 Welcome. Thanks for joining and nice that you're here. I'm Malta. We start the presentation and we will have time for questions anytime you want, but also at the end we can go through any details and we'll be happy to answer your questions. Compared to the talk before, this is more like a theoretical thought experiment about what maps actually are or what do we understand as geographic web maps. But in the second part, I will also present experimental implementation on how we can publish these kinds of new web maps. So first I want to shortly start about the perspective. Does anyone know what semantic annotations basically are? Cool. Would you tell it? Exactly. Say where it's from. Yeah. Exactly. For example, the answer was it's basically metadata on an object and to store various attributes, right? What did you say? Exactly. And people use it on a web for example through URIs, Uniform Resource Identifiers. And these URIs refer to well-defined terms and concepts describing the world from a specific domain and we use it to contextualize data and give it kind of meaning so machines can interpret it better and decode the data successfully. So my perspective is more about map makers. There's actually a working group. It was the open annotation group which joined now with the W3C web annotation group. This is a kind of annotation from a reader's perspective. If you're confronted with a web map but the reader you want to annotate it, this is kind of a federated architecture for annotations. My focus is on people who create maps using these tools to create these annotations in advance for their readers. So just to make clear that the perspective is here about a different creation of web mapping documents. So yes, my assumption is that basically when we have complex visualizations such as geographic web maps, if they would be annotated we would be supported in more effectively reading what the visualization is about and maybe what is also hidden through the visualization but actually contained in the information which informed the visualization. So I think there's kind of a potential not yet tapped for semantic technologies also to facilitate human communication based on these artifacts like maps. So turning any information into a statement like if somebody says that there's a house, we have this kind of gets a bit complicated because someone said there is a house or there is something which we could call structure and we have kind of the who and the where, the coordinates, the something maybe there's an individual entity representing this house. For example, this university is kind of a defined concept, the University of Liverpool, Brussels, and it maybe exists in many databases. For example, Wikipedia or Wikipedia and if you manage to map to create maps which already referenced this item, then you can use maps to interlink many data sources. And then we would have also the label in English language we could be a structure for the society, a civic structure like I don't know how to translate it better. Basically, we can look in the definition it's a more abstract term for a house. And furthermore, to attach to this house there are many other attributes which annotate kind of on these levels. For example, sources often an organization or a person like there are many vocabularies to describe this. So if we talk about semantic annotations on maps, I decided to use the schema.org vocabulary because it's kind of a vocabulary which allows us or tries to describe everyday life concepts, like terms which we use in everyday life. And there are five very abstract entities. For example, organization is basically anything which is kind of from university to business to all kinds of companies. And it's a very hierarchical taxonomy about how to describe the world. Creative work, there are book articles, media files. Intangible is something which has no material existence. Event is also like an event and a place. We can look it up. I wanted to click here and see some examples. If you have never seen schema.org, it is a public vocabulary developed by a consortium of major search engine developers. And they use it to describe data on the web. So the Google knowledge graph also understands this vocabulary. We will see here. So here we have the hierarchy, for example, of places. A civic structure, maybe an airport, aquarium beach, bus station, campground, cemetery, event, etc. And we can say, for example, event venue has the geolocation. Let's make another example. For example, these are the relations which describe how the building relates to the geographical reference values. So we say, if I use an organization and I have geodata on it, the area serve property is there to express that the geographic representation, for example, the geometry, represents the area served by this organization. Or a birthplace is an attribute of a person. For example, you want to say this geographic reference value is the birthplace of a person. And drop off location is actually a bus stop, for example. So you can do a coordinate and put it in relation to kind of a civic structure like a bus stop and say, this coordinate stands to the content in relation as of drop off location. And there are some generics like geo, which don't specify a specific relation between the content and the geographical representation. And yeah, this is all also in the repo. And this is just to tell you about what kind of annotations or statements I want mapmakers to make about the world. So if you make a map, you're implicitly making these statements. But if you use this vocabulary to make these statements, you have a well-defined meaning and you can publish this information machine interchangeable and application interchangeable. So you can reuse this information, not only individualization, but also in other things. I will show some examples. Yeah, please. How does this represent hierarchies? I mean, the ULB campus here has many buildings. These buildings have different institutes. And maybe the ULB has also another campus far away from here. No, OK. It's not a spatial hierarchy, but a topical hierarchy. So the schema, I mean, that's kind of a class name. And then the institute is what the relation between the institute and the university is probably the institution is a part of the university. But the institute is also independent of the university. So on building in a building way, it's probably easy because we can always refer to like the spatial structure between that. They are one big thing is if we have the area of the university, we can also say and this includes all other spots. We can still interpolate and say, OK, this institute is part of the building. So in spatial reasoning, it's more simple to aggregate this. But in topical reasoning, we actually would have to need to express the relation between an institute and the university. And this is kind of a complex relation sometimes. It's not easy to describe because they are also acting on their own. It's not that the university is the umbrella for everything. So I just meant the schema.org's hierarchy that it allows to very abstract classifications, but it also allows very concrete classifications. And it is a kind of hierarchy. OK, this is a long excursion. For the first two slides, I have to pick up and get to some examples. Sorry if it was a bit theoretical. Let's get started with the next chapter. So yeah, we talked about this kind of statements we want to make. And this is kind of an example for web map, how it's maybe technically seen for technicals. We have a model of the words you did, right? And we have kind of a projection layer, which we now do to base maps or base layers, so-called, which we use. And I'm just concerned with annotations of all the overlays. So it's an interesting future work on how, for example, this schema ontology or vocabulary relates to OSM tags, for example, because this is on the base layer. And this could be matched or aligned to each other. But I basically just wanted to allow the map makers to express, describe the content in the overlays. And this is also the focus of the implementation in leaflet. So I needed to find a term for the web mapping document. It's a central concept for me, because I also wanted to see the geographic web map as if it weren't a visualization. But if it were a list, and I could sort it, I could explore it, kind of, different attributes. And I work at the Department of Focartigraphy, and some people really don't understand my computer science approach, that I want to get rid of the visualization and just read text, kind of way, because it's a less ambitious way to describe the world for me. And, yeah, so web mapping document is basically trying to decompose a map into a list of elements. It needs to be standard conform and machine readable. The idea is to reduce the complexity and just engage with certain parts of the map so I can analyze it. And it may be used to snapshot a map. This is also coming from a research project where we study participatory maps, which change often over time and have a lot of different authors in it. I will show an example in a minute. And this is hard to, like, archive even, or there's no possibility for programmatic reasoning about this kind of collections. And, yeah, if we develop such a model of a web mapping document, this would allow us to develop interactions for maps which are reusable across all kinds of maps. It's kind of an application developer approach. Before I can write user interfaces for it, I need to model the map, and that's kind of the idea behind the web mapping document. This is kind of a web map, which is a good example because it's just visualization. And behind this is a real database, and you can also engage with the data in a different perspective. But this map has a lot of different content types, from ideas to problem descriptions. And each kind of bubble is also a different author behind it. So there's kind of a very complex amount of information aggregated in this visualization. And the website also provides a means to engage about the ideas people have about the city. They try to engage citizens of the city of Hamburg to engage in ideas and exchange ideas about how to improve the city. And they also have alternative visualizations for the same data. So it's not just only the map, but there are lots of mapping applications built nowadays, like Carjo or Mapbox, where they just collect information in maps, but it is not accessible anywhere else or anywhere else. So I hope to also to approach these companies with these free software from Leaflet to enable their users to publish maps, also machine-readable and exposed for other application developers. So I had to study a bit HTML and how to express content of multiple authors in one document. I investigated some terms and these are basically all the old approach and these are all the tags which actually allow for structuring the content depending on that there are many authors. I decided to go for a mix of articles and address. So because of the content in the article, now describing a web map element should be independently and reusable of the map and it is used in. One example of such a content, this is a map of a festival I made and this represents an artist in the band with a concrete track and if the publisher of this map, which is occasionally me, also knew about semantic annotations for map maps, I could open this map in my VLC or use it somewhere else or save it as a playlist and do something else with it. So it is an example where people spend much time organizing information in maps but then it is kind of lost in it and I see it as a practice, like an information organization practice that people start to do. It's somehow happened that we start to organize also information in maps. So my hope is to contribute to a document format which makes it more reusable. So what is a web map element? As I said, if we want to have a list of web map elements, if we want to see the map as a list of elements, we have to define what is an element and this is kind of my definition is similar to a point of interest but can also include areas. It's not specific about locations and it's basically everything you overlay a map. And specifically about my understanding is that a web map element is both together the geographic representation, like the geo data used in it, for example a polygon or a point location, plus the content. Just the content description, which actually is represented by this geographic information, together make a web map element and both are kind of, as I said, statements about the outer world from individuals. And if we start to see web map elements as distinct items, we can connect these information or elements to the spatial index of the web, which is not yet possible if we annotate them. We can also show selected information on a specific web map element or amount to users. And we also maybe, which is not yet, like I have no evidence for, but theoretically these kind of maps should be indexed by major search engines and we could actually find maps through a topical description of one element. Like we can see what kind of geo data someone used to represent this content in which map, right? So it makes a different, it makes them open up for public research. Yeah, so that's, I will shortly present the vocabulary I used. So this is basically the schema.org vocabulary to describe a web map element. And this is a relation I explained in the beginning. So in which relation does this content stand to the geographic data? And this is basically the denotation for the geographic data. And this is schema vocabulary and this is all Dublin core vocabulary. Dublin core is kind of an old established metadata standard to annotate any kind of information resource. So it just contains very basic terms to describe data sets or data entries. And, oh, I should have exchanged. Sorry for the, yeah. So this kind of overall is the 20 terms or so which I propose to annotate a web map element with in text form again. And yeah, so I also wanted to make an implementation and I think this is the next slide. So I will now go into the practical details. Five minutes, yeah, I have five minutes left before we can open up for questions. So this is a standard leaflet example of a marker and it's basically an image tag with a resource. So there isn't really no meaning in the document about there's no spatial references exposed. There's no topic classification, even if there is a title. The title is, I think, it's not even in here because it's another layer. So it's hard to guess something and which is totally fine because leaflet was not made for this and HTML also was not made for this. I don't want to be misunderstood. I just chose leaflet to implement because it's very easy to extend and a lot of people use it. And so there's a potential that a lot of people can just change the HTML output if they install my plugin. And this is a map where we see one thing. So my plugin contains two components. It's not really well illustrated. But the one component is to generate the new kind of HTML model. And the other component is kind of a first example of a leaflet control which engages and operates on this model. And here, for example, we have a full text search through the HTML. And when I select an item, I can control the viewport of the HTML. This is kind of possible. We can build these kind of controls with the new model of the map and actually have kind of an ecosystem of map interaction dialogues which work on every map which implements this model. So there is an example of my map and a structured data tool for... I'll give a quick demo before I jump into there. Let's see how it works. So this is basically 400 public sculptures. So photographs of public sculptures. It's not the sculptures themselves represented in the map but the photographs from the Wikidata database. And as you see, you cannot really read anything here. There's no really information because it's hard. The labels would overlap on stuff like this. But now I can use the leaflet control and say, can I find the tomb of Oscar Wilde? It's in the Scherler pairs in the famous cemetery in Paris center. And I can... So this click here actually gets the spatial data from the HTML. It has no relation to the leaflet runtime. So it really extracts the references from the DOM. And this is kind of the dialogue for the metadata. And I also see this data set comes from... This element, this web map element, is actually a Wikimedia file which is associated from Wikidata. So this is kind of a map where I went this far that each element has a URI for the entry in the Wikidata database. So it's a map which just interlinks other databases. So this is one example of... Where I see now I can read these are 412 photographs in this map. Before that I couldn't find out what is this map about. Now I see the description of the map itself tells me this is a map about photographs. And another example is here for example... In this map I see two creative works, one city, one person, one organization and two countries, which is actually not true. It doesn't make sense. It's no problem when you have a multipolygon. It tends to be counted as many countries, but since they all have one identifier, we can interpolate and say this is just a bug in the software in a way. I say, for example, the creative work is an overlay. So we can actually also load images of other base layers and start to annotate them. And this is now expressed as a bounding box in HTML. And I can... Now this is a historical map of the same area, right? I will check out the HTML here. So, yeah, here we see. I make it a bit bigger. A moment. So this is represented by an article. It's an article of type creative work. And the location it depicts... So this is a special relation. The content location means it's a relation to express that this article depicts this area, which is of type geographic shape. And now there comes the bloat, I think, with a... Oh, yeah, no, this is just a bounding box. It's not so much. I mean, the polygon of USA is probably 80 pages long. So it really is a lot of data. Yeah, but this way we can actually find... add these elements to the spatial index of the World Wide Web. And just a short example of how leaflet annotate works. As I'm running out of time, a few examples. Maybe... No, time's up. So now we have time for five minutes for questions. Okay. Anyone has any special questions about the leaflet API? It's leaflet.annotate, the plug-in, it's on GitHub. And you can now start to annotate markers, overlays, circle markers, and pop-ups with using schema.org. Sorry for... yeah, please. The what kind of schema? Yeah. No, it's another taxonomy, another vocabulary. And this could be the schema.org vocabulary has a specific purpose and the OSM vocabulary has a specific purpose. And we could research into a mapping. It's just one time we need to align certain things and it would be interchangeable. So the question was if I'm reinventing OSM tax, but I see it as a free-form taxonomy and schema.org is a controlled vocabulary. So we could actually benefit if we align these two vocabularies and this needs to be done just once or whenever one or the other changes. It's nothing I understand as a... I'm not trying to reinvent this. I'm just using the schema.org because the search engines understand it, right? One more question? Okay. Yep, thanks for your attention. Sorry.