- From: W3C CCG Chairs <w3c.ccg@gmail.com>
- Date: Tue, 27 Oct 2020 10:33:16 -0700 (PDT)
Thanks to for scribing this week! The minutes for this week's CCG Verifiable Credentials for Education Task Force telecon are now available: https://w3c-ccg.github.io/meetings/2020-10-19-vc-education Full text of the discussion follows for W3C archival purposes. Audio from the meeting is available as well (link provided below). ---------------------------------------------------------------- CCG Verifiable Credentials for Education Task Force Telecon Minutes for 2020-10-19 Agenda: https://lists.w3.org/Archives/Public/public-credentials/2020Oct/0045.html Organizer: Heather Vescent and Wayne Chang and Kim Hamilton Duffy Scribe: Present: Kim Hamilton Duffy, Leonard Rosenthal, Nate Otto, Jim Goodell, Taylor Kendall, Kerri Lemoie, J. Philipp Schmidt, Jeanne Kitchens, Simone Ravaoli Audio: https://w3c-ccg.github.io/meetings/2020-10-19/audio.ogg Kim Hamilton Duffy: Okay, IP note, anyone can participate in these calls. However, all substantive contributors to any CC work items must be members of the CC G with full IPR agreement signed Kim Hamilton Duffy: And I'll include the in the chat. The link to join. Kim Hamilton Duffy: And to do that you will need to ensure you have a W three web account for such a simple signing up for an account. And again, it's free for community groups. Kim Hamilton Duffy: So if you want to read the text of the W three see community contributor license agreement that's available at this location and pasting in call notes. These minutes in an audio recording of everything set on this call are archived at our, our GitHub repo at this location. Kim Hamilton Duffy: And we use IRC or yes we do use IRC to queue speakers during the call. So we are still using IRC today so use it if you can. It's an easier way to maintain order. Kim Hamilton Duffy: But if you try to add yourself to the queue in zoom over here, we'll, we'll try to figure it out. We'll try to move you over to IRC. Kim Hamilton Duffy: Hey, so we use IRC Q, as well as to not today, take minutes all attendees. She type presence plus in IRC to get your name on the on the Kim Hamilton Duffy: Attendee list and the transcript and in IRC type q plus. That's the letter Q symbol plus new space to add yourself to the queue with an optional reminder Kim Hamilton Duffy: And see if you're not on IRC simply asked to be put on the queue or you can type. So in the end, zoom message will add you Kim Hamilton Duffy: Please be brief. So the rest of the group get a chance to chime in and you can always keep less again. Kim Hamilton Duffy: And this meeting is held by voice, not by IRC off topic IRC comments or subjects deletion from the record. Hey, again, no scribe. Today we're trying out auto transcription Kim Hamilton Duffy: OK, so now let's move on to introductions in reintroductions Kim Hamilton Duffy: I think I see a few new aliases on the call. Kim Hamilton Duffy: Let me see actually first if if anyone would like to introduce themselves add yourself to the queue. And I'll call on you. Otherwise, I can start just calling on aliases that look somewhat new Kim Hamilton Duffy: Kristina, could I call a new to introduce yourself. Hi. Yeah. Can you hear me. Yes. Kim Hamilton Duffy: Yes. Please go ahead. Kim Hamilton Duffy: We may have lost her um Kim Hamilton Duffy: Let's see. Yeah. Can you hear me. Kim Hamilton Duffy: Yes, we can. Oh, sorry I'm against the video is a tree. Hi, my name is Kristina. Um, I worked at Microsoft, where I'm working on the US team identity standards and then you standards architect. And they do work was several customers who were getting education and credentials on space. And also an aside, other side project my passion project at work in an NGO where we do again educational credentials in Zambia and, you know, I've been super interested in joining this call, but I'm basing to it's midnight here so you know it's not always the best call time. But yeah, super happy and made it here today. Kim Hamilton Duffy: So yes, thank you so much for joining us. It is a very inconvenient time we're working on making some more time zone. Kim Hamilton Duffy: Friendly options. So we'll stay, stay tuned for that. Um, let's see, I think, I think Kostas has been here before, but maybe I'll ask him to introduce himself Kostas Kim Hamilton Duffy: Okay, we'll skip, actually. Anyone want to reintroduce themselves before we move ahead. Kim Hamilton Duffy: This can be useful if you're, you know, if you switch jobs or you're focusing on a new area and you're wanting to tell people about what you're working on. Kim Hamilton Duffy: Okay, let's just go ahead then. Kim Hamilton Duffy: Okay, so before we get into the main topic I wanted to remind everyone that, you know, so the modeling verifiable credentials for education. Kim Hamilton Duffy: Document that we've been working on reminder that it moved over to spec text. So if you click that link, you'll see the, the current work in progress. And you'll also see there's a lot of Kim Hamilton Duffy: Issues. So, actually. Kim Hamilton Duffy: Let me share my screen and forgetting that I can do modern things now. So here is the the new spec and then we were using just get hub to track everything. So all of the comments that were sort of tracking issues in the Google Doc are now over in GitHub. Kim Hamilton Duffy: So that makes it easier to kind of keep a threat of conversation and get them updated in the schema. So please review the issues here, and comment on them. I'm going to start doing that as well. I think you can sign up for notifications on the repo just by watching it. Kim Hamilton Duffy: If you're having problems. If you're not familiar with using GitHub reach out to me, we might Kim Hamilton Duffy: You know need to do a quick sort of tutorial or something like that on a future call Kim Hamilton Duffy: And if there's time at the end of this call, we can start working through some of the issues, but otherwise we will be coming back to this and future meetings so Kim Hamilton Duffy: It's a good idea, though, to start reading through the issues available and, you know, start writing any concerns or thoughts that you might have. Kim Hamilton Duffy: Okay, let me stop sharing now. And with that, I think we're ready to turn over to Leonard and so Leonard Kim Hamilton Duffy: I will you know if you have slides to present, things like that. You can use the screen share and again I will be maintaining the queue. So in general, we asked people to Q themselves so they have a concern. Kim Hamilton Duffy: You know, our question or comment as they go unless the speaker chooses otherwise. So just Q plus yourself to IRC and we'll get to your questions at the end of it. With that, I'll turn it over to Leonard Leonard Rosenthal: Hey, thanks. Leonard Rosenthal: Thanks everybody for joining today. Let me. So what I'm going to talk about. And I'll tell you where I am who I am and where I'm coming from. So Leonard Rosenthal: I'm not a I'm not an educator. I actually am not involved at all in the education space, but I do spend a lot of time in the area of Leonard Rosenthal: Verifiable what's now verifiable credentials. But I think about it, not necessarily from the point of view of the identity aspect or the credentialing aspect, but about the data and the information on that you want to store that you want to transport that you want to archive. Leonard Rosenthal: And how that particular aspect plays out. And so the reason I think about these things other than they're very interesting is that my role at Adobe is I'm the Senior principal architect for PDF and also for our content authenticity initiative. Leonard Rosenthal: And so these things overlap very well and also aligned nicely with with the work that you folks are doing as we think about this idea of what is the right container for our information. How do we want to bundle things up. And again, pass them around for various reasons. Leonard Rosenthal: So obviously, you know, we're not talking about anything new. This is not rocket science. We're not inventing the wheel. We don't want to reinvent the wheel. The, the idea of this portable database, if you will. Leonard Rosenthal: Is not new. And so we've been looking at this for quite a long time in the world in general. Leonard Rosenthal: You think about this in the world of EMR electronic medical records and the need to exchange things between systems. Leonard Rosenthal: Recording data dashboards. I'm going to use one of my particular favorite use cases is around stem publishing. So the ability to publish rich content. Leonard Rosenthal: With math and science incorporated government or industry data sets. I'll give some examples of that. And of course, what we're here for today educational credentials or credentials of various types. Leonard Rosenthal: And at the end of the day, obviously, the point here is we want to we want something that can be consumed, both by a machine and by a human. Leonard Rosenthal: Having to have two separate items means they're going to get out of sync. You might lose one. Which one is actually the real thing. So how do we bring these two things together. Leonard Rosenthal: And why PDF. So what, why are we even thinking about this as an option. Well, it should be fairly obvious. Everybody knows what it is. The internet knows what it is you can pass them around in your email clients, put them on the web. Leonard Rosenthal: You know it's been a scattered mind type for 25 years every modern iOS platform knows about it, you get a PDF viewer baked into every OS you purchase today. Leonard Rosenthal: I see you can't get away from it. It is ubiquitous as the web and web browsers are, and we'll talk about that momentarily. And it's also an open standard. It's been an open standard since 2008 so over a decade now. Leonard Rosenthal: And that means that you know we can leverage it from other open standards so we don't have to worry about any proprietary nature being provided here. Leonard Rosenthal: Also want to introduce to you. For those of you may be, may not be familiar. In addition to the whole PDF standard, which we call ISO 32,000 that's the number at the ISO Leonard Rosenthal: There's a number of what we call subset standards, where we've taken the big PDF and we've reduced it we've said, What are the core functionalities that are needed for specific use cases. So the two I want to bring up one is PDF a PDF for long term archiving. Leonard Rosenthal: This is a standard that his name implies, is how do I use PDF for storing documents long term. And in this case, long term. It's not five or 10 years Leonard Rosenthal: But we're thinking 50 years 100 years, have you think about documents that need to be stored. So my two favorite examples here in the United States, the Leonard Rosenthal: National Archives are by law required to keep documents until the end of the Republic. So with the assumption that the Republic, the United States isn't going away anytime soon. That's a fairly sizable lifespan. Leonard Rosenthal: But it actually doesn't compare to the Nuclear Regulatory Agency, which is required to keep documents for the for one half of for the afterlife. Leonard Rosenthal: Of each of the radioactive isotopes. So for any of you who know your science, you're talking potentially 10s of thousands of years. So we think real long term. Leonard Rosenthal: The other thing, of course, it's extremely important that I know many of you on the call are deeply involved in is accessibility and so PDF UAE is the PDF standard for ensuring universal accessibility. Leonard Rosenthal: And these standards have been around quite a long time and they are mandated by law, much like wake AG is mandated so as PDF you a PDF also in over 50 countries and all of the states here in the US. Leonard Rosenthal: So fairly well known. If you're not familiar with them to know. And obviously we're here as part of the W three see. So let's talk about the open web. Leonard Rosenthal: PDF is a part of the open web platform. It has been part of the web for more than 15 years as I mentioned before, it is natively supported by all major web browsers. And it's also a Leonard Rosenthal: Normative reference in the HTML5 specifications. So if you crack open that speck in your search for PDF, you'll find that a couple of times. Leonard Rosenthal: So it is part of the wet. Leonard Rosenthal: And of course there are thousands of independent implementations and this goes to the fact that it's an open standard, whether you're an open source like commercial you have your choices. Leonard Rosenthal: So I mentioned. So that gives us your why are we thinking about PDF. How does it fit into the ecosystems. There are clay here. Leonard Rosenthal: So this idea PDF as a container. This is not, it was not something I necessarily I'm coming to you for the first time with Leonard Rosenthal: It's something I've been working on for quite a long time. As you can see here, one of the first uses of this was for the US Census data. Leonard Rosenthal: That GPO the government printing or government Publications Office here in the US started bundling the census data inside of a PDFs in 2003 and then they proceed to digitally sign Leonard Rosenthal: Or certify that document, they really were at the bleeding edge of this technology and have been doing it since. Leonard Rosenthal: The Brazilian government. I worked with also in the mid 2000s, to ensure that all of their transcripts. So here's an example where they take Leonard Rosenthal: They recorded every one of their meetings in an XML grammar that they define they take that XML file, it gets embedded into a visual representation. That's the PDF Leonard Rosenthal: Again they sign it. And this goes up on their websites. So if you want to get any transcript from the Brazilian government of any government meeting. That's how these things have been distributed since 2008 on the EU patent office has been doing something similar. Since 2012 Leonard Rosenthal: On perhaps the most recent set up things as around the invoicing so in 2014 the German government created a German standard called zero preferred Leonard Rosenthal: The French followed suit. Although it took much longer to get it out with something called factor X on both of which align with the EU invoicing standard which is een 16 931 Leonard Rosenthal: And basically this is the same sorts of things. I've been talking about, which is let's take an XML chunk of XML data. And again, XML is not special, as you learn these are just all folks that have chosen XML with their format. Leonard Rosenthal: We take a chunk of XML. We put it inside of a nowadays a PDF a format. So going back to that long term capability. So we want to ensure that these invoices. In the case of the EU and German and French standards will live for a long time. Leonard Rosenthal: So we're going to put that in there and now we've got this long term archival document it meets all of the mandates of the various countries, Germany, France and most actually I believe all of you at this point. Leonard Rosenthal: I'm now has PDF a mandated as well. And we get this beautiful container of both human and machine readability. All right. Leonard Rosenthal: So these are great examples. How does this apply to some of the things you folks are thinking about, well, I was doing some research and apparently the folks at Open badge factory had the same idea. So like I said, this is not new. I'm not trying to, you know, give you anything new. Leonard Rosenthal: So I found their web page. Leonard Rosenthal: Where they talk about exactly what I am, which is you start with an open portable format which supports rich metadata digital signatures. Leonard Rosenthal: You bake your badge into it and away you go. Now, I couldn't find any details of how they're doing it so Leonard Rosenthal: As I go through the rest of the presentation, I'm going to talk about my ideas as to how I think it should be done and why and the pros and the cons, but again, somebody else has already started going down this path. And I think that's wonderful. Leonard Rosenthal: And you can kind of think of this, although I you know I think it's fairly obvious. If you're familiar with the IO R le are rappers in your community. This is the reverse. In that case, Leonard Rosenthal: There you put a PDF inside of the JSON. Here we're going to put that JSON or XML inside of the PDF. So all we're doing is reversing. What is the container or the rapper technology but same idea, the same principles. I'm sorry. Is there a question. I'm happy to take it. Kim Hamilton Duffy: Um, I think someone just didn't meet themselves. I'm going to go through and check for that. So, go ahead. I'll see. I'm sure everything's muted. Leonard Rosenthal: Right, no problem. Right. And we keep going. Leonard Rosenthal: Alright, so Um, Leonard Rosenthal: Again, why are we doing this. So just like the Dodge factory said PDF is a great format because Leonard Rosenthal: Not only represents content that's what we all know in and are familiar with it for, but what most people don't know. And I'd like to spend some time really talking about Leonard Rosenthal: Our this semantic nature PDF and the metadata capabilities, a PDF, because that's really where we're going to leverage to connect the semantics and the metadata information coming from the BCS Leonard Rosenthal: To the PDF content. And so why other people thought of this before. Really, it's because people just don't know what PDF can do. We're used to getting Leonard Rosenthal: These very simple documents without structure without metadata and we're just viewing them. Historically, we would have printed them out. Leonard Rosenthal: And then we sort of were done with them. We just don't think about them because most people don't produce the kinds of PDFs that we want to produce. So let's talk about what we can do, when we do it right. So let's do that. Leonard Rosenthal: One of the key things we want to start with is the metadata platform that PDF leverages so this is also a standard of its own. It's called SMP the extensible metadata platform. So I so one 664 Leonard Rosenthal: It comes in both an XML serialization and a JSON LD serialization, so it's a data. It's a single metadata model. It's based on RDF it's compatible with RDF Leonard Rosenthal: And we take that and again we can see realize it either is XML or JSON LD, and it is doesn't matter the serialization is irrelevant to the data model that share Leonard Rosenthal: This same standard SMP, not only is used by PDF, but it's also the standard for all common asset types used on the web and in general today. So if you go look inside of a JPEG or a pain or an MP4 file or, you know, add one app for whatever your favorite Leonard Rosenthal: Format is for images audio video 3D um it's now part of GL TF. For example, which is the up and coming 3D format standard Leonard Rosenthal: Uses x m p for its metadata. So this is the standard for asset and document metadata. I'm SVG, for that matter, supports SNP Leonard Rosenthal: As its metadata format, although it's not used ritually in that world. So great format. It's indexed by the web and by desktop search engines. Again, not only in PDF, but in these other various formats. Leonard Rosenthal: So we got a great standard to build on inside a PDF SMP exists in two different locations conceptual one, is it the document level. So I can say here's everything about the document. Here's my ex MP for the document. Leonard Rosenthal: Go. I can also attach it directly to individual objects. So this comes in handy when you think about, for example, and we'll talk about this combining multiple documents together or perhaps putting Leonard Rosenthal: Images are multiple images on a page or across the document or each one of those images, each one of those objects, each one of those pages has their own SMP associated with it. Leonard Rosenthal: And that gets carried along throughout the process. So we don't necessarily have to worry about one and only one chunk of metadata. We have it bound to each individual item. Leonard Rosenthal: And that's not unlike, of course, how you want it on an HTML page connect schema.org information to individual objects within the HTML page. So very similar concepts. Leonard Rosenthal: Apply here only differences. Again, we've standardized on this SMP I've also been working so they're just mentioned schema.org I've been working for the last few months. Leonard Rosenthal: With the folks@schema.org because we now have this JSON LD serialization, we're trying to coordinate how we take various schema.org schemas and properly serialized them index empty. Leonard Rosenthal: I'll point out, and it's working, it's going along, admittedly, slowly. The biggest problem we're running into just as an FYI. Leonard Rosenthal: Is that even though it's called schema.org there actually are no schema files for any of the data that's up there. So whereas for x m p we have schemas for everything. Leonard Rosenthal: We don't have that from schema.org so we're trying to figure out how we utilize these non machine readable descriptions of the various schemas and how we can apply them in this very machine readable fashion so SMP the core metadata within PDF Leonard Rosenthal: Um, I talked about structure and semantics. So this is something, again, most people don't think about but PDF has the concept of logical structure and structural semantics. Leonard Rosenthal: And if you look at them. We call them tags. If you look at them they map really nicely to HTML tags and that's by design. Leonard Rosenthal: This goes back to, we did this actually back in the HTML three days just came out in 2001 and it's evolved since and semantics don't change, there are simply course semantics paragraph headings list tables, tables of contents. We do a few things that Leonard Rosenthal: We've aligned nicely with the ARIA D pub work so on both ways. Ra di pub reflects some of the semantics that we have in PDF that weren't in HTML. Leonard Rosenthal: And we've also gone the other way. So we actually support ARIA D pop as structural elements as structural elements within PDF Leonard Rosenthal: So you can leverage some of those as well. So this way you have the semantics that you can express in your web content. You can also express directly and equivalent Lee in your PDF content. Leonard Rosenthal: And so there's this really nice model. One of the things that we allow in PDF that I know is not a always supported in the HTML world, but we think is extremely important, especially as we think about specific use cases like we think about education. Leonard Rosenthal: And the like sometimes you want custom tags. You just need them. Um, you need to be able to reflect your own things and then map them accordingly. So here's a couple of examples. Leonard Rosenthal: For example, I want to call this a diagram or I want to call this a Drop Cap, because that's semantically really what it is, but under the hood. It's a figure Leonard Rosenthal: So that's okay um footnotes and then notes, of course, a really good example. That's something where r e a d pop has certainly addressed some of the concerns. Leonard Rosenthal: And the list goes on anything that you want to address through custom Leonard Rosenthal: Tags. You can do that in the map them equivalent Lee and that's how for example, we were able to do and I'll refer back to our a deep up and already itself. Leonard Rosenthal: So in PDF, the ARIA tags are custom tags set because they're not part of the core PDF Tag Set, but that's okay because again we can utilize custom tags and therefore incorporate any of these additional things that we want. Leonard Rosenthal: And this would also potentially leverage if we needed one specific to the educational community. Leonard Rosenthal: So how does this play out. So I apologize. I just did not. I didn't have time to actually redo some of my examples with a specific in this case. Leonard Rosenthal: With a specific educational credential, but I'm hoping that you'll get the idea from from my previous stem presentation samples. Leonard Rosenthal: So the idea here is, again, we have these tags on the lap. And the idea here is to point out that the pieces of the document. Go to the tags. So, some of them are obvious like headers and paragraphs, but Leonard Rosenthal: Tables figures those things kind of obvious this, you know, not a big deal to should be what we think of as semantic mappings. Leonard Rosenthal: Where it becomes interesting though is when we take this the next step, and that is the PDF as a container which is again where we're going for here. Leonard Rosenthal: I can embed any type of file any type of data that I want in here. So here we see an example or I have a bunch of XML files. I have some SVG files. I have some CSV files embedded into this PDF Leonard Rosenthal: But just the during bed, it is irrelevant. It's that's not as important as the fact that they are connected directly to Leonard Rosenthal: The visual and semantic elements so that table of data is directly connected to the CSV file that represents the data. So not only do I have that visual representation, but I have the actual data that most likely created that table, or at least represents that table. Leonard Rosenthal: In the case of the math and now at the math at the bottom. I have the math ML embedded directly into the PDF. So again, not only do I have visual representations, but I have full semantic representations Leonard Rosenthal: In other open standards tightly bound, so I know exactly as I walked through the PDF, as I look at each of the components. I know what things are connected to what and one of the advantages of Leonard Rosenthal: That is you put Leonard Rosenthal: This together. You put together the common semantics with HTML, the ability to bind these other standards to the individual elements. Leonard Rosenthal: Then we now have a standard for what's called deriving HTML. And what this means is that rather than having to do arbitrary conversions of this stuff and getting non deterministic results. Leonard Rosenthal: I can take a PDF. I run it through this well defined standardized algorithm. And I get HTML out in a deterministic manner. In fact it round trips, both ways. So now I can go PDF HTML. HTML to PDF maintaining the full semantics. Leonard Rosenthal: Including additional attributes. So for example, if you look down. You see the language. So language tags, for example, those are things that were present on the on our semantic elements. Those come into the HTML. Leonard Rosenthal: On the figure. If you look down at the figure, you'll see things like placements. This one is obviously a bad example from an accessibility standpoint, it doesn't have all text associated with it. But that of course would come over Leonard Rosenthal: You'll see them in the case of the image, rather than bringing over a raster image because I had the SVG, I can just bring over the SVG. Leonard Rosenthal: Same thing with math. I know I can bring the math. I know over and so I'm no longer bound by this, you know, people describe PDF sometimes as we're data goes to die. Leonard Rosenthal: That's because you just don't know what's in there, you know, and as you see none of this is new. This is not stuff that just showed up in PDF yesterday. Leonard Rosenthal: All of these are things that have been there for years, decades. In some cases, they're just not leveraged and what we're trying. One of the things that Leonard Rosenthal: I've been trying to do for many years, and there are others amongst us doing this is again going out there and talking about what we can do and talking to communities like yourselves about how to leverage it so Leonard Rosenthal: You know, as I mentioned, even even individual attributes, I talked about these are deeper even things like already a fan. So, for example, Leonard Rosenthal: Here on the right is what the Wikipedia page for RDF a looks like it even has an exhibit a sample that's the actual sample off the page. Leonard Rosenthal: You can see that the PDF equivalent, as I mentioned, I can Leonard Rosenthal: Go. Leonard Rosenthal: Bi directionally and maintain all of those semantics. Even the RDF so you can see that even the individual RDF a items. The fo as properties. Leonard Rosenthal: The relationships. Everything directly out of the HTML into the individual semantic components of the PDF one to one mapping Leonard Rosenthal: So very, very huge opportunities here for how you want to express all of your information and encode it and we'll talk about that. So I really just wanted to set the stage for all of these things that are possible based on these PDF technologies that most people are not aware Leonard Rosenthal: One more. We were talking about this is of course very big in the open data and linked data community is linking Leonard Rosenthal: You don't just want to have everything internally. So you can link, of course, both into a PDF out of a PDF and within a PDF Leonard Rosenthal: I'm all using the common models. So you can do this either using native PDF linking capabilities or Leonard Rosenthal: We fully support you. Our eyes with fragment identifiers and that's codified in RFC 8118 and replicated in the ISO 30,000 standard as well. You can point to pages. That's old school. Leonard Rosenthal: But you can also point to individual structural elements. So if you want to link directly to a table or a cell of a table or to a figure Leonard Rosenthal: On you can do all of that. You can also point to the embedded files themselves. So if you wanted to link all the way into, say the JSON. The represents your credential that's embedded in a PDF. The I posted on the site somewhere. Leonard Rosenthal: No problem that is fully supported in the fragment identifiers Leonard Rosenthal: And so that, again, you can host that PDF with your embedded JSON and the PDF visibility, but have the links, not just to the PDF that's nice. Leonard Rosenthal: But I actually want links into the PDF and into its content and into its semantics and into that JSON embedded data. For example, representing the credentials. Leonard Rosenthal: And so you can, and that also works. Of course, the other way around. If you want a bridge from your embedded data to your content. Content to the embedded data and back and forth so very, very rich models for Lincoln as well. I mean all standards based Leonard Rosenthal: And of course, one last set of things we want to talk about is, and I mentioned this earlier digital signatures. Leonard Rosenthal: PDF digital signatures are standardized both at ISO and 1453 and with Etsy for AI this and what's called the pod as standards. So you may be familiar with the other artists standards caught as and shot as or Katie's and GDS depending on where you come from. Leonard Rosenthal: Pot is is the PDF version of that. It's been an Etsy standard for almost 15 years now. I think it's been that long since we we started that process. So quite a while. Leonard Rosenthal: And all of the things that one would expect to see in the digital signature standards so Leonard Rosenthal: We have approval signatures. We have certifying signatures. We have cereal signatures time stamping capabilities and again all standardized with the various Leonard Rosenthal: Organizations that one would expect. So this not gets us take that and put the signature put the binding around it to ensure not only who did it come from, but whether or not it was campus. Leonard Rosenthal: PDF does also support security encryption and DRM but I'm not going to say much about that. I know many people have feelings, one way or the other. So we'll ignore the political aspects of that and just say it's there. Leonard Rosenthal: One other aspect which is not a feature a PDF, but I wanted to talk about, because I do think that in the aspects of education, as I think about at least Leonard Rosenthal: How I see I'm predict educational credentials being used to me. One of the things that is a key feature PDF that I think will be a big when Leonard Rosenthal: I'm in the educational community's ability to aggregate your credentials together. So I think about this, for example. Leonard Rosenthal: As one of my kids is putting their resume together. In addition to having their resume in that same PDF, they could bundle in Leonard Rosenthal: There the crypto their credentials from their university. They can bundle in the credentials from the individual Coursera or whatever classes that they've taken and then that goes along as part of their resume packet Leonard Rosenthal: And PDF that's just logical everybody thinks about the ability to do that. They do it almost without thinking it's it's really one of the most common operations done Leonard Rosenthal: And so to me that ability to do that to create that custom CV your resume using off the shelf tools. We don't need a custom service. We don't need a, you know, Leonard Rosenthal: I have to go to a specific place if I have my PDF of my credentials, which hopefully I do. That's part of being able to have them portable that I can store them. I can keep control over them. Leonard Rosenthal: And I can bundle them together in the way that I see fit and distribute them as I need to. Leonard Rosenthal: That's a big wind of this whole open air. Whole idea of open credentials and sharing that information on in relevant manner. So I wanted to point that out. Leonard Rosenthal: To me, at least, and you guys may tell me I'm completely crazy, but to me. I think that's a great use case for the use of PDF, especially for VC and education credentials. Leonard Rosenthal: Okay. Leonard Rosenthal: So how do we get there. Leonard Rosenthal: Can we just stop in any questions so far. I do want to make sure everybody is okay on the technologies and stuff before we sort of I put them together into what I think is a least a couple of good suggestions. Kim Hamilton Duffy: Yes, thank you. This is great. And actually we do have someone on the queue. I'm sorry my cutie and Nate Europe. Nate Otto: Hey, thanks. Um, some really interesting stuff in this presentation and I mainly so I'm pretty familiar with the open badge factory implementation. Nate Otto: The Open Badges work group to review it and update the validator to be able to accept that format for a Nate Otto: Container for Open Badges data. In addition to SVG and the N G which are currently supported, but their approach really has a separation between the data that is Nate Otto: The Open Badges actual metadata and the visual presentation Data, whereas a lot of the slides that you've been showing here really shows that there are potential possibilities for actually presenting the verifiable credential data in the visual format and not just using the Nate Otto: The PDF as a transmission format essentially for the real data. In the case where like open badge factory does where they're not actually displaying the credential data there. They're just packaging and in there in the metadata. Nate Otto: I'm struggling to understand like Nate Otto: Why it's useful to try and go further and display the actual data on that is in the VC in the human readable. Nate Otto: Content of the document. Is it the same data or are you just read capitulating the data is sort of my main question to you. Leonard Rosenthal: Oh, such as a question. Yeah, it's not so I would say. And actually, all I have an example of Leonard Rosenthal: Open Badges how for example i would i again i didn't see their specification. But if I were doing Open Badges for PDF how I would do it. So I'll actually show that. Next, I think, is my next slide. Leonard Rosenthal: Um, but to the question. It's not the same data. There's data bundled in JSON or XML. There's data on the PDF side so that is two separate pieces of data, but they can be again tightly bound Leonard Rosenthal: To each other. And the reason they're not the same data, you know, there's pros and cons to that one side says well you edit one you want the other edited, but I look at something, especially in this particular context. These are not documents, you're going to edit. Leonard Rosenthal: Their somebody is delivering to you in most cases you completed a course you graduated from university. Here's your diploma. Leonard Rosenthal: Type of thing. Here's your, your certificate in editing it would be bad. And that's part of why we want to sign these things. So the fact that it's two pieces of data. I don't think it's a big deal because it's Leonard Rosenthal: Not to be edited, but what you want to do though is connect the dots and the reason to connect the dots. Leonard Rosenthal: Is for these downstream workflows. So for example, the aggregation workflow case I gave you before. If I want to put Leonard Rosenthal: Up. I have three or four badges and I want to put them all. I want to aggregate them onto a single page. So for example, I want to go into Leonard Rosenthal: Google Docs or, you know, pick Apple pages or whatever it is. And I want to place, you know, for these badges on a page with a nice big title you know my accomplishments for 2020 Leonard Rosenthal: I need to be able to do that. And then when I produce the new PDF out of that I need those badge date that badge dated go with it and Leonard Rosenthal: You know, in a sense, that's what you accomplish with the pinion bedding and the SVG embedding today or baking to use your term in Open Badges Leonard Rosenthal: We need to be able to do that same thing in PDF. So when I place those PDFs on another page and I produce another PDF. I needed to go with it. Leonard Rosenthal: And that works. If it's the whole thing. But now. Okay, so that's easy if each of those badges was individual now I've got a document which has four badges on it to follow my example, each badge with the big data. Leonard Rosenthal: Now I need to take one of those out for some reason. Again, I need. It's not the whole PDF. I need to stay with the badge so Leonard Rosenthal: Same model, I think, you know, it's not a thing. It's the same model that you want to accomplish with those other formats, we want to do with PDF Leonard Rosenthal: Badges tougher is we'll see better not tougher badges is, I hate to say this less interesting because it's just a picture or an image of figure Leonard Rosenthal: But when we think about something like and I apologize. Actually, this is the one example I wanted to do and I just did not get a chance to do it with other things going on. Leonard Rosenthal: Um, you think about a diploma. To me, the diplomas and more interesting case because it's got text. It's got connection information to the university. It's got all of these other visual elements. Leonard Rosenthal: On the page, other than just a simple picture. So I think that's where it becomes more interesting than Nate Otto: Just a simple picture. Nate Otto: Yeah. If you've read Open Badges at all. Leonard Rosenthal: Sorry, and visual. I'm sorry. I apologize. I didn't mean to. Leonard Rosenthal: Denigrate my point was it at least the the examples that I've seen. And I said, All I've done is read the papers and the pages. Leonard Rosenthal: They're usually some sort of visual graphic. It's not like a diploma wouldn't normally be used as a badge, but I could please tell me I'm wrong, like I said, it's not my area. Nate Otto: Okay. Leonard Rosenthal: Great. Good to know. Thank you. I appreciate that correction. Kim Hamilton Duffy: We don't have anyone on the queue. So I'd like to go ahead and Leonard Rosenthal: Go with the rest. Absolutely. Alrighty. Okay, so, um, we talked through this. The whole idea is let's not reinvent the wheel. Leonard Rosenthal: Okay, I think I made that point clear. So let's actually talk about how we might not reinvent that wheel so Leonard Rosenthal: I actually went, as I said, I thought I would try to use some existing examples. Leonard Rosenthal: So the one I did get through was. I thought I would redo, at least in my mind, the 3D print master example from the Open Badges version to specification. So I picked the graphic of my own because there wasn't a graphic provided so I grabbed the graphic. Leonard Rosenthal: I took the JSON right out of that particular document and I embedded it into the PDF. So this is what we've been talking about before. I have an embedded JSON. Leonard Rosenthal: I've connected that as I mentioned before, directly to that image so that images has directly bound to it that JSON and PDF binding terms. Okay. Leonard Rosenthal: The other thing I've done is I've given it a relationship. So in PDF terms when you attach something, you have the option of giving it a relationship. Why did you attach it Leonard Rosenthal: And so in this case I gave it a customer relationship. So I wanted anybody coming along later either a human or machine to understand what the point of this embedding was, why did I bother doing it. Leonard Rosenthal: So in this case I gave it a. I said, You know what, it's a VC egg credential. So the, the four characters and the underscores is just PDF Leonard Rosenthal: Syntax that's a requirement for custom data is you have to prefix customer information with a prefix there so Leonard Rosenthal: I picked VC D. I thought that was a good choice. And then this is the credential. So again, it can be anything you want. I just pick something to to represent it. And the point is now anybody or any machine coming to this. Leonard Rosenthal: Knows exactly what this data is what it represents. And how it's found. Um, the other thing I could have done and I didn't Leonard Rosenthal: But, and would have been very easy as I also could have added the schema, so I couldn't or set of schema. So I could have directly embedded all of the necessary schemas, the JSON LD context files. Leonard Rosenthal: Here into the PDF and we already have a standard relationship type, which is schema. So I kind of put in all the schemas. Leonard Rosenthal: And marked all of them as schemas as well. And then it would have been a completely self contained Leonard Rosenthal: PDF that wouldn't even even, even for the purposes of resolving the app contacts in the JSON LD, we wouldn't have had to go on out for any external reference Leonard Rosenthal: So again, I didn't do that. In this example, but it's certainly something one could do. So what are the advantages of this approach, as I mentioned, it's compatible with the aggregation workflows that I talked about earlier. I think that's extremely important. Leonard Rosenthal: And then, the fact is that it's searchable by PDF tools. So today, PDF tools, you know, your viewers your browser will search inside of attachments Leonard Rosenthal: Unfortunately, it's not a future of web search engines. It's something I've been trying to get Google and Bing and others to support for years and and I'm hoping that Leonard Rosenthal: Someday, that they will do that. I think we need a good use case. And maybe this is one of them to get them the web search engines to start looking in PDF attachment. So a Leonard Rosenthal: Lot of great benefits to doing this approach, but I wanted to make sure you understood. It's not 100% perfect that does have potential downsides, again depending on what your pros and cons, what you're looking for, as your requirements. Leonard Rosenthal: So, the alternative is to put it all inside of the X MP, so we can take it and we can embed it either into the document either into the document level SMP or into the Leonard Rosenthal: Individual object level SMP, we can do that either as a single see data block. In fact, if you look at, at some of Leonard Rosenthal: The ways that some people embedded. For example, if you look at the SVG making spec for Open Badges. That's how they put the JSON in desk VJ Leonard Rosenthal: Absolutely nothing wrong with that. If you're starting with XML data. For example, you might, however, want to simply incorporate that into the XML of the SNP, you would have to do a little bit of transcoding Leonard Rosenthal: If you can't use it. Exactly. Because you have to comply with the RDF syntax, but it's not a huge lift, Oregon. You could also do that same see data block, even for XML data wouldn't be a big lift. I prefer the latter. And part of the reason for that is going to Leonard Rosenthal: That web search if your goal is to get this searchable and index the BI Desktop and web search engines. Leonard Rosenthal: Then you want it, then you don't want it in a big see data blog because they're not going to search it index that but if you've got to properly transcoding into the SNP, then you'll get that Leonard Rosenthal: Automatically by the search engines on the other thing thats related here is there's an open issue. Leonard Rosenthal: With the WG on trying to expose X MP directly to the browser and use your agents and even to the JavaScript. Leonard Rosenthal: And so if this if we can get support for that particular issue and get that adopted then having this information in the SNP also means that your breath. You can write scripts and do everything directly in your pages to actually build on that embedded data. Leonard Rosenthal: And as I mentioned before that now works, not only with this model of putting it into the SMP not only works for PDF. But you can do the exact same model, the exact same serialisations for again pick your favorite asset format jpg ping web P, etc, etc. Leonard Rosenthal: And that way you got one model one approach, regardless of format. So rather than having to do something different for paying something different brands VG something different for PDF one model fits all. Leonard Rosenthal: The other advantage of that is. And again, this is why we're trying to get it. We're starting to see this already happened. It's been happening on the web. As I mentioned in search engines for decades. Leonard Rosenthal: But now we're starting to see other people expose it and the most recent one is Google image search Leonard Rosenthal: So before the data was indexed by Google image search. So you could search from the various things. But Google didn't expose any of that information and they just started doing that recently. Leonard Rosenthal: With their new license double images feature. So now if in this case if your image had license data they displayed in Google image search so Leonard Rosenthal: In the future, the more information we can get out there and standardized formats like this and we can start to get the search engines, the browsers to understand it and to expose it Leonard Rosenthal: Then we can continue to build on that and we're, again, we're not creating anything new. We're leveraging a set of existing open standards. Leonard Rosenthal: We're just evolving that and trying to connect the dots, really, we got the standard over here and the standard over here. We just wanted to, to talk to each other. Leonard Rosenthal: We can really build up this ecosystem around embedding this information into arbitrary asset types that are all part of the open web platform. Leonard Rosenthal: The only downside today. So this is great. And I believe read I truly believe that these are the directions and i and i like i said I'm pushing a lot of these myself and various other groups. Leonard Rosenthal: The one downside to this is today, it makes aggregation harder. So we go back to the aggregation use case. Leonard Rosenthal: It's not impossible. It's just a little more difficult. And so there would be some work that would have to be done. Leonard Rosenthal: Honestly, in the PDF side specifically because that's what we're talking about aggregation to just make aggregation more aware of this particular scenario but that's actually easy to do. Leonard Rosenthal: It's not rocket science. We could actually make those changes into the PDF specification going forward. Leonard Rosenthal: And then we would get all of the advantages and none of the disadvantages. And so, you know, directionally. I think this is something that, as I said I wanted to push. I wanted to put forward and and the pros and the cons accordingly. Leonard Rosenthal: So that's what I wanted to talk about. I really wanted to just give you that lay of the land for PDF and and how you know I see it fitting into your ecosystem. Again, I, I'm a Leonard Rosenthal: I'm a guest here. I know you're I know, as I said, education is not my area, but I'd like to think the other areas are Leonard Rosenthal: So hopefully this is giving you guys some some thoughts and I look forward to working with you and hopefully trying to make some more. All of this a reality. Kim Hamilton Duffy: Thank you so much letter that was very informative. Kim Hamilton Duffy: I actually had myself on the queue and feel free to add yourself to the queue to ask any questions so I you Kim Hamilton Duffy: Were talking about digital signatures are starting to one thing I wanted to ask about their just mostly in terms of my lack of awareness of, you know, what's being done in PDF. So okay, so Kim Hamilton Duffy: Let me, let me walk through my reasoning so naive Lee, it seemed like you know in terms of what verifiable credentials support right now, in theory, you could just, you know, sign a verifiable credential embed that in a PDF Kim Hamilton Duffy: But that's, that's a limited that's some value but you know there's value in having a signature on the whole PDF document. Right, so that gives additional Kim Hamilton Duffy: You know tamper resistance properties to the whole package. Now, that was the area where historically, I was unsure about like, which ones were sort of more Kim Hamilton Duffy: You know, centralized solutions like our flexible digital signature formats supported and what would it take for those to sort of be compatible with Kim Hamilton Duffy: You know the verifiable financial data model. I think that the X MP option to that you described at the end might be the the clearest path for compatibility with the verifiable credentials data model as it exists right now. I was just curious for your thoughts on that though. Leonard Rosenthal: So that the answer is that it is, you know, the there's nothing about the the VC data model that is incompatible with your PDF, the PDF signature model, the only Leonard Rosenthal: I would say the only difference is that PDF supports a I'll say more limited set of signature algorithms. And that's because again because PDF has to serve. Leonard Rosenthal: For the longer term, we don't just adopt the latest and greatest Leonard Rosenthal: We take a very long term view of any technology that we adopt and so the signature technologies in use today and PDF or the classical one. So this is traditional you know RSA public key, it's elliptic curve cryptography. Leonard Rosenthal: All of which are great well established and again internationally standardized algorithms. Leonard Rosenthal: A lot of the newer blockchain connected, if you will. Algorithms are not supported because again, they're just newer they haven't gone through the same levels of adoption and international standardization. Leonard Rosenthal: On that the others have. And so that's why we've chosen the ones we've had, but to that end I will say that one of the things that I'm working on. So I mentioned these Leonard Rosenthal: The Etsy jada's standards, the advanced signature standards from Etsy. One of the things that I've been working on is a job. So in order to align Leonard Rosenthal: The current support for J WT and AWS in the seas and the JSON based serialization of the that model you need an equivalent in JSON, if you're going to do signatures in the EU. Leonard Rosenthal: Or you can't do it without it has to comply with these standards. So I've been working Leonard Rosenthal: As a member of axiom and and the, you know, VC community to connect those dots and to get that Jada standards. That's in process right now. Leonard Rosenthal: And that, again, they're all aligned. It's really just about how do you put it, is it in a PDF is it in a CMS is it in JSON otherwise all the signature bits are the same. Does that make sense. Kim Hamilton Duffy: Yes. And in fact, I'm so that's good to hear that you are already working on that because that is the Kim Hamilton Duffy: I'm not pronouncing is Jada Jada is the x ad S one, because we had identified that in this group as the fastest fastest path to allow verifiable. Kim Hamilton Duffy: How you say to support an addendum to the verifiable financial data model that would allow use of LD signatures, you know, based on XML serialization verifiable credentials. So I think we should sync with you on that because we're very interested in that approach. Leonard Rosenthal: Absolutely, I'm happy to have a conversation about this and again using PDF and pot is get you that same result as well. So this is where you could even Leonard Rosenthal: Take the JSON today, put it inside of the PDF sign the PDF with pot as and you get the same alignment and benefits. But yeah, we can all because the shot as hot as and and Jada is coming up are all aligned. It all works. Very nice. Yeah, happy to have those conversations Kim Hamilton Duffy: Great. And we have one more person in the queue. Chris when you're up. I Kim thanks and Thanks Linda for presenting this. This is really good to see. I wanted to follow up on what Kim said Only because I'm not familiar with it. Can you bring your own keys to a PDF signature. For instance, you know, a Fido to Yuba key type of situation or, you know, smart card Estonian ID type of signature and still have a PDF format. Absolutely. Leonard Rosenthal: Yeah, yeah. The keys can come from anywhere. We don't the format doesn't care where the keys come from. It's simply the algorithms that you use, along with the keys and in some cases those keys specify the algorithms. Leonard Rosenthal: So yeah, not a problem at all. In fact, the European smart cards, for example, European ID cards. Leonard Rosenthal: Government US government cards, all the news for decades. Yes. No problems takes worse. No, good question. Kim Hamilton Duffy: Well, we don't have any Well anyone else on the queue. So I think I'd seen a question about advancing the work items on the spec. So in the couple minutes we have left will turn to that and thank you again. Leonard. This was incredibly useful and looking forward to continuing the discussion. Kim Hamilton Duffy: Um, so yes, Nate. I saw you in in IRC talking about the VC do models spec and you know how we can use the GitHub repo to advance issues. So yeah, just I'll cue up since we're limited on time. If there's any specific thing you wanted to discuss Nate Otto: Sure. And apologies if my internet connection is a bit unstable. Nate Otto: And also Thanks Linda for a great presentation. The so there's a bunch of issues on there. I'll put a link to the issues page and some of them. I see you've created Kim as discussion threads for various different Nate Otto: Sections of the document, and I would like to try to make some good progress over the next few weeks couple meetings where we Nate Otto: Advance. Some of these different things. One of the, the efforts that we've been talking about his use of the schema.org has credential property. There is a proposal to add has performed. I'd like to Nate Otto: Have that flushed out, get some more justification and like, what is it that we're going to put into a schema.org pull request, what in the schema.org world is the object of that. Nate Otto: That claim and then I just wanted to point out everyone else to the fact that there are discussion threads. Maybe we can make some progress by Nate Otto: Having some good discussion about these different facets of the documents so that we can take the next steps on each of these things and also start looking across the different examples, a little bit better at. See if there's consistent models that we can use. Kim Hamilton Duffy: Thank you need. Thanks for drawing attention to that. Yes, I am hoping people will start taking a look. Start thinking about these, we will be coming back to these in future meetings and then also Kim Hamilton Duffy: If there's interest, we certainly can meet outside of this ongoing meeting to just have working sessions on those. So yeah, we'll come back to that. And we're at time. Thank you so much everyone and thanks again Leonard for the great presentation. Okay. Have a good week. Everyone
Received on Tuesday, 27 October 2020 17:33:38 UTC