 So, good morning everyone, I'm Avis Serrano, I work as an assistant researcher in the Universidad de la Complutencia de Madrid. We are currently researching on wikis, on communities. We have developed a wiki ground, it's a work tool to visualize collaboration in wikis. It's been my work for the last seven months, it's the first time I present it in public, so I hope to have you like it. So, first of all, what is wikiran for? Wikiran is a web application that helps you to visualize metrics, data of your wikis, the history data. And it facilitates the analysis and research on the evolution of wikis or a set of wikis. So, let's go in, it is at wikicron.science, it's online. And here we see a lot of wikis, they are mostly from wikia, it's a hosting service. But all media wikis are supported, I have added this brand to it, as an example, I think it's over here. And, first of all, we need some wikis to select, they have been processed and uploaded. So, we are going to select two wikis, the legal French and legal Dutch wikis. They are both big wikis, very active wikis and they have a large community. And they are about the same topic, right? So, we might think that they have a similar behavior and similar numbers. We want to compare their activity, so we have to go to metrics and select some metrics of articles. We are going to focus on content, because we have pages and content pages are articles. And edits, so that it's in articles and press compare. So, here we have some graphs, they are time series. And they start from the creation of a wiki. So, if we see this line, we have this data, this number of articles from the 13 month since the wiki was created. So, the Dutch legal wiki started with more articles, it has 165 articles in 30 months since the creation of the wiki. And French wiki is going a little bit below in the same time frame, right? So, this is the 30 months since the creation of French wiki. And it has only 56 articles, which is a third time less. But as we see here in these graphs, we can see the evolution of these wikis are differing, they are changing through time. And the French wiki is going with us deeper, upwards growth. So, we see that since I think it's month 44, month 43, they have a different growth. So, the French wiki Lego wiki is going more productive, it has more articles, it has more edits. And we can see these months, from 40 months until 100 months, it's becoming a much more productive wiki. They are generating more content. So, we might think that the French people are more people or maybe they like more Lego game. So, we might think that there are more users deep in the wiki or more active users. And then we have to select some metrics about users. So, I'll go to users, total users. Ok, maybe you are trying to visualize a large wiki and it takes some time if you are using it at the same time. So again, it's users, total users. Ok, so as I said, maybe there are more number of French wikis, wikipedia, wiki users. But it is not the case because there are much more French vets, wikipedia, wiki users. And also more registered users and more anonymous users. So, it is not that there are more users that they write in French, but there are less users that are editing that are making more content. So, they are more productive. I'll select 18 articles. Here we will have much more 18 articles in the French wiki. But the number of users in these months is going up also in both. And it's more or less the same slope for both wikis. So, to visualize the productivity of the wikis of the wiki users, we have added some metrics about distribution of work or participation, we say. And they are belonging to this list. First, we have the unique coefficient. It is a well-known coefficient to measure the distribution of wealth in a society. And it's also used to measure the distribution of work in a community, like these communities, like wikis. And we are going to select that. And also there is a 10 to 90 ratio. The 10 to 90 ratio is a ratio which divides the amount of work of the top 10 percent contributors by the 90 percent contributors. So, it's a number who measures the difference between the top 10 percent, what the top contributors make against the other 90 percent. So, if we press compare, it takes some time because we have to calculate all these metrics. So, let's remove this. And we'll have the unique coefficient and 10 to 90 ratio. The unique coefficient goes from zero to one. And when it's close to zero, it means that the committee is perfectly equal, like everybody has the same amount of work. And if it's close to one, it means that there are less people, doing the most of the work. So, here we see that the French community is becoming much more unequal. So, less people are doing close to one. Less people are doing most of the work. They are very unequal community. They have more work than the other less work contributors. And, well, it's clear if we see this metric 10 to 90 ratio, because it means that we have 120 times more work made by the top 10 percent contributors than the other 90 contributors, remaining 90 percent. Both are unequal wikis, but the French one has less users doing much more work. They are more productive. And we can conclude with this analysis that French wiki has less users, but they are more involved and they are doing much more work. But also the French wiki has an equal distribution. So, the problem of these graphs is that they can be misleading, because we have a scale that fits to the bigger one. So, here we have this axis, but this looks like flat almost, right? So, we want to see the difference. Here we see that actually the wiki goes down at some point. So, it means that at some point there are less concentration of work, or maybe there is less work. Because if we again see this graph, it's in articles, we see that more or less at the same point it goes down. The number of pages in articles goes down as well. So, it looks like these wikis need some people very involved in doing a lot of work than the other users that maybe they still are contributing are not that productive. So, let's go back to the presentation. The main features of wikigron is to have a web tool and supports comparing several wikis at a time, but we haven't seen any other tool that allows to visualize in the same graphs and in the same different metrics and selection of wikis in one screen all the data. And we want it to allow any media wiki, also wiki media projects. And we have added a lot of metrics and we are still planning to add more. So, talking about the commit filters, we want to add a lot of metrics, a lot of wikis, and we need to have a new UI, a new design to be able to select search and filter a lot of them. And we also want to add URL with parameters so people can share their analysis. And we want to be able to have a feature that people can just enter the link and the tool will download the data, process the data, and add to the tool directly from the web tool. And finally, we want to add some metrics about single wikis, like when you want to only analyze one wiki, the evolution of one wiki, we want to have metrics only for one wiki. So, the conclusion is that wiki ground can help to visualize data, to visualize the evolution of the wikis, but it doesn't provide you an answer of what that data means, because data doesn't explain itself, right? So, you still need to go deeper in the community, in the activity, what is the reason of what happened, maybe it's a fun movie wiki and it has ended the movies, or it might depend, right? So, it doesn't give you an answer, but it can help you to find questions or to compare with the different evolution. But hopefully, we want one day to have a set of variables, metrics and numbers that can tell you how is your community going, how is the health of the wiki, but it's at least far right now. So, we are working in the University of Complutencia of Madrid within the ASIA research group. We are developing wiki ground, it's fully open source, it's in eHub, you have documentation source code there, and the address is here, wiki ground to science, whether you can mail me or just, we can talk right now here. Well, thank you for listening, and if you have any questions. Hi, I'm Aaron from the wiki media foundation. So, I'm curious, like when you look at this tool, what is it for your research group? Is it like a novel technology or something like that? I guess I'm asking because so in the wiki media space we have a thing called wiki stats, which regretfully you can't just upload a wiki too, but it can handle a much bigger wiki and it has many more statistics, but can do comparisons and that sort of stuff. So, I'm guessing like how do you see this technology positioned from a research perspective? Okay, right now we are focusing on wiki wikis just because they are a very wide variety of wikis and a very wide variety of topics. So, wiki media projects are kind of big, just as full, but we don't know if we can... I said some of them, like wikiversity totally didn't work and wictionaries are kind of medium and wiki sources picking up steam in some places and not in others. Yeah, we were looking closer to wiki just because we have a very wide variety and we want to know if... We know there has been some research on wiki pedia or some wiki pedia projects, but we want to know if we can generalize the conclusions to all kind of wikis. I have seen some of the stats wiki media provides and I think they are focusing on what the wiki media foundation needs, more new commerce or more contributors or stability. We are actually more focusing on the sustainability, the growth of a community, but not depending on if it's a dictionary or it's more like, okay, if people come here and have free time, what they do. I mean, I think at some point it has to come first at some point. But right now we are focusing more on a very wide variety of wikis and metrics more on distribution of work rather than new contributors, how long they spend. At some point it's going to... I would like to have a dashboard with indicators and to work that for any wiki. Any other question? Yeah, I thought it was really helpful to include the inequality metrics too and like Erin, I was thinking that would be a nice thing to suggest for wiki stats. But I was wondering, did you include some kind of threshold for eliminating people who aren't active or is that just all users? Okay, yeah. We have removed all the bot activity because we think that can make noise because we want also only human activity, right? But we haven't thought to remove the activity of... You mean like less... Or maybe people who we just created are going to account 10 years ago and so what? That's what I was thinking because wiki stats also has active... Well, not just wiki stats, but everything I've seen that the statistics on wikis has some kind of threshold for active users that isn't really standardized, which I think is interesting. There are different thresholds which make interpreting difficult. We don't have a threshold for calculating the distribution of our metrics, but it's like a minimal number of users to make sense of this metrics, but it's an interesting point. We also know that these metrics are very aggregated. If you have a genetic coefficient from the last 20 years, maybe the man who started and made the first 10,000 wiki pages, it will have a very big coefficient, but maybe it's not the real scenario from the last year, right? We're thinking to have three-month statistics only this three-month distribution, but it's very interesting that the users... It's also complex to know because how do you know it's an active user? Is it because Has made one edit in the last month? I don't know, maybe. Very interesting. Okay. Any other questions? Yeah. I'm Nicolas from France. How do we install... I think France is a very great tool. How do we install this tool? It's a Python web tool. I mean, it doesn't mean Python, but it has a web interface. So you install it locally, and then you can use it locally, or you can use it with a domain, with an external port. That explains it in the repo. In GitHub, it's explained... Yes. I hope it's clear enough because I don't know more people than installing it and developing it. If you have any problem, just open an issue or mail us. I mean, we want also people to see... I mean, work around, talk around. I will try. Thank you. Yes, so... Sorry, I was just realizing that I think it would be really interesting to know what, say, the people who are managing WikiaWikis or who are taking on that role of community manager in a wiki. When they look at the stats inside of Wikicron, what do they see? For example, when there's a shift in the gene coefficient, can they tell you a story? Does that help them think about some things that are to do with their community health? Have you done any user studies? Had people look at this and see what they thought? We are using it for research right now. But, yes, again, at some point it would be useful to have also a user perspective on what they want to look for, what they want to find. So, yeah, that's interesting because, you know, we are using Wikis. We have investigated some people who want to see their own wiki has to download and process the data. That's also a barrier that has to cross. At some point, again, one interesting feature is to have the ability to just write the URL and do everything for you. Thank you because we have to do this as well. This is just based on dumps, right? There's no actual real-time component. You have to keep uploading new dumps, maybe? Yes. It's more about the history dump than real-time analysis. Again, at some point, if we have a dashboard, we would like to have an extension of the mediability and get data into it. Not at this moment. Okay. Thank you.