 Welcome to the second Transcribes users workshop. I'll start again. My name is Melissa Terris, I'm a professor of digital cultural heritage at the University of Edinburgh and I'm delighted to be part of the REIT project over the past eight years as we have developed the tools and techniques for automated handwriting recognition and the Transcribes platform. So this is the second meeting of the users of the Transcribes platform. We're now at over 17,000 registered users of the site. This is a really fascinating cohort and a wonderful cohort when we said that this meeting was happening, it's sold out within two days. So people who are here, thank you for signing up quickly. The issue is the hunger and the interest that there is and now using our digitized content in a way where we can get access to the writing. So what we're going to do first off is we have a couple of opening papers. We have a paper from Gunter, the leader of the project, who's going to talk about highlights of over the past year and what's been happening with the REIT project and what's been happening with Transcribes. And then we're going to talk about the REIT co-op, so that's the financial model and the community model where we're going to work towards working together with the grant clinicians and then we'll have time for some questions for all of us too. So without further ado, I'd like to welcome Peter to come up to talk about REIT and Transcribes highlights of 2018 and future outlook. Thank you. I realize that some people are here who have never heard anything actually directly from Transcribes and REIT, so I repeat maybe just three sentences that REIT is an EU project which started in 2016 and here you see the list of partners, as you see a number of universities and archives from over Europe and we got quite a lot of money, 8 million euro, which was spent about 60% for research, the actual computer science research, about 20% went into the Transcribes platform and about 20% into dissemination and network building. Our goal from the very beginning was to set up the research infrastructure which shall cover the whole workflow from digitization to getting enriched text and being able to search a collection. So we really wanted to cover the whole workflow and our ambition was that you get the chance to do all steps within the platform without the need to use other tools. And one important objective was to set up the culture of cooperation and this means for me that I believe that this technology, which is amazing and as you may know I'm not a computer scientist, but I've studied actually literature, German literature, I'm working since 20 years in this domain of civilization and so I feel like a translator between technology and humanities. But this technology really will change everything and I know that our claim that we will revolutionize access to these documents could be misunderstood as a simple marketing claim. On the other hand it is really my... I'm really convinced that this is happening and it's not only happening because we are doing something but because so many people are doing something and because the whole development goes in this direction. But to make it a real success I believe that we need to incorporate across the different sectors and there are different sectors with different interests on historical documents. So researchers, many of these people have different interests compared to archives and libraries and private persons looking for their family history have different interests and of course computer scientists also have different interests. But we can meet in the middle or we can meet with the platform and that was really the idea more or less six years ago after the click of meeting in Valencia. We said okay that's our mission and that's where we want to go. Yeah, Melissa already mentioned it. It's really great to see that we doubled our registered users more or less every year and a bit more than 2000 in 2015 we expect that this year 17,000 people will be registered in the platform and it's really amazing how many people are approaching us and how many really really interesting projects are considered or tested or maybe in preparation and so on. So the idea that text recognition is really at the heart of many people seems to be a valid idea. I have summarized just for this September some of the activities in the platform to give you an idea what happened just a month ago and you can see that usually about 170 people currently more because we were on the news are registering per week. These about 450 people are actually active users per week so we are really working with the expert client and they run more than 1000 layout analysis jobs and about 600-700 handwritten text recognition jobs. What is also interesting is that about 100,000 or 92,000 images were uploaded by users just in September so about 20-25,000 images per week and this is independent of any large project so it's really what users are doing with the expert client and passing the button and working with the clients. But what is even more interesting is that some of you know that they have the permission to train models so we didn't make this really available feature because of course we do not want to over stress our servers and training is of course something which requires a lot of computing power. So if people are approaching us and writing an email that they want to train their own models we give them permission and you can see that in September more than 140 HDR models were trained by users within 70 collections of the 2K2 models in the collection and I think about 24 different languages so we see that people are working really with old languages with Arabic, with Hebrew, with of course all European languages and very much with very peculiar and special scripts like there were also people from high-fair I think working with Bangla and so on and it seems to work so of course with every model it makes sense but a lot of them make sense and this leads me to one of the also one of the main key points in the project that the data are really the most important thing from my point of view. I mean of course we are all happy with the software and what is happening there but software in Arabic this will change so today you will hear that we have now the third generation of HDR processing in transcribers and the progress is great but what really remains from my point of view is of course the data and the knowledge of how to process this data but the data as we all know and as we can read it every day in the newspaper other than the new internet economy is very much based on data and I think it is one of our main again missions or objectives that you are the only and you remain the only of your data and we are those who will support you to do most out of your data but it is still your data that you want to have here a kind of transparency which is maybe not completely always the case with other internet companies or whatever. So the current host data, your transcriptions and of course other things are highly valuable and this really surprised me that the models which were trained in September had 16525 pages of control so as you know to create one page of control is of course some work so one hour is maybe a good estimation so people worked just for September 3.8 years on control's data I think that's something which shows the value if you just multiply it with 20 euro you come to 107,000 euro and so there is really value generated in the platform and that I think goes on more or less every month and hopefully it will increase. We have concluded a number of MOUs so Memorandum of Understanding with more than 70 institutions all over the world also here we see a lot of interest and of course we want to transform this into what we now follow my talk here this is what is the future and I will quickly talk about that in some minutes. So in some way I believe we have achieved after these three years our mission which was to make the whole process available to non-technical users and to allow you to carry out the whole workflow. I'll just give you some quick images to remind you of what can be done so we are currently producing the first series of this content so civilization made easy with the smartphone a lot of progress was achieved not only with the tent but especially also with the DockScan app so it's really working better now and there will be a number of features which are very useful to scan documents and so on. Then I already mentioned the layout analysis so this is an image I got a typical image I got from a user two weeks ago as you can see this is a contract from Austria and digitized with camera and I think a year ago there was no chance to process that kind of document and how it should invest a button and you see that every line was recognized perfectly and yes of course at the binding you will have troubles to recognize but everyone will understand but in principle it will work nicely with the next recognition but it also works with tables like this and finds the lines and so on then transcription that's something you know and there was not much progress in this of course but this is an interface which is not known to all of you it is the training interface and it is really a simple tool to select pages which were transcribed and to start the training process so that's actually the easiest path in the whole process and of course the funny part because what you get is kind of learning curve so it's a visualization what the computer does and of course you get the figures how good the model is and one of the progress was this year that is the new version you can also do some advanced narrow rate measuring you get not only more figures which is not always useful because you need to understand what is the difference between the figures but before it was much easier just with the character array but anyway it also enables you to export the data to get an understanding of if you have more pages then you can get better of you and work in a more systematic way this year I have to say that one of the highlights will be the improved recognition so Umbram will talk afterwards about HDR Plus and it's now some of you already know it from personal contacts with us that this new HDR is coming and we are able to install it it is now running the training process is running on a GPU server so this goes rather fast and the results are really amazing you will see the figures afterwards but I just can tell you it is similar to last year's baseline detection it's really a major, major progress so this is one example of a text from the 18th century German text and you can see there are some red words but if you have a closer look you will see that often it is just one character and the character array for this page which is below 5% which means that even people who are not able to read this text are able to understand what is written here the same works also for printed text you can see here it goes down to really amazing besides just trained on 100 pages then I already mentioned that the web interface is coming so this is a quick preview on simplified transcription interface in the web which is working very nicely and we will hear a talk from the Netherlands, from the City of Amsterdam they will put it online for cloud users and it will be one of the tests for this interface I'm sure that they will like it and it will work very nicely then what was implemented last year also was this keyword spotting feature that some of you maybe tried out also I have to say that not too many people are actually using it I will talk about this in the afternoon a bit more but it enables you really to find words which are badly transcribed and of course exporting is possible as well one progress was here that you can come and see your own XSL T transformation and get your own export now some words in the future of transcribers we were this year talking a lot about the governance model and I have to say that four years ago when I was writing the proposal for a read we were writing the proposal for a read together but this section was on my part because I thought ok I have to write something about the sustainability of this research infrastructure and I wasn't happy to write that we would found this being of as a company but association it's hard to write a company as an association a foundation is of course sometimes a solution there are a number of European initiatives which are like a Europeana which is organized as a foundation what were the criteria for our governance model so I thought it should be something that cooperation is at the heart and to remember the culture of cooperation that's something we want to achieve then I wanted to have an open model so that people can join and institutions can join and that we get a kind of group feeling that we have synergies under our forces then it should be able to allow our CAHPS libraries but also SCOBAs volunteers and so on work together and of course also to be able to make some profit but the profit should not be for the shareholders but it should be for the stakeholders so you here in the room and the idea was then to set up something with a cooperative society and fortunately we found out that there is an EU regulation which gives us a good chance and after once we will hear more about this from Professor Dellinger here in Vienna who is an expert on that okay this is the governance model but of course we also exist as a model to keep this as a thing to expand it and so on so what are the sources of income because it was another thinking which we did in the last month and of course everything is under discussion and this conference is also very much dedicated to get feedback from you and in Frank and Rogen we also are I think putting the things on the table and I have before you feedback so of course we would like to have a freemium model the services are free for single users and so the barriers are low to best and run some processes on the transcripts platform so we want to be open here but on the other hand if it comes to professional use if a specific amount of data is reached then some processing fees probably will appear of course the amount has to be discussed I wouldn't like to say some figures now but we are open for discussions then of course some institutions are approaching us for a specific project so customer projects, large-scale projects where more work is done for a specific customer and we have done our tests with the high performance cluster in its program we can say that we can easily process 200,000 pages a day or even more so the sources are here and the opportunities are here to be earned of course we are happy and this is something which is in an academic environment very usual that we are part of grant applications so of course we are looking also from our side for grant applications but we are also happy to be part of your grant application and actually all the people are coaching us for this so that's something I believe is of course interesting for us and we will sell the scan tent and hope that it will be something which is useful for a number of people so these are some ideas on how we would like to generate only four different schedules back from or from it all if it becomes the definitive governance model so 2019 and beyond of course we want to found a great goal together with you we are already involved in the new site project so this is an EU project dealing with newspaper digitalization and newspaper enrichment again the colleagues from Bostock are there as well and we will have the pleasure to benefit from their work somehow as well so architecture recognition and layout analysis for printed documents especially newspapers will be something which will be available in 2019, 2020 then we are very happy that we get the first project with the National Act for Finland so the ministry decided that they want to give one screws a try and we will set up first large scale or large scale but at least the first project is 800,000 pages of core track codes and make them available with keeper spotting so train the models run the models and set up a specific site for this and a message which I got just two days ago that we also get funding from the Tyrolian government so a local call but the nice thing is that we are already able to convince them that the mid-code is in foundation and there we will set up a project with disaster documents and we are currently negotiations with a number of institutions in the Netherlands in Ireland, in Canada and in Greece and some other countries also for setting up a large project so I see actually a good chance to make a successful transition from the CodeCheck to an ongoing service but of course it is a challenge it is also new for us we will see but we have some good opportunities so thank you a lot for this Thanks for that, good job we will go straight on to the next presentation which is about more about the mid-code and then we will have time for a couple of questions before coffee table so I would now like to introduce Marc Castelliger I am sorry from Austria from the Austrian Riva as in the band it is not Riva's association it is the umbrella organization of Riva as in Austria whatever you can say Thank you sir, I should have practiced that before I can have the language Good morning everybody I am Marc Castelliger as was mentioned before and I am grateful to have the opportunity to talk about a total different world I think I am coming a little bit from another world as you can judge by the tie probably the only one I know has spotted another well, it is a different world it is the legal world but maybe it is of some service for you well, my career started also at the university so I am a little bit familiar with university procedures I have worked there for 15 years and became an expert in cooperatives and edited this book for example about Austrian cooperatives it would be much easier for me to found an Austrian cooperative very easy, very much of experience but you are intending to do something new the first the building of foundation in Austria which is quite an interesting project also for me I am now working almost 16 years at the Austrian Rifeisen Association which is as I have mentioned before an umbrella organization for the whole Rifeisen Group in Austria there are many banks but also institutions the agricultural field of business and the Austrian Rifeisen Federation provides audit services for all the members this is important this will also fit and apply to an SCE becoming member of the Austrian Rifeisen Association the audit services we also do training and we give for example legal advice that is my duty as you can guess well let's start the agenda we are planning to do in the upcoming 30 minutes I want to give you a short overview of what are the legal characteristics of cooperatives then the second part will be why is an SCE the proper legal form for what you are intending to do and the third part will be a short overview over the draft statutes of the weed co-op which have been prepared in the last months well so let's start with the legal characteristics of the cooperative I found the definition of cooperatives prepared by the international cooperative alliance and they say cooperative is an autonomous association of persons united voluntarily to meet the common economic, social and cultural needs and aspirations that we owned at democratically controlled antitrust this definition is fitting pretty well it's just what a cooperative is we will have a closer look to that the first part is is all about meeting your own needs meeting the common economic, social and cultural needs you can see that a cooperative basically is a self help organization a cooperative always interferes makes business with their own members it's normal to have related parties transactions it's for a limited liability company or a joint stock company it's always dangerous to have related parties transactions their own shareholders you have restrictions it's always necessary to deal at arms length to treat their own shareholders as if they were foreigners dealing at arms length this is not the case in a cooperative a cooperative is built for doing business to make well business with their own members to provide services for them to create member value and this creation of member value should be a direct creation of member value it's not that we make profit in a cooperative and then we pay dividends to the shareholders but we create the member value directly by making something that members need meeting the needs of the members that's the purpose of a cooperative immediate creation of well value for the members and the second characteristic which is very important for cooperatives is the open membership it was mentioned before by Dr. Müllberger that you have a high interest in being open for the access of everybody interested in the TransTribus platform so I think that really a cooperative the adequate legal form for you because it's much easier to become member of a cooperative than to have access to a limited liability company accessing a limited liability company would create a need to change the statutes if it's goes along with an increase of the share capital this is a rather complicated procedure in a cooperative you just apply for membership and the board of directors says ok yes, come in it's very easy and it goes along with a variable capital you're just paying more shares so the share capital rises and then if somebody wants to leave it's also possible you just leave, you get your shares back after 6 months the minimum period allowed from the law you get your money back and you leave the cooperative if you say ok you have promoted me enough it's it's enough from that side so the capital is always variable and this is the reason for a very simple approach to membership transactions in capital companies limited liability companies or joint stock companies the transactions are always at the market value it's always very difficult to find out what is the enterprise and what is the worth the value of the enterprise you have to expand lots of money for experts telling you ok what's the value the market value and in a cooperative you do all this stuff very easily because you do it at nominal value it's in our draft it's 250 euro one share you pay in 250 euros and if you leave you get out 250 euros quite simple well connected with that one is that you can't have a purpose to make any profits or to make money by increase of share prices there is no increase the shares are always 250 euros that remains the same throughout the history of the cooperative unless you change the nominal value of the commandment to your status ok the next point was the autonomy and the democratic control we always speak of self-administration in cooperatives and this feature of self-administration comes out in the people the members themselves are the only ones being able to be elected in the board of directors we have to be a member of the cooperative to become the board member in the cooperative the people themselves the members themselves administrate their cooperative and the decisions in the general assembly follow the democratic principle in former times we had one man, one road as a slogan in the democracy field also in national politics and it's the same in cooperatives normal strict model is one member one vote regardless of how many shares you have for the SCE this strict concept is softened somehow it's not so strict in the SCE you can have up to five votes for the bigger members but everybody counts also normal natural person becoming a member he or she also has got one vote so this is similar to our democracy where we don't have more votes in public elections because we are paying more taxes or some of us are paying more taxes this has been abolished I think in the 1910s something like that the origins were also in our national politics like that so this is our democratic control democratic way of decision making in the general assembly of equality well the SCE regulation is as a regulation directly applicable it's european law directly applicable and it is more or less a common essence of european cooperative laws I had the opportunity to find the negotiations over the SCE regulation some 16 years ago I think the negotiations finished then it was really an amazing process with people from all over Europe fighting and arguing why their cooperative law is the best one and we should keep their own ideas and this is a european compromise it's the essence of all the cooperative laws of Europe and so as I've seen in the first report from Dr. Müllberger you are coming from all over Europe so it will be really adequate to have such kind of european compromise as your basic legal form best of european cooperative laws in the SCE regulation for example you have the chance to select whether you want to have a dualistic system with a managing board and a supervisory board as we are used to in Austria and Germany or you can have the one-tier system the monistic system the private sector is one administrative board as a whole doing the managing and the supervision all in one the registered office of your cooperative if you decide to have one could be as I've learned in Austria this has some importance because being registered in Austria means that this is an Austrian SCE and the Austrian cooperative law the Austrian well executive laws will be supplementary applicable to that cooperative so it's the european regulation and supplementary also an Austrian law for example this means for you that an Austrian SCE would have to apply for membership in a so called revisionsverband we the Austrian association are such kind of revisionsverband revisionsverband means we are organizing the audit for the cooperatives being our members we have auditors as our clerks and they do an independent work nobody can interfere with what they are doing there is nobody who says down or up it's not possible independent audit of these clerks of our association and they are well trained and they do more than a normal auditor does the normal auditor in the company world only checks whether everything is legal and whether the accounting is correct and the accounting principles are complied with in a cooperative there is an additional feature in a cooperative the auditor also at least in an Austrian the same in Germany in an Austrian or German cooperative the auditor also checks whether there is the fulfillment of the promotion purpose and he or she would also check the expediency and the economic efficiency of the management and this creates or it is supposed to create some portion of extra trust for all the members if there is an economic auditing or what is done in the cooperative is the money used well that's the question is the management really encouraged are they doing well this is the question a normal auditor wouldn't like to interfere because he wants to be selected one more time don't tell the manager that he's doing rubbish well in a cooperative managers sometimes get this message from the special auditor okay then we have a short time for a short look for the main contents of the draft statutes that have been prepared first the name to be read co-op sce with limited liabilities nobody has to be afraid that there will be extra payment in case of insolvency our cooperative is just 250 euro per share you can lose that obviously the cooperative should go bankrupt but not anymore and by the way it shouldn't happen at least austrian cooperatives are on the last rank of all the legal forms in the statistics of insolvencies maybe also because of the strict audit okay then the sce will be set up in the monistic system as i've heard in your last meeting this was preferred by a felt majority well it is possible to have this monistic system with just an administrative board doing all the management and also the supervision then most important the purpose of the sce well everybody should find his or her interests in this clause this should be what you really want to do the purpose of the sce is to satisfy the needs of its members and to promote economic, scientific and cultural activities in particular with regard to digitization digitization or i don't know digitization, transcription, recognition and searching his or her documents some other purposes follow or subject matters the ongoing operation of the transcribers platform for example the provision and operation of the necessary technical facilities then i felt that you also want to make business with non-members people from outside it should be possible to use transcribers also for people who are not member and this has to be stated in the statutes expressively that you want to do business outside if you want to be open you have to have this clause the capital stock often sce has to be at least 30.000 euro it is denominated in euro and as i told you before the capital stock is reliable you don't need an amendment to the articles of association if the share capital rises for example because new members apply for membership and pay in their shares so you will see many changes here in the other direction you have a minimum requirement of 30.000 euro so it may happen that some members wanting to leave the cooperative will have to wait for their money because the amount of share capital would fall below this threshold of 30.000 euro and this is not possible this is not allowed falling below this threshold so they would just have to wait until new members join or old members pay in more shares so that it is possible to pay back the shares for them without falling below the threshold i think there was an automatic translation spread in preparation for this meeting the automatic translation in this point is rather misleading so maybe you understand better what is meant when you read the slides okay then for the shares they will have the nominal value of 250 euros that the amount number of shares an institution has to pay in differs according to the number of employees the institution has i think you can discuss the figures it's just a proposal but it's maybe a smart approach to have some some relation between the number of employees the number of users of an institution and the numbers of shares that have to be paid in a natural person just one share just 250 euros to become a member well there is also provision in the draft statutes about voting rights and these voting rights should also relate somehow to the number of shares so that the bigger institutions having had to pay in more shares will have greater voting rights up to the limit of five votes for the biggest of all there will also be possibility of investing members who are not eligible for the use of transcripts they are just supporters of the idea of transcripts but they are not able to use it for themselves and in that case we speak of investing members it's possible to invest in such kind of cooperative kind of strange investment because it's not for making big profits but maybe you have institutions, foundations maybe if you apply for grants anyway that somebody will give money to the SCE and then it will be possible for this investing member also to have shares and to have some voting rights here we would suggest to have greater amount of shares and in the relation less voting rights and it's also important that the whole group of investing members if there will be a whole group I don't know the whole group of investing members under no circumstances can have more than 25% of the voting rights so there is no threat of being governed by people who are not really interested in transcribals but only in their money this can't happen or should not happen in an SCE cooperative then we have a place of the General Assembly will normally be at the registered office and the registered office is planned to be in Innsbruck so Innsbruck is a beautiful town and maybe everybody wants to come to Innsbruck every year but if not so if you don't want to spend your whole budgets on the expenses it should also be possible to have electronic participation in a virtual General Assembly and I think this could be a smart idea well I think for further details just join the workshop following after the coffee break thank you so much