 Yeah, hello everybody. It's a great pleasure for me to tell you somehow a continuation of our project two and a half years ago. I already reported about this citizen science project which started actually in 2019 and it actually was finished last year, this year in February. And yeah, so some things have happened in the last two and a half years, so it's okay. So the title was Cati and Franzi. I have to explain it, I think. Cati stands for Catasta Tirol, but it also stands for Catarina, so this was one thing. And the other thing is that Franzi is a Catasta, so this is Franz II and it's somehow a bit funny to call him Franzi, because that's the smaller form of Franz and yeah, this was there. Main idea to have clear names for the interfaces. What do you see here was actually the main idea of the project that we wanted to use Transcribos to recognize all the documents of a collection which I will show you later on a bit more in detail and to set up a first search interface so that people can very early search the documents they are interested in, in this case land registers. And the idea was to have first this searchable version, actually this was the very first read and search interface. And then to go on with citizen science and to go on with more, that the scientists add more data to the database. But actually it was so successful or the project was so successful and the citizen scientists were working so quickly that we said okay we will not publish this search interface but simply focus on the citizen science project. So that's the reason why you will not see very much of Transcribos in this project. Transcribos was this first search interface, but you really will get some information now on the citizen science project itself. The background is what we call the Franciscan Cadasta. It's the first land register in Austria, so a real land register combining maps. It is a rather high resolution, 1 to 2880 and the register where the owner is mentioned, the name of the owner, the number of the parcel, the type of usage of the parcel, which is very interesting of course if you know it was a house or it was a meadow or a wood or whatever. And the size of the project could be rather large, so the whole of the Austrian Hungarian Empire was actually recorded with this Cadasta and it took about 50 years actually and it's about the size of today's Germany. But of course we are concentrating just on a very small part of this, just the northern part of Tyrol, so actually where we are here. All together these are likely more than 30 million parcels for the whole Austro-Hungarian part, but about 1.5 million here for Tyrol. The digitization of the maps actually took place some time ago. It was done by the Bundesamt für Eichenvermessungswesen, so by the Austrian government, but the land registers themselves, so where the interesting information was recorded, this was something which was not available so far and was part of the project to digitize these documents. These were about 180,000 pages which were needed to be digitized and then ran through Transcribos and made available via searching. Now the actual CADTI project is the Citizen Science project and we simply support the users a little bit with AI so they can see the transcription but most of them can read the current writing rather well, so they usually don't need it and it's really a manual task. The manual task consists of actually two sub-tasks. One is to underline the parcel in the map and the other one is actually to record the data from the land register. As I said the project started in late 2019 with a prototype version and became really progressive some months later, so mainly in early 2020 and now it's running on a very stable basis so that we hope to conclude it in 2025. Here are some results which are very similar to what we have heard before. We also never tried to involve a lot of people but we tried to involve people who are really committed. In our case this was rather simple because we have here entire the situation that each Commune appoints a person which is called the local chronicler, so someone who cares about the local history of the Commune and these are typically very similar to what we have heard before, retired people, typical teachers or public servants and others who feel somehow responsible for the history of the local region and most of them were really easy to convince that they take part in this project. As you can see here, five of them did about two-thirds of the work. Also this is very typical for this kind of citizen science project that only a small portion of people is really, really enthusiastic and working very hard and there is even one person, she's really amazing, she did never stop. I don't have the figures for now but at least within nine months she spent about 1,100 hours. So it's a lady and she's about 65. I think she has, she's not retired so she always stayed at home and she simply loves to put in data. She's a very modest person, would never try to get the audience for her work but just does it for her own reward. These data, where we had a closer look for the nine months from September 2020 to June 2021 and as you can see about 312,000 entries were made and this is, yeah, was somehow a fifth of what we expected that will be necessary. But yeah, the success, so from our point of view, the project is really a great success and we tried to identify some of the success factors. First of all I would say it's really also the type of source. I will show it later on to you. It's simply nice. It's simply very beautiful maps. You get such a lot of information out of it. You have access to these maps which is not that easy usually to get on the internet and you can connect it with the data so that the information is even more fruitful and more interesting and that's, from my point of view, really, really one thing. It's not boring. It's interesting. Then yeah, these local people who are interested, who are willing to work and the main thing really which is important for them is this kind of acknowledgement that they get both on the local side, so the mayor of the commune and friends and other people know that they're working on this, so there is a kind of thank you culture from this side. Of course, we tried to support them and we also, I think it was also a good decision to tell them that we will not prove read actually their work because it would have been too much work for us. We got 200,000 euro for developing the technical background but we got hardly some money actually to support them from this dedicated people and so we trusted on them and on their quality and they are really, really trying to provide a good quality and they know of course each other so they are also asking each other if there are problems and of course they come to the main supporter. In our case, is Gerhard Siegel, a historian who is also personally interested in this and that's the person they always can ask. Also Miriam supports them on the technical level. There is some acknowledgement by the university so we invited them for a similar public event and this will go on of course once we go to the public in a more official way. And the other thing was that I think we provided a simple interface, also a very simple organization. People know I'm responsible for this and that, mainly my own commune. They have to really commit themselves so they have to write an email to us and say I want to do this commune and then they are somehow responsible. They can add other people to build up a workforce but usually they work one or two people in one commune. And then the interface didn't change very much so after first adaptations that stayed stable and it's working rather simple. I will show you. The first level support was done mainly by phone and email. For recruiting the people we also used face-to-face workshops where we invited them and introduced the program to them. Here you see the start page, an overview of the map of the commune where you need to type in the parcel number and here's input for the protocols. But I try to give you now demos so that you get a real understanding how it works. If it works, now you have to say yes. It's like the anniversary. I'm not willing to let us do that. Sorry, no. I can try. Do you have it in mind? Okay, so this is the start page what people see. And yeah, the information is rather simple. It's called a web platform for users. It's just in German. And here you see the progress which was done in the last 24 hours. So about 934 entries were made, 9,000 in the last seven days, 35,000 in the last 30 days and nearly 800,000. So we reached already half of what we think that is the whole number. You see the most active of the last seven days, the most active projects. And if you want to take part, then it's really very, very simple to write an email and to get the links to the communes. If you're logged in, then you can see this is not public but just for people who are registered and you can see who has worked what. So you can see that Gerhard Siegeri did nearly 180,000 entries, 20-22 percent. It's not totally fair. She is much more busy and has done a lot more because she did the harder stuff. So here 135 more. And all the others you can see also per project. Project means here in this case commune. And you can also see the activities from the very beginning of the project. So this was the better phase. And then people really started to work. They were actually finished. It was not summer in this case. Nothing happened. But it was mainly that the maps were not available. And then we were able to connect all the maps. And so people were really happy and working enthusiastically. And again, there was a workshop and new people started to work. Now, if someone wants to work, he would go actually to the commune and here see a page with the tasks. So a short description. Again, you could see the other users who has worked on that. The activities for this project, some information and so on and so on. And then here the off government tasks. One task, as I said, is to mark the number within the parcels. So that's the nice task. You see the map. And what we do is to underline the parcel number and to provide the data. And that's simply done here. So you just put this line and type in the number and that's it more or less. So that can be done quickly. And the other task requires a bit more time. And it's about the land registers. The land registers divided between buildings and real estate. So here you see also the status of the parcels. So we could go for new and select the page. Here you see the land register with the parcel number, the names, the profession of the person, the address, then the type of usage of the parcel, the size of the parcel, and sometimes some remarks. And three years ago we didn't have actually table recognition. So we used a very, very simple system. Today of course we would solve this in a much more convenient way. We would run or we would train a table recognition. And then we are sure that we would get very good results on the table cells. But at that time this wasn't possible. So we followed a similar approach that we said, okay, you mark the line where you are working or where you are working and then you get a form. But it seems to have a problem because it opens in a new window. Usually it's shown here. So I think that has to do with the presentation. So here opens simply a form where people can type in. But the form could be pre-filled if we would have actually the table recognition included. And one of the important things is that people can copy the data, which makes it much faster for them actually to work. And because of course the owner and sometimes the usage type and so on and so on stay the same and you can optimize this. So this is Cathy and as I said people are working with this and entered 800,000 records. So let's go on with this. Francie is now the idea to use this data and to make a public interface for searching. We are working on this since more than a year and we had several attempts. Probably at the beginning we tried to build up a kind of dashboard. We were very ambitious. We wanted to have filters, diorograms, tables and all should work together. But that became really a bit complicated and not really user friendly from my point of view and also a lot of bugs were included. So we made a cut and said, okay, let's do it in a very, very simple way. And just provide a browse and search interface but you are able to search for all the data from the records for names, professions, commons and so on. Of course to link the database with the map and to provide all data for free download as a CSV file for the public. So we orientated all also a little bit on Google Maps and so maybe you could call it Google Maps for disaster data. Yeah, that's how it looks like and I will show you now the prototype. So what you can see here, the red, so the colors are maybe not that nice but red means that it's still under in progress and blue means that the commone is already done. And here's a simple search interface and we can type in a name like Josef, a very popular name in Austria and the first level shows you 187 commons I think where Josef can be found and we could look at the commone. We could also directly search Josef and the commone. Now it should have, yeah, so we find here just the Josefs in Wilton and we can show the parcels. So here we see now the parcels selected from Josef, I can't read it from here unfortunately, but you see the names and you see also here we means Wulgul, so that's the house name. In Tyrol many people had the names also of their houses and you see also the profession, so E.G. Bauer, so Pharma. And you could even select this and then it selects just the farmers and in this way we try to make it convenient for the users to see what is actually there. If we click away the search then we get all the data which are already typed in and we can also have a closer look. We could filter E.G. for worn-gabiders or for building or for other parts and of course we can look what is this parcel or if I'm interested in this parcel it should also work. Yeah, so that's the simple interface more or less and that's what we will provide to the public likely in the next two or three months I would say. It's already working but some things probably need to be improved. Yeah, I think this was it from my side. Thank you. Perfect, thank you very much. Are there any questions? Yeah, microphone is coming. Yeah, exactly. So thank you Gunther for this wonderful presentation. As I told you yesterday we had quite some success with using your own baseline model training to actually automatically transcribe these annotations on maps. Is this something you would consider doing before releasing the product or perhaps in the future as an update to the current system? I think for this project we will not touch it. We don't have any resources anymore so it's really running on a I do it as part of my university job and also the others are doing it just as part of their job so there is no money around. But so yeah it would be nice because it would of course speed up the process for the volunteers but yeah. Right, I think we had a question right in front of you. Yes, thank you Gunther. I was wondering I cannot read the URL but is this a public thing we can look at at home? Yes, you can. Can you spell it or increase the font size? Drop me an email and I will send it to you. I will keep it a bit. We don't want to make it really public but of course if you're interested I'm fine. Yes, we have heard now let's say different strategies how to handle the quality assurance of let's say the the crowd working place from let's say having proof ring or having not. I'm wondering if you have considered double keying in some project maybe you or the former speakers can also give an answer. Thank you for this question. I think that's really important. It probably depends on the data of course but I think citizen science projects really are helpful if there is no proof reading from someone else. Because you will experience that Yolanda Griezmaier which typed in 130,000 parcels I'm sure she could do also proof reading for others. So it's what I want to say is it becomes much more expensive if you plan your crowd sourcing project with this extra reviewing phase. But if you like Wikipedia you couldn't pay Wikipedia if there would be 20 people or 500 people working on that. Just to see if everything is correct. I mean the system is really open and if there is a mistake they can change it. And of course if you have maybe data where death and life maybe is dependent on it I would use double keying but usually it's not that bad. So I clearly say that it is also more motivating for the citizen scientists to know that they are not controlled but they have the full responsibility and it's a kind of trust you show to them and that makes them work much better than if they have the feeling oh well someone will look at it anyway. And that's also a bit the difference between students and retired people that the retired people usually work very carefully and with students the enthusiasm will go away and it's not so important for them. Okay I've got a question here. Hi Gunther this is a question, GIS related question? Behind this I mean that's interesting. The map comes from the ArcGIS server and probably that's known it's a it's a big company in the US which is more or less a monopolist on this. So the data come from the official Austrian disaster map and we even and they are completely geo-referenced and the main challenge actually was the stitching of the single images to one large image so they are geo-referenced rather well also stitching worked rather well it was not done by us it was done by the public agency in Vienna and we could even make an overlay with the modern data so that's sometimes we can do. And how did you manage places that no longer exist or have been completely built over now? All right so say if the modern GIS says that that's now a reservoir how can how did you introduce more geographical data into yours or did you reuse an existing gadget here or I'm wondering where the place name data came from did you reuse some or did you create all the place name data yourselves? I'm not sure if I understood what you mean all these data are historical data and so but what I can say is that that if I would now use a modern layer it would not be perfect but I guess something like two to three meters difference it would work so you could not make really yeah I mean these are still legal data so it's really the first disaster in Austria and it has still legal authority and some people then start to think on borders and so on and sue each sue each other therefore we are a bit careful with the modern disaster. Okay got one more question? Yes Gwinter. Truskwivers has seen a lot of developments in the past three years and we talked about how we three years ago we're really happy with results that would not be satisfying today so there were a lot of developments and you have seen a lot of projects by others so I was wondering if you could start this project today would you change anything in strategy or workflow? Yeah I would start with the table recognition that's clear and the interface itself already has the ability to take the data directly from the page file the page XML so if we would now know that this cell is the family name cell we could also directly integrate it into the form so that the citizen scientist just needs to correct. On the other hand they often like to transcribe and not to correct yeah so I mean the writing is okay and recognition would be good but yeah it's also something we learned that they often really like to simply type. Okay I don't see any more questions so it is time for the lunch break then again I think you know the drill food is outside for free of course and a vegan option is also available. See you later in the afternoon. Thank you.