guys from Atelier de Cartographie want a grounds from our university organizations to design and develop a pedagogical mapping tool. So the whole idea here was to provide students with a tool to do these kind of thematic maps. So we started tripartite collaborations where we have teaching cartographers. So they are at the same time really doing cartographies and maps at a high level of symbology and also teachers. We were more likely the digital methods specialist and also people used to drive free and opposite software projects on the IJAL methods with the IJAL methods as a mean. And then I hired a guy called Arnaud Petzel from a company called Apix with a web application developer specializing in cartography, cartographical tool. So I have to really say thank you and bravo to Arnaud who actually did all the hard work of coding everything you are going to see here. And he really made a really great job. So if you need a good developer in this area you could check his website. So our first objective is to make things easily as easily as possible. So we want our tool to be accessible for newbies. When I say that it's like more likely like licensed students, students two years after baccalaureate in France. We want to create a map of course. For that we have to make sure to have a mean to get the data for the map, to upload data. And then we want to be able to choose with the map projection we are going to use. And we want this map to convey a method. So we invest a lot into means to actually choose and tune visual forms to map the data on the picture. So to sum up we have basically three steps in this tool. We are going to show it after a point. First one is to upload data of course then to choose the map projection and finally to add visual forms. In the uploading data step what we, the features we've developed are data type recognition. So you just input a CSV and then try to understand whether that number columns or latitude column, longitude column, geographic names columns are any kind of simple qualitative values like strings. And we also the tool propose geographical names alignments. So we are trying to recognize from a column with geographic names which one we have in our base map. Then we ask the user to choose the map projection and to, so we really want this tool to convey pedagogical also information. That's why we tried what the cartographers proposed three different features to guide the user through their choice of map projection. The first one is ARIA. So those are projections respect the areas. The second one is distance and the second one is angles. So actually this is all the map projection we, the tool propose so far. So we have, well you know that I'm sure. And so what they did is actually they decided to qualitatively decide which one were the good ones to, if you want to respect the areas or as a distance or the angle. And then it leads to people just to remember that the Mercatar projections which is most commonly used because it's squared, it's quite convenient. It's only good for angle but really bad for distance and areas. That's very one important point for my colleagues from Atelier Cartography. But you can still use it if you want. And the last step is to tune the visual forms. So here basically the idea is first to pick a variable from the data. You can have many different columns in your CSV. You pick one and then you say like well what I want to do with that. I want to add the symbols on my map. I want to draw colors on areas. And then you can tune the colors, the size. You can define classes. You'll see that how to map the variations into different classes using means and all the kind of statistical treatments. And then finally you have your map. You add the title. You add the author. You add the source. You tune the legend. And here you are. You have your CSV. CSVG. Sorry. R-U-P-N-G. All right. So remember that map. It's the one they've done. I've picked up as an example. The first thing we're going to do now is to do live mapping exercise. So I find the data sets they've used to do the first one. So this map has been done basically in Illustrator, I think. I should have asked my colleagues about that. But I think it's mainly Illustrator. So now we're going to do it with Cartis using the same data. And here we are. Okay. So this is the data sets. So it's basically total dioxide emissions. So I'm just going to filter out. Thanks to LibreOffice. The year we want to map. So let's go. Let's take that one. One more there. Yeah. So I'll just copy. And let's go to Cartis. She's not done that. All right. So here is Cartis. The first thing I will do is to change to English for you guys. So the first thing the tool proposes you is to select a map. So here I'm going to work on the world countries the 2016 time. It proposes you many different ways to describe countries. So we have labels in different languages. We also have like easel codes. But you don't have to worry about this. If you need those to prepare your data, you can download this model to do your cleaning. So the cleaning has to be done preferably before. But if you just upload your CSV. So I have it here. And here you have the colon recognitions, data type recognition I was talking before. And as you can see, what you can see here for instance, and the year has been understood as a latitude information. It's a numeric one. And then here you can see that it doesn't succeed into casting into country names for 24 cases on 220. And then here we have a little guide to let the users actually help the algorithm while not the alignment process manually. So we try to do everything in an automatic way. And if it's not then, well here for instance, we say in French it's not recognized because we want the end in four letters. And this is true for everything. It's like almost there. Actually we are thinking into adding kind of a fuzzy recognition mechanism here. We have the technology. We just have to put the button here. But you know, we haven't done that so far. So this is important because if you don't know that, you will not recognize China Mainland as China. So you will have like missing points in your map and you don't want that, right? So this is, yeah. So I'll switch to the point where I've done that. OK? It's done. Right. The Montserrat, I couldn't find Montserrat. So it means that you have more data in my files than I have on my base map. OK. I can drop one. I'm fine with that. The rest is pretty cool. So we can go into the visualization. And this is where you close your eyes because I want to do that with you. All right. So I have my base map here. So it's for now empty. Here I can go back to choose different kind of base map if I want. I have to click here. You can also tune the base map if you need to, you know, for instance here, you can change the longitude, right? And then this is where the plus button brings you to the second step, which is basically to create visual forms. So the first thing I'm going to do is to take the total, no, I'm going to take the per capita CO2 emissions and then I'll use. And this is where the tools propose you to decide which forms, visual forms. Oh, sorry, I'm French. Sorry about that. Yeah, better. So here you can choose different visual forms. So you have two ways. The first thing you have to choose is whether you want to consider these variables as values or as categories. Quantitative variables, qualitative variables. And then the second choice is do you want to use symbols or areas. So here I'm going to use areas, continuous scale. And here we are. Here we have the distributions of these values and you can choose different ways to discretize your values, regular interval, nested means, and quantiles. You can choose the number of classes you want. So just to show you that that actually works, right, you can change everything. You can also choose the color patterns. So we pre-compute a lot of different color patterns to propose the user. So if I want to do something similar to the first one we had, I'm going to use a scale of red. Yep. Good. Or you can use hatching if you want. You remember this time where we are using hatching? You can do it. You can reverse the scale if you like it better. Right. You can also see your data aggregated if you need. And here we are. We have our first visual forms, which is areas colors. And then I'm going to speed up because I'm late. Then I'm adding a new visual, a new variable to the total emissions. And now I'm going to use symbols. And then for the symbols, I'll actually use circles. And then I'll make them empty. No, sorry. I'm using the alpha, right? And then I'm just using the scale, using the USA as my maximum, I think. Maybe not. And here we are. Let's say we're good. Right. OK. Ta-da. So now we have our map, our visual forms. We have a legend. That's pretty cool. And we just, OK, I added the titles, the source of data, and the author here. You can see that they are added on the edge of the map. And then from here, you can decide to put the legend or to discard it, to add borders to do, or you can put the verticals back if you need. All that kind of things. You can choose a resolution and then bam, you can download. OK. So, sorry about that. Here we are. OK. So we had, this is our results. And so in a few minutes, I tried to do the same thing that we had in Illustrator, right? With Cardis, like that. OK. To finish up, another kind, quite of nice features that actually took a few times to do from our note to develop this kind of base map where you actually have this continuity in the geospatial areas. So we have the France Metropolis 10. And then we have Departement d'Outremer, which are like islands in Caraïbes and in Lille de la Réunion. And so here, I used geolocalized data sets and the geolocalization works on different areas, although there is a discontinuity between those. Doing that was not exactly easy because it was not planned at the beginning, but we haven't thought about that. And so the whole engine of putting the points on the map needed this special multiple geospatial areas. Under the hood, we have a full web client application. It's umber.js, so JavaScripts. We use D3 for the projection and many different visualization utilities. You can fork that on GitHub, of course. And because it's full web, your data stays on your computer. So we don't have any server features. It's only web, full web, which means that it was quite easy to provide with an electron version for offline use, which means that you can also download a desktop application of Cardis and use it on your computer without internet. There's actually a talk about electron layers this day in another room. And that's it. Here, Cardis. Any questions? Yeah. A more legal question than a technical question is when you get that out, how do you know if you can reuse it to clear it? Okay, so I'm not a legal person. So your question is, are you allowed to reuse data to create a map? I think for what I know is, the first thing the data set I'm using here was, I think, a DOI. So it's really to publish, so there were no problems with that. But I think a map can be understood as a statistical rework on data. So you're not, if you publish a map without the data sets, you're not publishing the data, so you're not need to have the right to publish it because you're publishing your work of interpretation of the data. So this is a different point. And the second question was? That was good. I had a second question that you asked. Okay. Yeah. How do you do if you have lots of different data? Maybe you want to show natural resources like say, I want to map oil, go to gold, silver, uranium and whatever. Yeah. A lot of different symbols. Okay. So two things about that. First is this tool is to create a thematic map. So your map is not an exploratory data analysis tool. If you want to do that, use a GIS. It's good for you. You will use that once. You know what you want to tell your readers. And your readers will not support like too many information in one map. So you have to choose for them how you're going to tell your stories with maps. Which means that you might first put like two, three, five layers of symbols. You can do that in Cartis. Actually in the visualization here, this is one layer. This is another layer. You can add many layers if you need. But I mean, more than three is going to be a mess. And then you can create multiple maps. And then you're going to build a map, basically. So it's really like, those are your producing documents. People can read. And you have to make sure what you're doing is going to be easily readable by the people. So you cannot put too much information. Yeah. Okay. So no, not yet. Probably not that soon. Because our targeted audience, oh yeah, the question was, can you build your own base map in Cartis? So technically you could. But we are not going to put money into the user interface to do so. Because our targeting audience are students. Which doesn't know how to do that. But I mean, the whole system is easily manageable by our team of cartographers to add base maps. So the process to do that, we'll contact them, pop up the base maps. They will tell you the guidelines of which kind of base maps they can handle and then put it back in a source code. And you can also fork the source code, learn how to do it and do it on your own. But there will not be user interface to do that. Yeah, I think so. I hope. But this is a joint collaboration. So I'm speaking on behalf of my teams of cartographers. I don't know exactly what their editorial directions, which direction they want to go. But I'm pretty sure they will accept contributions like new base maps. Yeah. And can you restrict the map that you're using to a certain area of the world? Yeah, of course. Look, it's as easy as... I don't want to see Asia for instance. Can I just say Asia and then it figures out Asia by itself or do I need to do it? So you have two ways to do it. You can do it like this. And it will export only the viewable part component. Or you can once, when you choose a base map, we have base map on the specific countries as France as I showed in my slides. You might have a base map for Asia. Okay, thank you.