 Hi. Hola. Thank you for being here so late after all these things that we learned and shared during these days. OK, my name is Shani. I did this work with Sandro, who sent cheers from Brazil. So he's enjoying. This phrase is well-known in Spanish, in Argentina. And we use it to refer the group of people we join to do things that we enjoy together, like the Zoom in team, the book club, the CSBConf organized team. And the idea is to talk today about collaboration in a special kind of group that we call community of practice. A community of practice is a group of people who are passionate about something. They know how to do it. And they get together to learn how to do it better. And why I'm talking about this? Because June last year, I become the community manager of a community of practice that is called our open site. Our passion, open and reproducible research to everyone built by everyone. We know how to do that. And we get together to learn how to do it better, building technical and social infrastructure. What is that, Shani? Well, one of the things is our packages to help scientists to do science. Research software engineers call packages, send those to our review process. We review them. We help them to get better. And then they become part of our suite. Developers have the support of the community and the staff of our open site. And users get high quality software to do science. We also try to make the way data tools and practices more discoverable. We have a project that is called our universe that helped to publish and search more than 18,000 packages, our packages. We do our biggest effort to create a welcome and diverse community. We have several projects for doing this. For example, our champions program that is focused in region like Latin America to help people to become contributors to open source software and our multilingual publishing project, which is trying to localize and translate a lot of our materials and documentation into Spanish and other languages. And we try to build capacity for software users and developers using, for example, all like books that we work, everything open, guidelines, templates, software, and different space. Everyone, the people who do that is pride about the work we do. And we promote advocacy for a culture of data sharing and reusable software. Everything we do is open. So we hope that it's useful for people to use. So now I say that I become the community manager. Again, what is that? Here, most of us think about the person who handled Twitter or Instagram. But it's not exactly that. A community manager in the context of a community of practice is someone who facilitates the activity of the members of the community. So it's more like an enriched activity, more than this social aspect. We have to deal with Twitter and Instagram and MasterDone and all that, but it's only one part of the work. We do program management, program design. We have interpersonal iterations, communication, technical aspect. So yeah, that's tried to be me. And the thing is that when I start this job, I say, OK, how am I going to do my work as a community manager? How do I do a good work? The first thing is to know your community. So let's analyze our open side community. Let's do this together. You that only are going to be a better community manager, if you know your community, you probably are going to be a better member too. So it's really nice to know part of what groups you are. At the open side, we register a lot of data about what we do, how many blog posts we write, how many community calls we run, how many people attend the community. And that is very useful. I give us an overall idea about our open side. But communities are built on connections. It's people doing something funny together, something nice together. And that kind of data don't give us insight about the pro and the connectivity inside our group. So we want to know our community connectivity. And why? Again, Chani, come on. Why are you doing this? Because we want to have effective interventions to improve collaboration, information flow, knowledge reuse, knowledge co-creation, knowledge transfers, who is building what and who is using what and who should be working with who. We try to answer questions at a given moment in time and all over time. How do things change? Who is connected to whom? Who is not connected? Where and who are the hubs? The people who brings together other people? When about the cluster? The people who work together a lot? Where are new connections forming? New patterns of connectivity? How does our network before and after an intervention? Before and after our champion program? Well, what kind of data we need for that? We need some data that show relationships, that show people working together. And there is where social networks come into play. And again, I'm not talking here about Twitter or Instagram, but I'm talking about networks. Networks that have nodes, in this case people or members, have an edge, something that they do together, the link, for example, write a blog post together, go out to ship. And sometimes that can have a direction. For example, someone who attend a community call, their direction will be in one direction. And perhaps not to the other, but these people receive information and that's it. And we can map networks. We can have all our members here, the little people, and we are going to have things that they do together. And we can start to explore this pattern and to start to identify interesting things in the network. So for example, the degree is the number of connection that a person has. So higher degree, these people do a lot of things with a lot of people in the network. The multiplexity is when two members do a lot of things together. So they do things together more than once. I don't know, they write a blog post, they translate a chapter. The between the centrality, it is the people who are always in the path to connect to other people. This is the people we really don't want to lose from our network. This is the people that it is important to keep because otherwise we can split the network if we lose them. And the closeness is the people who are close to the other members. The last step you need to get to the full network you have. If you want your community to know something, you have to be sure that these people get the message because they will spread the information through your community. Okay, keep this in mind. And the cluster, so the people who work together a lot. So this can be by topic, this can be by geography, this can be, it is really interesting to see who is working with who a lot. Keep this in mind to see how we apply this to an example. How we can collect those data. How I know if people do things together. Well, one of the things we have at our open site is a contribution guy. Okay, Shani, this is amazing. It wants to be part of your community. What I can do? Well, we have a full book that tell you the different things that you can do. So for example, you can write a blog post, you can review a package, you can maintain a package, you can speak at a community call, you can be a champion, or you can host a co-working session. A lot of these things can be done by more than one person. So you can start to identify the nodes and the edge. Why a blog post? Node, author, edge, culture, ship. Champion, node, mentor, mentee, edge, mentorship. And so forth and so on. Where is that data? In a webpage, in GitHub, in our universe. Let's see an example with blog posts. We have our blog posts using block down, this is a young, the head of the file. And you can see the title, the author. So, hey, here's the people who are working together. These people write something. And the date, when they do this. And here is the code in our, for reading all the blog posts we wrote and get that data. So we create the list. We create a table with the information we need in Spanglish, you know. We look, we look through this list. We get each one of the document. We read, okay, okay. We read the young. We put the data in the table and then we create a CSV. Then we transform this information in, this is the format that a network need. We need a form and a two. And one that we can do that with that code that is there, and then we can plot the network like this. So this is our full network of people who write blog posts. The size of the bubble is how many articles they write. And the edge is with who. We can analyze this by time. The network in 2014, the network in 2022. That's nice, this grow up. We can add the names of the people. So we can start to identify. And I can calculate these things that I said at the beginning. For example, I know that the higher degree in this network is Noam Ross. I know he's a staff. But if I looking for the higher degree of someone who is not getting paid by our open site, I know that is Laura de Chico. So I can start to know my actors. And I can see all contributions together. You see all these table, write blog posts, maintain a package, everything. That is our network today. And then I can start to see the clusters and identify. So for example, the cluster, there is our champion field cohort. So if I saw any of you there, I will go back to the title. I will know because you have now with these people that you have anything to do with the champion program. And the same with, for example, with our university star interview team. So I can start to know who work with who and who I need to try to connect with who. To connect with who. So what if you want to do the same for your community? My tips. Define the notes in your network. Can be not only people, can be countries. We map this with countries. So we know each one of our member where they live. So we can see where we have a presence. It can be organizations with the affiliation of the people. You can map organizations and you can see with who are you actually working. And with who not, which is pretty interesting. Define the type of connection that you have in your network. Start with your path for contributions. Identify which one can be done in teams. And probably you already registered information about those type of connection. You can automatize a portion, yeah, like this code in R. Try to formalize the work so you can repeat and reproduce. It's hard to capture all type of integration. You are going to miss something. And there is some collaboration that has some open close privacy of the data that you don't want to put on the network. Knowing the know helps you to understand the cluster. I mean, I know that people are champions. You probably don't. So that knowledge that you have about the network help you also to understand. And in the case you are new, like me in the community, as the people that have been on the network very more time than you. You can take a snapshot of the model of the network so you can compare different times. I want us to know if our champions program will change at network. I hope it does. And if it's changed, it changed in the way we really want to. And share what you find with the community. Share it. And with other community managers. Thank you so much. That was excellent. Thank you very much for that talk and for that analysis. Very interesting. We have time for questions. We have one in the back, hold on. I'm so excited about this talk. This is so amazing. I've never seen a data analyzed for the community this way. And I can't wait to do that for my communities. My question, I mean, of course, I can admire you for like five minutes, but my question is how do you actually... So there are people who exist in multiple communities, right? And those multiple communities also impact how they interact within a community. Would you consider actually adding neighboring communities in your things? Because I can see that there are lots of overlap in different communities we work in. Thank you, Malvika. Yes, I was actually talking with people at Bioconductor, which is another community that work with R&P packages and the carpentries to try to replicate these for them and then see how that works. One of the things is, if you can imagine, we need to identify each member like a unique person. So we are using names and emails, trying to do that to say, okay, this Shani, it is the same Shani that is here and it is the same Shani that is here. And we now are using only public data so to take care of our members and asking for permission when we show things like this. But I think that could be awesome to see that. Yes, I also agree that we have a lot of overlapping and common interests. Yes, next step. That's great. I love that slide, really, really nice. So what's like the most surprising thing you realized about your community that you didn't know or you didn't suspect perhaps before? Okay, our open site for several years has an uncomf. I don't have the data of the uncomf here, but the clusters appears anyway. Those things that you see on pink and green and purple are people who met on the uncomf and they start to work together. So even without the data, you can see some kind of impact or result of that activity on the network. That was a plus for me. I didn't expect that. And because I wasn't there when that happened, people that was on the network was the one who told me, oh, this is the uncomf in Australia. Yeah, like that. So we give sense to that group of people working together. And the other thing is, I was doing some analysis of content in other language than English. And we need to increase that, a lot. Yeah, and the other thing is that all the higher contribution, the higher degree and everything is stuff. So that also shows the importance to have funding, to have people who can put time, pay time on the development and sustainability of the community. That wasn't a surprise, but it was nice to be able to see it with data. I've got a question. How long did it take you to do this? And can you share your code so other people can do it quick or faster? Yes, we are going to create the repo with all the call after we check for some quality. But yeah, yeah, actually I'm trying to create functions. All these, what I show here, I have like 13 copies of the same. So I'm trying to create that as a function. So it's more tidy. And how much time was to, it's more than coding work is to the concept of the north and the edge and to try to identify where we have the information. That take like, let's say if I will do all together a week of work. And then the coding takes more time, perhaps two weeks. And then the analysis is not finished because this is kind of the first scratch, the easy things that you can see. So yeah, thank you so much. Can we get another round of applause?