 There's two speakers. I'll be presenting to you the first part of our talk and Davy will present the next one. We've titled our talk A Funumentary with one big reason. We want you to have fun in the next coming half an hour. So we're going to present, we're going to speak a little bit, we're going to present a little movie, speak a little bit, we're going to present a little movie. So, let's go. Davy and I, we work at Cinematik. Some of you know, some of you don't. Cinematik is the Belgian Royal Movie Archive actually. It contains movies that are more than a hundred years old from the first Lumière movies from 1890 up until a movie that was made yesterday. So, I will talk to you about the presence of what we're doing there and the technical stuff we're doing there. And Davy will talk about the future, which is very interesting. First of all, I'm the technical guy, Davy is more the political guy. Who are we? Who am I? Maybe. So, here we go. A company called The Republic and Cinematik asked us a little while ago to help them, technically. So, that's what we do. Actually, we started working internally there. So, actually, we're part of becoming part of Cinematik. But first and foremost, I don't know who of you, a little poll questioner maybe, who of you do know the Cinematik? So, let's say 50-50. For the other half, I don't know if you have an idea of what the film archive looks like or what it contains. I brought you a little sample to entertain you again. It's all movies, but it's also new movies. It's movies that all together consist of three petabytes of data, which is a lot. And I'm not speaking about the analog stuff. This is only the digitized stuff. So, we're talking about a lot of movies. Now, if we're talking about a film archive, what does it look like from the inside? You probably think, for those of you who've never been there, that it more or less looks a bit like this. Which is normal, of course. But for those of you who have been there, actually, it looks more a little bit like this. So, just to say that movie in the year 2018 is not analog anymore for starters. It's not a small file anymore. It's becoming bigger and bigger. Even scan resolutions are becoming bigger and bigger. So, we do, Cinematik, do contains films that date back to 1890. But we're in 2018, which brings us to the question, OK, we've been around for, let's say, 100 years. How do we go forward? How do we jump into the future? And how do we go to 2100? Actually, this is the main question that people at Cinematik said, OK, we want both to innovate, but also to let the archive grow. Not only by digitizing the stuff that is on analog and inside the archive, but also the digitized stuff that comes in. What do we do with it? So, we said to them, we're going to link your data. This is actually the major topic of this talk about linked data, and more or less what it means to link your data when it comes to linking it in a film archive. First of all, what does the software looks like that we've been using for the last 100 years? You all know it. It looks, actually, it is a screenshot of this software, but it could have been access, it could have been an Excel sheet, that's a little bit pimped, whatever. We call it FoxPro. It's a FoxPro, it's a Microsoft thing from the 90s. We're still using it today on the Windows 95 machine. So, we said, let's say, it was the summer of 2015, yeah, the summer of 2015, we said, okay, we're going to start preparing for the future. We're going to rewrite the software from scratch, and we're going to link it. So, we said, okay, you have tons of data, you have tons of interesting data, even tons of interesting metadata. What are we going to do with it? So, we drafted a software, and I give you a sneak peek of the software that we've written today, which more or less looks like this. First thing that we wanted to get rid of is the fact that metadata is boring. If you hand metadata to an archivist, and you say, okay, it's not only metadata, but you can add videos, and if you have PDFs or images or whatever about this thing you're creating, you should be able to just add it. Like if you have a film about the Winter Olympics of 1904, and you really have a real scan that's digitized the Winter Olympics of 1904, you've been working on this for six months. The first thing you want to do if you write your little form about the Winter Olympics and the excellent director that's directing it and so on, you want to be able to add your film to this metadata because this is the real passion. So, we created the framework and we baptized its trial. And we said, okay, this is what we want to do. We not only want to link our data, but we want to mix media and metadata in the same form. Some of you probably think, oh no, another linked data framework. Why are you inventing something new? There's tons of stuff that's already out there. Well, Sterlo is not really a piece of software. It's a philosophy. So it's a philosophy of something called Cinematheque and us together and we got creative and we said, okay, let's think like pirates. And when you think like pirates, this happens. So we said, okay, we're going to take what we can and we'll give nothing back. So this is the Sterlo philosophy. We're a small institute. We have major plans where, oh my God, one of the aims is to become the best video tech institute in the world. I'm not kidding, this really is one of our goals. But we have no money. So we said, okay, we're going to think like pirates. But first of all, a pirate needs to survive, of course. But a pirate also has to innovate because it lives outside the law and so on. And the third goal was to become the best video tech institute in the world. This is the heritage sector. We're not talking about commercial sector because that's a little bit impossible. But we said, okay, we're going to think like a pirate and we're going to give back, but we're going to give back later. So we're going to wait a little bit. We're going to take, take, take. And then when we're ready, we're going to give you back. So this is why we created Strahlov from scratch, actually. I heard in another talk today. We want to learn by doing. Actually, this is the main reason behind Strahlov why we started it from scratch. We wanted to learn linked data in a video context by doing it. So we said, okay, what is linked data? We started studying linked data. We started studying our own archive. We started studying where pirates, what other archives had, what we wanted to have and so on and so on. So this is an example, three examples actually of what we wanted to achieve by using linked data. First of all, when you enter, this is a country, when you enter Belgium in a form, there's tons of data you can steal. We're pirates, so there's tons of data we can steal. On the right-hand side, you see a wiki data. Sparkle query or the results of wiki data, Sparkle query. On the left-hand side, it's a genames map. So there's tons of data you just need to make a link. If you want to use Belgium, you just need to make a link between what you have and the concept of Belgium in someone else's ontology. And you get all this for free. So this is rule number one, how we're stealing other people's data. Rule number two, we have a lot of keywords. So if I have a movie about bullfighting and we want to annotate this movie about, this is about bullfighting, we said, okay, we're not going to enter bullfighting because it doesn't have the translation to French and to Dutch and Russian and so on. We're going to link to the concept of bullfighting in wiki data. So we said, okay, on the right-hand side, you see all the translations and all the possible languages that have translated the term bullfighting on wiki data. So we get all this for free. And on top, by linking bullfighting to our little video, all the sub-statements and the super-statements and the fact that it's an activity and this is on the San Marco Square and so on, we get all this for free. So this is nice, we're pirates, okay, we're getting there. Last thing is, when we're talking about corporations, persons, actors and so on, we're not going to enter Judy Dench in our database. We're going to link to the concept of Judy Dench on wiki data because we see here she's married to a guy that's name, I don't know, and so on and so on. Her name translated even or transliterated even to other languages. We get all this for free. So these are three theoretical concepts that we wanted to integrate in our little framework. So actually that's what we did. So I'm going to present to you, I'm going to do a live. I made a little video about it. So this is a live version. It's about a monsieur qui a mangé du taureau, the film we saw earlier about the guy with the bullhorns. And we're adding a fact to our page, it's the country of production. Actually it's not Belgium, it's France. So by just entering France and we're saving the page and we're clicking on our little link, it made this perfect link to, in this case, to the geoname concept, the geoname concept of the Republic of France. And we get all this for free. So all the translations of France and so on, even the coordinates that are probably here at the bottom, the Russian and so on, the coordinates of the country, the bounds of the country, we get all this for free. This is like one second of work in the new framework. Second thing is our bullfighting concept. So we're adding a fact. We could have added whatever we wanted. We could have added video or images, whatever. We say, okay, this fact is about, it's keyword. We're going to enter bullfighting, mistyped even, and with adult text this. So it says, okay, I found bullfighting on WikiData. Is this what you want to use? Yes, this is what we want to use. So we have a keyword now. We made a link to WikiData about bullfighting. If you click through to the WikiData sheet, this is what we get. We get all this for free. So, okay, we're there. We're three years later and we say, okay, we've done quite all right for pirates, right? So before we give back, you should always give back more than what you got with adult. So we didn't stop. We created this as well. So we said, okay, as pirates, we have a code, of course. So we're going to give back more than we got. We got all this data for free, but we're going to open source our code. And we're going to add all of this functionality that annotates the movies in real time and tries to detect using machine learning, all the things it sees in the movie. And on top of that, we're going to integrate the concept of not only saying this is a bench, which is wrong. This is a cell phone and a person. No, we're going to link this to the link data concepts. So if we want to know a cell phone, actually it's a regular phone, but without the cord and it uses radio frequencies and so on. So we get all this for free again. Second thing we wrote is an OCR engine that's integrated into the statistical engine of Stralo that detects all the subtitles, that detects all the directors, all the and so on and so on and names in each and every frame or every few frames. And we link it back into our Stralo database. So we said, okay, as good pirates with good code, we're going to give back later. This is more or less the time we did it last fall, November 2017. We said, okay, we're going to create a website, Stralo.com. The first goal of the website was to make all code open source, which is actually a checkbox here, but it doesn't render. Then we said, okay, Cinematec is parsing all their data sets of the old FoxPiril stuff into the new Stralo code and we're going to release this data set. Actually, that's what Davy is going to talk about. And then every data set needs an ontology. We're going to release this ontology and on top of it, we're going to link it to other European ontologies in Europe. So this is the project website, all code is open source and it's quite easy if you have a little bit technical knowledge to put it up yourself. Davy, we'll continue. It might be strange that I'm going to present a future after this presentation but also just to clarify, I'm not a political guy, so dealing with politicians I'm more looking at how can we take that philosophy and translate that into something that our film archives can use but how can we like exponentially let that grow within our network of European film archives as well. Second thing, I'm not a level head so this might go wrong so I might need help just to let you know that. So what we were doing, I'm sorry, we started thinking about all this and what we feel a modern archive needs to do is because we have all this cultural heritage these films, this data and these photographs, these film posters, we have all that and we wanted to see how can we open up that to the actual, what we consider, owners of that content which is the Belgian public. So for instance this film archive is very much your film archive if you're from Belgium, if you're not from Belgium I'm sorry, you have to go to your own national film archive but so how can we give them access because we've digitized a lot of films but we've never really been able to give access to these films within copyright laws and I'm not going to talk about that today but that's another talk but how can we give access and how can we also provide more information about your own history how can we provide and contextualize all this information that we have in these films that we have. So then we decided what we need to do to become this philosophy that we had that Brown talked about how can we become a really good modern film archive so we need to do and we need to reach certain goals and these are the goals we need to provide access to data sets, we need to open them up we need to provide access to our public domain content which might be simply you can just say just upload everything to YouTube but we wanted to make sure that and we're also doing that that we can do it through different channels and as much channels as possible and we wanted to incorporate very much a sort of playground for innovative ideas because we were like 20-25 people working at this archive and we have very, I mean we have clear ideas of what we want but we need more input, this is clear, I mean other people have other ideas that are very open to that and of course play to your strengths we are a film archive, we will never be a digital company like Brown's but so we need to look at that and then as I said how can we bring this to other film archives how can we have a network that grows exponentially one of the things that we're doing is because we're a federal archive and as you might know in Belgium the federal institutions are always a little bit complicated especially if you're talking about finding funds and finding money to build on this so what we normally also do and also because we are very much embedded in this European association of film archives and global, the FIAF film archive funds we also also very much look to European projects and setting those up together with other countries and one of our projects that we did where we felt like we need a couple of things we need more information about things and we also need better data sets we need better qualitative data sets meaning we need to clean up our data sets that we have but we also need to link them and we need to create as much new information and new data that we can in a shortest amount of space and what we found is that we wanted to create a platform which is what we're doing with IME Assets now, it's not ready yet so don't go searching for it, but you can find a website you can't really find the platform yet, give us a year so what we're doing is we're creating a sort of digital workspace where researchers but also the general public can watch these films, annotate these films add other information about these films and not just because normally metadata is linked to an entire film right now but to shots, we automatically segment our films into shots and we automatically add information, metadata to these shots so if we ever need to find something that we don't need to go through the entire film yet again and watch the entire film 20 minutes just to find a 2 minute shot and one of the things that we took from what we were doing was implementing this linked open data and I'm now adding the open data because we're a European project and we felt like this openness and opening up this data set could very much help other people and would also help us in the long run linking and bringing this project further to other stuff this is where it might go wrong this is a very short phase, so don't worry so what we wanted to do is we took all the images that we had almost all of them films from 1890 to 1960, 70 someone of 9 European cities and we're connecting them and we're allowing researchers to work on those images and we're automatically transcoding them as well but annotating them and letting other people add information to that as well this is one of the projects that's actually helping us a lot in understanding how linked open data can help us as an archer but help us also with European problems one of them being multilingualism there's a lot of multilingualism that is because we all speak different languages in Europe so adding linked open data keywords as with the translations would help also what you see here on the left is the north translation in Brussels like it used to be, it was broken down unfortunately and built up in the session you know now we have images of this and if we have images of this to find that in the current like Google Street, it might be difficult so how to deal with that and also during the war is a lot of streets were called the Adolf Hitler Strasse on Street we had to replace those after the war of course as well so how do you feel if you have images of that how do you know which street that was during that time linked open data has solutions that can help us with all this and one thing that we're doing and we're still trying to figure out how to launch is better the best way possible is this is our future version our future vision is what we call studio cinema tech is where we want to become this studio for collaboration on innovation on these films and for allowing other people to access our content to access our data but to help us in making better stuff for the people to discover these films but also for our researchers for our digitization people inside I'm going to close it down now, don't worry and we're working together also with other organizations and do the same like back for instance and we're trying to figure out how to place ourselves within that entire thing so if you ever have ideas I'm very much welcome you know where to find us I'm the first one, but I'm the second one I'm sorry yeah I was talking about the picture but anyway so if you have other ideas we'll come back to you so keep an eye on us we're trying to build this archive for all of you and we hope you can give us an input as well thank you very much Adrian I have any questions, maybe some small questions otherwise you can always mail us, yeah when you want to add a keyboard to a movie the keyboard doesn't exist in Wikipedia for example do you add it to Wikipedia? we internally do in these European project this was a huge thing where we're still discussing I want to add it to Wikipedia I'm not as a free text because I hate free text if there's anything I really hate free text metadata but we're working with universities and researchers and I think data is still something that's very new to them so they love free text this is a competition between two different visions but for us internally yes if we add new keywords we already have 20,000 or 2,000 so I don't know how much more we can add but we will definitely do that first to Wikipedia data so in that collaboration project how do you actually work together to manage these ontologies in a collaborative way and you use some kind of... that's really complicated is it because of the spoke solutions that you need to build or could you be there to use existing... yeah this is the spoke this is something that we're building in this platform where you can research the data we're adding linked-up data services to that and then for instance the data that's there can help enrich the film and the metadata that you want to add it's a really much more complicated answer that I can give you now I don't think we have enough time but I would gladly explain that more thoroughly over millions of the data but this is something that very much that we're building this entire platform I'm sorry to have to be a bad guy because we have to...