 So, hi everyone. My name is Axel Jean Coran from the University of La Rochelle in France. I'm a research engineer working on the News Eye project for three years now, and we'll be talking with Jean-Philippe Moreu from the National Library of France today about the use of IIIF in the News Eye project. Next slide please, Jean-Philippe. Thanks. So, the News Eye project is a horizon 2020 project which started three years ago, and the objective of this project is to create tools and help digital humanists to analyze both in a qualitative and quantitative way a large mass of newspaper data. The project website is available at www.newseye.eu and the demonstrator basically gathering all the outputs of the project is available at platform.newseye.eu. So please feel free to check these out. Next slide please, Jean-Philippe. Yeah. So, this News Eye project, there is eight partners in four countries, so France, Germany, Austria and Finland. And on the right-hand side of the slide, you can see the main workflow of our project. So, you should start reading it from the bottom. So we start with digitized newspapers that were given to us by three national libraries, so the National Library of France, Austria and Finland. We selected a set of about 185 issues, so which represents about 1.5 million pages. And the objective is to try and create new algorithms to extract information from these newspapers, create tools to analyze them, and of course create a user interface for users to be able to use those tools. So starting from the bottom, we started with text recognition and article separation. From there, we try to enrich semantically these data by trying to identify name entities, trying to identify the stance of those entities and trying to link them to knowledge bases. This all this information is gathered in a database, basically. And from there, we have two more basic steps of creating tools. The first one is trying to identify topic models, trying to create topic models on those data, trying to identify trends and so on. And from there, there is the team in Helsinki is in charge of developing a personal research assistant, which goal is to help users automatically identify interesting facts about the collection of documents, or trying to identify trends in this collection and so on. And all those tools can be seen and used through the user interface, which is available at platform.newside.eu. Next slide, please, Jo-Philippe. Yeah, thanks. So this is another view of the demonstrator. So this is a digital library-like website. I guess we are all kind of familiar with. So you have a text search in which you can enter keywords to express your queries. The system will return a list of documents that can then be filtered again with different metadata, the language, the newspapers, the documents come from, also the name entities that are available in those written documents. And when you click on a result, you are then sent to the show page of these documents. Jo-Philippe, yeah, next slide, please. So this is the show page basically of one article. You can see that all the issue is present in the viewer, but the article is highlighted, and you have more information about this article. So you can get the automatically extracted text. You have a list of metadata and the name entities that are available in this particular article. Next slide, please, yeah. And so AAAF was really something that was very interesting for us because it's a nice way to, at the same time, integrate images from national libraries, but also to export new data. So using AAAF. And so in the News Eye project, we used AAAF as a consumer for the different thumbnails in the research results, but also the viewer on the main show page. All the images are stored on the national library's servers, basically, which allows for easy integration, basically, of those documents. But we also use AAAF as a producer in several sense. So the first one is that the National Library of Finland does not have the capacity to create a AAAF server now. So for everything to be on the same page, basically, we also created a AAAF server to publish the images from the National Library of Finland. But more than that, we use AAAF to also export the different articles that were automatically exported, extracted from the newspapers, and also the name entities that we were able to identify. And we decided because there is not a lot of, there is not a lot of norm about how to share different kind of metadata, like name entities and so on using AAAF, as far as I know. So right now we just use tags in the presentation API to publish these name entities. And I think I will give you the floor Jean-Philippe. Thank you, Axel. Yeah, so as Axel just explained to us, the use of AAAF foster reuse of data. To our opinion, institutions can disseminate more of the data, researchers can share annotation and transcription, and institutions may even wish to benefit from the work of researchers using AAAF presentation API as an exchange format. It's the orange arrow in the diagram. AAAF collection are also very powerful too. Within the project, national or thematic corpora have been described as AAAF collection, which can be brought in any AAAF viewer. The same data can be accessed for calculation purposes or reuse as dumps or data set. The slide shows the French dataset as a collection, open in Mirador Free. As an example, this way, anyone can have instant access to the content. I'm showing the Marie Claire magazine. This issue as an example. So it's instant access. For researchers, the same is true. Research data set can be exported from the demonstrator as complete way of collection. It's still a work in progress, but each dataset will include two layers of annotation. One for the world issue content and another one for the articles part of a specific research dataset. Now, let's start the long spot of our presentation on storytelling for digital humanities. I can say that showcasing digital collection is always a challenge for any researcher who wish to publicize results to the general public. Within the project, digital humanities group has produced content to present a work done with the tools developed by the computer scientists. Screencase, notebooks, case to the description, Twitter feeds, blog posts have been created for different purposes and different user groups. If blog posts are very useful for popularizing research results, the heritage documents they contain are generally only used to complement the text. First, to highlight digital collection, storytelling tools can be a wise option. There are a couple of image annotation and publication tools already developed, such as Tories or Tesla, via look commenting on specific part of digitized content to produce a dynamic and interactive analysis. For all these tools, we have chosen to present exhibit, which is a true story telling tool developed by Nino scene, currently in beta testing and used on top of universal viewer. For this presentation, we have selected a narration from a blog post produced by Najma Omari for the project and which is part of a series of articles, by signaling the connection between the media treatment of the coronavirus crisis and 100 years earlier event of other Spanish flu. So let's try to open the story. Yes. Connections and links between the old and contemporary press make it possible to arise the interest of the public who have been able to discover the digital collection of the BNF. Post got 2,000 retreats and more than 3000 likes. Among the topics we covered from vaccination to wearing a mask and so on. We were interested in remedies and drugs claiming to work miracles against the fruit are like kidding. Not sure you're seeing anything right now. Yes. In 2020. Several scientists have also argued that cloaking can cure COVID. Direct links to demonstrator can enrich the story obviously, like this one targeting the matter issue. From 1918. Many advisement in the daily newspaper were about the merit of drugs, old-fashioned remedies. And to have the cure. This include pink pills or great cure, literally the fruit cure for missing recovery in your content. And some brands, some brands really disguise the ads at short stories. We say, which they printed medium portraits, meaningful to learn more credibility to into a claim. On this slide. We have link. Our story. To a Gallica proof of concept. The content base in that retrieval name, Gallica pigs. Using a request in a newspaper, just so it adds that the base. This, by the way, this was a liquid is also truly driving. And as you can see, we, we get a lot of illustration and portraits. So let's go back to our presentation now. And, of course, our topics can be addressed and I've been addressed during the project. For example, we had committed to gender case studies on women's right in particular through the press of previous centuries. The right to reappearance, the parents to vote or to practice journalism for women. All topics of interest. The story of the right to abortion. Obviously can also be told with the same storytelling to exhibit. And using again, illustrated French newspapers. We clear the newspapers. The tour is also very well suited to explain the operation on the 19th century press. The organization or the layout of the speaker. I tell us user to explore the media in an interactive and very natural way. Just to show you the last exhibit story. A typical newspaper layout. Using again panning and zooming. By exhibitors to reason. And that's all. Thank you. Thank you again for listening. I think we have plenty of room for. Yeah, creation. Yeah, so the first question that we have in Cuba is what annotation motivation do you use for this is an article annotation, annotations. That's a good question. I should take a look at that. I'm not sure to be able to answer right now. I can't remember. Yeah, sorry, I can't remember either. We have to check. Yeah, I'm sorry. Are there any other questions from the attendees. Okay, we have one other question when you tag an agent in the text. Is there an authority file, or is it just the media fragment of the canvas. We did not attempt to, to make an entity name entity or linking during this, this project. All right, any other questions from folks. You have one comment. That's a thank you. This has been really interesting. I just wanted to share in case you don't have that up. All right, great. So it looks like there aren't any more questions. So thank you both for presenting. So it's really fascinating. And thank you for everyone for attending. Just a reminder that we have recorded this and we will send it out in the weeks after the conference. So thanks all. And we will see you at other sessions. Thanks, Mike. Bye. Bye.