WEBVTT 00:01.305 --> 00:30.589 [Automatic captions by Autotekst using OpenAI Whisper V3. May contain recognition errors.] I have the pleasure of inviting you to this webinar about news in TSD. The date is February 28, 2025. For those who follow the recording of this information, this webinar will be uploaded and posted on the website for the webinar in the future. It is possible to 00:30.606 --> 00:52.875 on the text of the webinar in your video player. It is written in both Norwegian and English, or you can turn off the text. We will take this webinar in English, so I have to switch languages. 00:53.507 --> 01:13.695 So, please welcome to this webinar about news in TSD, chain of stocks for sensitive data. With me here at the University of Oslo, I have some colleagues working on TSD and we will go through the main part that has been changed lately in TSD. 01:13.712 --> 01:41.647 My name is Dagfinn Bergsager, I am the director of the department that has ESD. Anne Bergsaker, I work as a section manager here in IT. Jan Bråten, IT department. Hanif, IT department. Patrik Bølman, IT department. And I'm Leon Dukoe, and I work in Dagfinn's section. 01:43.456 --> 02:10.503 So, we have a program with the keywords that we will go through here now. And the first part we will talk about is that we have integrated TSD slash NetShare Map with Helsinorge.no. We have some documentation that I linked up in the program for today. 02:10.998 --> 02:40.452 And as you can see here, it describes the integration with HealthNorge. This is an integration that is only available for Norwegian hospitals and Folkehelsinstituttet, as they are the only ones that are allowed to use HealthNorge. And there is also a description here about what you need to do to use this integration. 02:42.602 --> 03:11.049 There are also a description of the data flow in the service and that you have to contact us if you want to use it and you have to have an existing agreement with Helsingborg. The price for using this is a flat price for as long as you don't have a lot of invitation. But I will talk about that when we talk about pricing. 03:12.637 --> 03:41.869 So Leon, you have a small presentation here about the integration. I will share some screenshots of what it looks like. And the whole process starts actually with a dialogue between us and the TSD project, the customer, in setting up the templates that will be used 03:41.886 --> 04:08.422 for sending the messages to Halse Norge. And on that part, I don't have any screenshots, but once you have done that, also in dialogue with us, we will configure the specific net schema to be connected to Halse Norge. And then you get this Halse Norge button, which you can then click. 04:10.316 --> 04:39.974 And then you can start inviting people to answer to this niche. And you do that by pasting their national ID number into this box and creating a task. And then at the bottom of the screen, you see here that you get an overview of when you sent the task, whether it has been answered, when it was answered, and that you can search 04:40.913 --> 05:07.602 based on the national ID number. And what happens is that we create a generic task in Helse Norge, which looks like this. And you get a link to the Netschema, which you can click. And when you open that, you have single sign-on, very important for good user experience. 05:07.619 --> 05:23.217 and then you can fill that in. And we store the data as we always do in TSD, and it is available as an HMR submission. You can use the HMR data portal to view it, do research on it, 05:24.531 --> 05:47.585 And importantly, we also send a copy, a PDF copy, to the respondent in Helse Norge. So they get two things. They first get an invitation with a link, and then they get a copy of whatever they submitted. And by doing that, we fulfill our obligations to respondents. 05:48.797 --> 06:10.691 Yeah, it's a really good example of different organizations in the public sector working together to solve the complex needs of contemporary research projects where you have clinicians involved, non-scientific staff collaborating with scientific staff. 06:10.708 --> 06:29.872 And yeah, getting people involved in research without having them relate to new systems. People are already using Hansa Merge and they know how to do that. Yeah, stop sharing. Sure. So, Mikkoff, so... 06:31.152 --> 06:51.613 If you then have a Helsinorge net schema that you need to be created, by now we will create it for you, and you create an invitation, it is DSD that actually creates this invitation, and you, user of that service number, will get a notification on the Helsinorge app. 06:52.039 --> 07:06.527 or the notification system from Helsinorge. And when they click inside Helsinorge, they authenticate as a Helsinorge user. And most people there using the app and they have a long-term authentication. 07:06.937 --> 07:36.442 Netsham has integrated with that. So we are 100% sure that this is the same person that is filling out the form and the data is sent to TSD and then TSD will send a notification back to Helst Norge that this data has been delivered by this user. And that data can be a receipt or it can be a processed report, for example, giving a result based on what you have answered in Netsham. 07:37.108 --> 08:06.920 And this is operative now. We have been running some tests lately, but it is definitely operative. So please, it looks like this is a very popular feature for many people. And can someone at the University of Oslo, for instance, set this up? Or do you need to have a collaborator that is a hospital? You need to have a collaborator that is a hospital. The University of Oslo 08:11.015 --> 08:37.944 someone with their own agreement with the health by now. So and then all the hospitals in Norway have this agreement and at least Folkhelsinstitut have it. Yes. So we will then continue to the next point about news in TSD the last last half year. I will now 08:40.640 --> 09:02.961 Just share my desktop and I have already logged in and are ready to log into my project in TSD. This is a classic project in TSD. I am using Windows as my environment. 09:03.166 --> 09:29.480 Here I have a couple of browsers available that most people have, will be the same in Linux virtual machine. So if I go to, for example, Firefox, it will now always start on this website, which is showing all available web services that is available inside. 09:29.497 --> 09:47.312 And as everyone know, I can't upgrade it because this is not on internet. And as you also know, I hope I cannot go to UIONO because this is not on internet, but there are some 09:47.329 --> 10:14.292 web services available, and they are listed here. The most popular is Netgema portal that looks a little bit like this. You can view all Netgema submissions. I can click on show forms, and you can store your password. It's secure. And it's, of course, not two-factor because it cannot use two-factor as it is not on internet. 10:14.906 --> 10:44.258 And here you have my submissions to my project. And we have an internal publication portal. So if you are going to publish files from your project, you can use this service. This is a separate service where you can give people access outside of TSD to separate files. Logging in with BankID. 10:45.333 --> 11:12.022 I will show the autotext later in this presentation, so I will not go into that. And tables, that is... It's a collaborative, audited, spreadsheet-like application, which can be used for many different purposes, and we have a project using it to manage their contact register. 11:12.039 --> 11:29.684 for example, and you can set up integrations to other data sources. Have we created this? Yes. Yeah. Okay. Yeah. In collaboration with, uh, with Nubu, one of TSD's customers. Yeah. 11:29.701 --> 11:47.159 We have one more question about the integration. And the question is, if you already have an agreement with the hospital, does the whole thing that you showed, does it have to be set up by someone at the hospital or can someone at the university set it up? 11:47.159 --> 12:15.196 There is a contact to Helsinorge, bestilling smotak or something like that, that is in documentation. Ask them. This is like, we have a good collaboration with them, but they have to ask them. There are some things that we cannot decide, such as the content of the message. And this has to be decided by the person or the institution that is responsible for the integration. 12:15.213 --> 12:44.292 So the answer is, ask Hilsun Norge and they will tell you whether you can set this up if you're employed by university or if you have to ask your contact or your colleagues at the hospital to do it for you. Yep. So sorry, no answer from us. Yes, then I go to the next point we have here. That is a new portal for 12:44.309 --> 13:10.350 So I will then share my browser. And this looks like this. Congratulations, Patrik. So what is the overall thing about this new portal, Patrik? 13:10.367 --> 13:34.428 Do you want me to share? Or do you want to share? Yeah, or I can click around or do you want to? Do you have it already? Can you just share it? I've got slides that you can share. Please share. So earlier we had separate services outside of TSD for management. Now we have 13:34.445 --> 13:59.206 this new portal. For the ones of you who don't know the self-service portal yet, it's basically addressing three groups of stakeholders. The ones who might be future users of TSD, the ones who are already regular users or the administrators and PIs of projects. Future users will just apply for a project, a TSD project that is, 13:59.923 --> 14:18.968 Regular members will review their data, have the ability to review their privileges that they've got in their project and change their password and OTP, which they will have to do from time to time to satisfy our security requirements. And the administrator has a large variety of things that they can administer inside their project. 14:20.401 --> 14:43.541 The new self-service portal has been quite a journey. And if you've been with us for a couple of years, you will probably notice the old self-service portal, which has been with us for quite some time now. If you happen to be a user of EduCloud, you may have seen the intermediate version, which is version 2, which is at the moment still running in EduCloud. 14:44.548 --> 15:12.363 And for TSD, we have now moved on to version 3, which is what you see, what you experience live now. This is the newest iteration of the self-service booking. There are quite some novelties. I mean, the most obvious one is the look and feel. It feels different than the old one. It's supposed to be more responsive. That is, if you drag things around, if you increase your screen size, it's supposed to flow with you on your various devices. 15:12.381 --> 15:28.183 It's got now email notification that is, for example, if you apply for a project, you don't have to notify your PI anymore. The PI gets notified automatically and can then respond accordingly. And once they approve your application, you will get a notification in turn. 15:29.445 --> 15:49.991 We've got integrated manual pages, which is novelty. So for example, if you want to set your password and you don't know how to do it, what the implications are, you'd click that button on the top right side. And then you see information about what this all pertains to, which might be useful if you lost context along the way. 15:50.811 --> 16:17.927 Something I'd like to point out that the new integration of self-service is a collaborative effort, as it always is between multiple sections. This one in particular has drawn upon the capabilities of our UX group, which has been with us since the very beginning of when this project was launched, where they guided us through the project in terms of localization, 16:17.944 --> 16:47.022 consulting how to do things, best practices, and so forth. And they did quite an amazing job in terms of what they invested in terms of time and effort. And you can see here that they built an entire design system for us that sort of is something we can draw upon if we then create our components for the service. They did that for multiple of our components. And this is their Figma design, for example, and this is the outcome that you see. 16:47.927 --> 16:58.797 For the future, we are aiming for increased stability. It's still a new product and there might be an issue or two that will come along with it. We are working on it. 16:59.667 --> 17:28.473 To report those errors that occur, we've got now started a new iteration that will focus on resource management. That is, the first part of that one will be the opportunity for PIs and administrators to review what they are actually paying for at the moment, what kind of resources they have in their project, virtual machines, HPC resources, review them, and later also change them, but that is in mid-term development. 17:28.490 --> 17:50.964 will integrate the notification center more closely but as you will have a list of what you have been notified about in the past and we will try to integrate the same system for for edeka and that's my yeah my demonstration yes and if i give my demonstration 17:50.981 --> 18:10.282 As you now can see, there are several features there, and we have now made this a system that is developing every day. For example, these two new boxes came last week. 18:10.299 --> 18:30.913 So there are several management solutions there and we have worked a lot with the single sign-on between our services which was and now you can also log in with Ideporten and you can choose what project you want to you want 18:30.930 --> 19:00.555 to log in. As you can see here, I have now logged into my project P896, but I can also change this to, for example, P11, where I'm also a user. And that I can do with my TST user or with logging in with Interpol. So we hope this portal will make it easier to manage the projects and will, like we say, 19:01.305 --> 19:25.418 there will be new features here coming for them. And talking about the users, we have done some improvement with the user support. So let me introduce my colleague Tyrell from 19:26.203 --> 19:44.138 from user support at University of Oslo. We know that TSD has a lot of customers that are in contact with user support every day about TSD. So Tyrell, please say hello. Hi, yeah. 19:44.155 --> 20:07.807 My name is Cyrill and I work at the UiO's central IT help, which is the main point of contact and support for TSD. And we are a team of around 50 people who tackle support across all services in UiO. But we mostly focus on TSD because we get a lot of questions about TSD. 20:07.807 --> 20:36.288 So we offer support through mail, chat and also over the phone. And we can also set up digital meetings to have support if there are issues. And our main focus is to ensure quick help. So over the last fall, our focus has been to make the communication between us and TST even better. And one of our main focuses has been to improve 20:36.288 --> 21:01.408 and expand our system for super users in tst and the super users are a support level between it help and tst and they are able to offload tst by solving issues previously solved by tst directly hence freeing them to work on all the many cool things we will hear about in this webinar 21:01.408 --> 21:16.852 yeah and we have bi-weekly meetings with tc and where our technicians and the super users talk about coming changes and keep us up to speed and where we discuss yeah problems and the status quo as well 21:16.852 --> 21:46.681 Yeah. And we also have regularly coursing and yeah, all introduction courses and intermediate courses for all our employees to ensure that everyone knows how to offer the best support on the service. Yeah. So if you have questions or wish guidance for current or coming projects, you can always contact us at ithelp at uio.no or ithelp at uio.no. 21:46.886 --> 22:11.988 Yeah. Thank you. Thank you. Yeah. And when Tyrell is talking about TSD, she's talking about like, for example, Hanif that is sitting back here working on direct support on a higher level. But we have TSD is very big for the University of Oslo and the IT department and everyone here has some way to work. 22:12.005 --> 22:27.483 working with TSD and we have a special focus on IT support and hope to improve every day to be better at IT, to follow up users of TSD. 22:27.500 --> 22:56.203 And I think everyone can agree that TSD is getting better every year, every day and every month. Luckily we are going that direction. We are scaling, we are getting bigger, we are getting more features and we are also getting more user-friendly every day at least. And we are also getting larger. 22:56.886 --> 23:21.135 Now we are working also in some projects to take a look into other solutions in Europe to see if we can get any experience from other people working with the same advanced data management or research system for sensitive data. 23:21.152 --> 23:49.718 And Hanif, you and me and some colleagues are part of projects that is looking into this. Can you tell a little bit about this? Yes, thank you. So I'll talk about some of the EU projects that DST is working closely with. And the first one that I would like to mention is the Spuhin project, which is led by Directorate for e-Health. 23:50.606 --> 24:20.435 And TSD is working closely with HuntCloud at NTNU and SAFe at University of Berrigan in defining standards for secure processing environments and data transport. And we are actively contributing in work package five, which is about requirements for secure processing environments. 24:20.452 --> 24:50.230 and Work Package 7, which is about secure data transport to secure processing environments. And based on this participation and using our own risk assessment tools inside DST, we are working on a new and updated risk assessment and threat models, and also on a new system description for DST. 24:53.387 --> 25:15.776 The second project that we are working on is EOSC in Trust, which aims to create a European network of trusted research environments for sensitive data to enable secure and interoperable access to sensitive data. The project is creating a common framework or 25:15.794 --> 25:44.002 in its current stage, a blueprint to standardize legal agreements, system architecture, operating procedures and interfaces for federated data access and analysis. And it is based on preliminary version of the DARE UK federated architecture blueprint. And in the first year of this version, TSD has demonstrated 25:45.538 --> 25:57.142 FEDERATION BETWEEN TST AND SAFE IN BERGEN AND THE BLUEPRINT ACTUALLY ORGANIZES THE ARCHITECTURE IN 25:57.841 --> 26:25.350 research analytics zone, which is something that Doug Finn just showed by logging into the TST project and working in, for example, the Windows environment. And then there is the secure data zone, which is the infrastructure. And finally, the query management zone, which is something based on the FAIR principle where you can publish and share data. 26:28.968 --> 26:57.295 And for the 2025, we will continue working on work package 11, 14, and 17. And for the training package, we have identified individuals who will be involved in governing, operating, or using new and existing TREs and map their training needs. 26:58.183 --> 27:27.978 And we are currently addressing the needs of the roles like TRE owner, builder, federation operator, data provider, and a TRE user. And all these efforts in TSD are contributing towards the development of Noorthrae, which will present. Thank you. Yeah. 27:27.995 --> 27:54.923 So, Nord 3 I will talk a little bit about later, which is a collaboration between PSD at the University of Oslo, SAFE at the University of Bergen, and Hønd Cloud at NTNU. So these are all T-R-E, and we are three. 27:54.940 --> 28:23.234 It's not a little bit funny, but this stands for Trusted Research Environment. But during your project, Hanif, are there many TREs in Europe? There are quite many, yes, but not, I would say, DSD is probably one of the best ones, which have the most advanced features. 28:23.251 --> 28:46.800 that are available for the researchers. This is at least my impression. Yeah, that is my impression too, that we are the most complex with the most users, but maybe there are some with more data, maybe there are some with some very big projects, but we have big projects too, but we also have a lot of projects. 28:46.800 --> 29:14.821 How many projects is active now? I don't know the exact number. Is it close to 2,000? I think yes. It is looking to be stabilized around 2,000 actually. But there are new projects every day, but we are finishing projects every day as well. So does every country in Europe have their own TRE? No, I don't think so. But we know that folks, do they have in Denmark and Sweden? In Denmark they do, in Sweden they do, in Finland. 29:14.838 --> 29:31.357 These are the projects that we're working closely with. And then we also have in Netherlands and Spain also, I think. Are they connected to the universities as well? Yes, most of them are connected. 29:33.456 --> 30:00.435 And what is the main differences, do you think? Do they give virtual machines like we do with storage and HPC? Yeah, so it's more or less the same, but different. Same, same, but different. Yeah, classic. Okay, so we are very tight into what's happening, at least in Europe. Are we working with America or Asia or something like that? Not in these projects. 30:00.452 --> 30:25.486 No, not really. We are still in Europe here. It's a little bit easier. So then I will talk a little bit over here now about NORTE, which is a collaboration between the University of Oslo, University of Bergen and NTNU. 30:26.254 --> 30:49.855 It just showed there in this other project created a direct connection between SAFe and TSD. We were able to send data from SAFe to TSD and the other ways. And that is a good start how we can share resources. But in the national budget for 30:49.872 --> 31:17.995 for 2025, there was also 5 million Norwegian kroner given to Nord 3. And that was based upon a letter that we have now got that looks like this. 31:18.012 --> 31:39.804 We are working very hard these days to talk with interest about this. I will not go quite through this letter, but what is special is that it is, of course, sent by Kode, as they own the universities in Norway, but it's also from Kod helse- og omsorgsdepartementet. 31:39.804 --> 32:00.145 and we are to show, to give a feedback on how we can organize a national system for handling sensitive data. We can assure you that we will still focus on research, that is our main focus, but there also might be 32:00.674 --> 32:30.657 be things here that could be used for other things. We will come back to what happens around this national initiative from CODE and HUND that is taking out of... We are very happy to be able to scale up our systems to be used more in handling sensitive health data. 32:30.674 --> 32:54.531 Then I think this is like what has happened till now for the news. And then we will go over to some news coming up. And a lot of this is we don't have that much to show, maybe. But Patrick mentioned that this is about... 32:54.548 --> 33:19.991 that we are working on a portal that will of course be a part of the self-service portal where you can see actual actual usage yeah and that creates um a lot of like benefit for us we spend a lot of time generating invoices for people um and we don't really want to spend time on that 33:20.572 --> 33:49.735 We want to give people that type of overview when they need it so that you can monitor your costs continuously. And at the same time, it creates the foundation for allowing people to change their resource usage when they want it. So that if you decide that now you need another virtual machine because you're going to process a lot of data, then you just say that you would like another virtual machine. And if you need more storage, then you can just 33:49.753 --> 34:17.483 say that you want more storage and if you don't want that storage anymore, you can reduce it. So yeah, more self-service, more control and oversight for people and we get to focus on improving the services instead of doing things on behalf of other people. There are obviously many benefits of this and we 34:17.500 --> 34:28.524 For example, we can see that some projects need a lot of RAM. For example, they need, because they are working on large CSV files and you need RAM, and if they have 34:28.831 --> 34:58.831 Many people working on the data set with RL or something like that. We know projects that order many hundreds of gigabytes of RAM for a period, but then they are stuck with it and they have to pay for it. And we are not making maximum usage of the resources as well. So this is, I set up a link to our price model, but this is the big change now. So you can log in to see, okay, I'm actually, I don't need all this storage. 34:58.831 --> 35:25.794 I don't need all this, I don't need this. So it will be a more dynamic pricing based on usage and what you need. We have a question here regarding prices actually, because there is no mention of GPUs and what they might cost. Is there any difference here in terms of using the HPC system? 35:25.811 --> 35:41.067 Then we can change over to the next thing that I have on my program. This is upgrading of Colossus. Colossus is a supercomputer inside TSD. 35:41.425 --> 36:09.104 that has been there from day one. We have always had a supercomputer and it has always been called Colossus for some reason. I don't know. Whatever. It's called Colossus. Let's just say that. And that has been a joint venture from Sigma 2 and some milliers at the university that has invested in that supercomputer. Colossus is getting out of date now. 36:09.121 --> 36:38.780 And we are upgrading it, and we are now ordering a new supercomputer with CPU and GPUs. And this is now only a Sigma 2 machine. So we will use this only for... Yeah, so Sigma 2 is even more involved now in this new Colossus, and it's a part of the Sigma 2 HPCs in Norway. 36:39.241 --> 36:59.360 And as a part, and the pricing to use resources from Sigma 2 in Colossus and in the national HPC supercomputers here in Norway, that is an accounting system that is handled by 36:59.377 --> 37:28.848 the people that work with Sigma2, like us, for example, that work a lot together with Sigma2. And this is pricing sent back to the institutions. You have to apply to use resources at Sigma2 and there are some deadlines where you can apply for resources. With Colossus, we have changed this now together with Sigma2. 37:29.206 --> 37:51.271 that we are allowed, on behalf of Cinema 2, we are allowed to give out up to 200,000 CPU hours, which we have an algorithm to make this over to GPU hours. 37:51.288 --> 38:03.609 And that makes Colossus. And then Colossus are made now available for all projects inside TSD. 38:05.640 --> 38:30.913 And because earlier you had to order your own virtual machine to be able to submit jobs and have access to classes. Now we have created an HTTP API inside TSD, so everyone with a project have in principle access to TSD. And you are allowed to use up to 200,000 hours without applying to Sigma tool. 38:30.930 --> 38:57.500 It will be a part of the general accounting for the institution, but this is not that big number of... It's not that costly. So we will now also change the price model that we don't charge for DPU and GPUs. You will pay what Sigma 2 says, and it will not go per project, but per institution. 38:58.899 --> 39:28.899 This makes us also able to use Colossus for autotext that I will show later, where everyone actually uses Colossus. Then we will have a small demo of a new system that we are launching called RISC. 39:28.916 --> 39:48.131 We know everyone needs Ruth Annalisa now and then and everywhere, especially projects in TSD. At the University of Oslo's IT department, we have 39:49.326 --> 40:07.722 making rules analysis here there and everywhere for many years and mostly with excel and Leon is not that big excel fan. I think life is too short to spend your time in spreadsheets that you don't know where they are or 40:07.739 --> 40:35.401 when they have been updated or by who. So there are many reasons for not doing risk analysis in Excel. I mean, it's great if you do take the time to do risk analysis. And if you happen to be using Excel, that's also fine. But one can do better. And if you have a lot of services like the University of Oslo and like DSD, 40:35.418 --> 41:05.265 then it is difficult to scale that work. So we've created a system based on the tables system. So we're using TSD itself to manage its risk analysis. A great example of eating your own dog food, as they say in technology. And right now I'm showing it. And this is kind of an internal alpha stage, which means that 41:05.282 --> 41:32.363 the service can change quite drastically, but we are using it in a real way in order to validate that it is useful and to discover the problematic things and improve that. And if we are able to use it ourselves, then we can be convinced that it will be useful for others. And so here I'm showing TSD's current risk assessment, and we have made 41:32.978 --> 41:58.336 a way of organizing your services into different categories and hierarchies so that you can get an overview and i can show for example the table which shows all of the potential attacks against tsd and then you can see that you can for example rank your risks according to their severity 41:58.353 --> 42:23.114 And what's really important here is that you have audit so you can see who changed a risk element, what has been changed, and also an optional message so that if I were to, for example, make a change, let's say, 42:23.131 --> 42:48.490 If I wanted to change the description of this, I could say why I am doing that so that my collaborator in the security team could discover, OK, there was a change. Why? Why? Why did it happen? And so we hope that this system will create a knowledge base which we can use to improve the services. 42:48.507 --> 43:16.459 And the goal is to adopt that first in your division, Dockfin, and then in the rest of the IT organization, and then for other organizations as well, if they want to. Yeah, because this is stored in tables in your project. Yes. Because everyone has this tables service in their own project. 43:16.476 --> 43:46.425 So all the data are stored in the project. So for the IT department, we want to make an IT department project for all those analyses for other systems. Yeah, and it has great potential because once you have collected all risk analyses together, you can start querying across different systems. So IT security could then ask for the top 10 risk elements across all services. Will, for example, Oslo Universitetssykehuset or Folkenhelsa 43:46.630 --> 44:08.746 be able to see their risk analysis for all their projects? It's not a cross-TSD project service. Now we have very few of those. The project separation is very hard. So with one project they can make their risk analysis, they can make it 44:09.241 --> 44:29.343 in their project. Yeah, the hierarchical system allows you to organize your risk assessments in a logical way. And next steps will be allowing login with other identity providers such as either Putten or FIDA and 44:29.360 --> 44:58.114 implementing access control so that you can limit who can write to different risk assessments. You typically want everyone in an organization to be able to read it, but only specific people to be able to update it. So, yeah. And Will, there's a question here. Can it be used independently of TSD? Well, in this case, we use TSD kind of as a 44:59.087 --> 45:03.882 as an application development platform. So, 45:04.411 --> 45:34.394 It will be almost completely independent. You will need to buy a TSD project. And that's it. Like, for example, the IT department will probably have one or two projects for different rules analysis. And then we can have 200, 2000 disk analysis projects. I mean, it will be almost independent. You will be able to log in with Ederporten. And you just need to pay for that project. And then you can almost kind of forget about it. You never see a virtual machine. 45:34.411 --> 45:54.019 Yeah, it's a web service. Yeah, because this is, as Leon showed it in his private browser, not in the TSD browser, this is outside of the web portal, it's outside TSD, but data is stored in it. 45:55.094 --> 46:14.633 Yes, then I will try to give a live demo of this new killing feature in TSD, which is autotext. We have a very popular service here at the University of Oslo called autotext, where people can transcribe 46:14.650 --> 46:44.650 recordings. And we're working together with the National Library, running all their models with Whisper. And it has been a pity that this has only been available in a complex way inside TSD. We are now happy to introduce a little text inside TSD, which is based on good user experience and web application. 46:44.667 --> 47:10.879 I'm not going to show, I'm trying to make a video here with my mobile phone. I'm using the Visio app that maybe someone has seen before, and I can hear, film and talk a little bit. And this is designed for, we'll call this Hey, and Hey and Hello. 47:10.896 --> 47:38.029 and I will send this as a sensitive video recording directly to my TSD project. So, let's see if I can transcribe this live. This will work. I will log into my project again and I will 47:38.046 --> 48:06.493 Here, go to the NetGem portal and show my forms. And the last submission is at the top. And here I can see that there has been a recording that looks a little bit like I am live or live. Here with my mobile phone. I'm using the... Always funny to listen to. 48:06.886 --> 48:35.350 So I download this file here and I will leave it in my browser. So I will go back to this and I will go to tft autotext, which is our new service that I can log in to. Everyone has access to this service. 48:35.367 --> 48:54.411 If you have 10,000 hours of video, please don't just drag it up. Then you can test out here and then you can contact us. There might be some invoicing if you have a lot of data. This is also based on this usage of Colossus, but for all institutions, this is going pretty fast. 48:54.991 --> 49:21.425 Here I can take this new file that I just showed you and I can send it to texting and now it's uploaded to use Colossus and as you can see it's already queued up in the queue for Colossus and hopefully there are some available resources at Colossus at this moment. 49:21.903 --> 49:49.599 You can see how you can use the features. We have some information about the text out on my, on, on, on web. You can Google it and it will be the most the same here because we are, we are running the same, same mods. So here I was just using, yeah, I used open whisper launch. 49:50.009 --> 50:17.961 I should, of course, use ENB. You can see how it's transcribing. Still waiting a little bit. I could use this from the National Library. That is what we actually have default in outside of TSD. So, ta-da, it's finished and I can watch. I hope I remember what I said. I said something. 50:18.422 --> 50:41.067 video here with my mobile phone. I'm using the VCO app. You can see it and you can download it in different formats as well. Hopefully this will be very welcome for all the users of TFT and 50:42.466 --> 51:09.650 If you're using the Dictaphone, for example, then you can upload your files here. And we have not made it automatically with Dictaphone yet because we think people want to test out different models, for example, and see how it works. So this is the first version and it's available for everyone. Will it replace the transcript option in NetSchema? No. 51:09.667 --> 51:31.049 The Netschema has their own transcripts in the HutoText outside of TSD. And that is not designed for the same sensitivity of data. And can you use this even without Feide? Is one of the questions. Yes. You have fun to date with TSD users. 51:31.067 --> 51:51.937 the internet schema with, for example, with your TFT users. Or you just upload files. If you have some encrypted files that you have not been able to transcribe now, it's time to create an upload link, upload the data. This scales. You can take terabytes of data, upload it, and you can... 51:51.954 --> 52:09.309 start testing it. Like I said if you have 20 000 hours of recordings please use this to test and contact us and we will help you to run these large jobs that we have done several times before for example with the therapy lessons recordings. 52:11.374 --> 52:41.374 I know that someone have been asking us about if we are making LLM models available inside TST and as we have a supercomputer that is able to run LLM models we definitely should have something. Can you say anything overall about this Leon, how we will handle LLM models and other 52:41.374 --> 53:10.452 We have at least landed a design and completed the risk assessment for that design. And it will work by us starting the LLM framework which can load different models on an HPC compute node. And that compute node will then be reserved entirely for that model. And then it will be available 53:10.469 --> 53:37.585 via API. And then we will have a web interface as you'd normally do where you can chat and interact with the model. There will not be like one big model in TFT that everyone can use. Everyone has to spin up their own models due to security issues. Well, I have to also say that we are starting with this in EduCloud. 53:38.831 --> 54:06.323 I think. Yeah, because we could have a dead view. Yeah, OK. Something can. But I don't think we have any technical blockers for doing it. It's also a matter of prioritization. How important is it for people? Or do they have other pain points in TSD that are more important to address? And yeah, I don't know. 54:08.012 --> 54:15.179 I think we will hear more about this when we... 54:16.067 --> 54:44.582 go to the next conference that we will arrange this summer. So, we have data at the UU. True2FM will be the 10th of June here at the University of Oslo. And this is the fourth data at the UU. Before that, we had some user meetings, but this is about HPC, NetGemma, and TFT. It's called user meeting. It's a little bit more... 54:44.582 --> 54:55.469 Yeah and it's very popular it will be live streaming like every year but also physically in Simula Auditorium here in the neighbor building. 54:56.561 --> 55:21.869 And we will talk more about LLMs and how we handle it for sensitive data at this conference. And there will be recording and I'm very happy to announce that we have the keynote speaker, first keynote speaker already booked in with the Hedvig Marie Egland Noreng from Pharmaceutical Institute at University of Oslo. She has been a big user of all 55:21.869 --> 55:43.353 HPC Negema and TSD for many projects. So we are very happy to have her as the first keynote. Then there are some sketches from the program here as well. And for example her koi for sensitive data. So just be looking forward for this. 55:44.667 --> 56:08.985 So, now we are trying to start with a tradition here. We can see that next year if it is a tradition that we will have a webinar after every half year and then this conference every year. So then we are able to make a small up-to-date as the development of TSD is happening so fast. 56:10.845 --> 56:29.275 Anything more that you should comment on the questions the last two minutes? One question. Can all members in a TSD project read the transcribed files when using AutoText? Or can members regulate access to only themselves or specific users when using AutoText? Very good question. It is either... 56:29.275 --> 56:48.797 yourself or the whole project and that's where we started because it was easy to achieve but we will of course that is a very good feature so you can then have many students using for example dictaphone and logging in and transcribing sensitive data and only they get access 56:48.814 --> 57:18.473 And later on, we will have group-based access so that your supervisor, for example, can access it. And we know that the national mobile app is very popular. So we hope that this also can be more popular in using for sensitive data. Yeah. Otherwise, I would say there are some very specific questions or some more technical questions. I would advise people to send us an email and ask. 57:19.155 --> 57:43.063 rather than to have them answered here during the webinar. Then on behalf of us, University of Oslo's IT department and everyone here and everyone around us and TSD, say thank you for following us at this meeting and see you at data at UiO this June.