 I'm Saiz Choudhury, I'm the Associate Dean for Research and Management at the Sheridan Libraries at Johns Hopkins. I'm joined by Mark Patton, who's one of our software developers, one of our other software developers, John Abraham's is in the audience as well. We're here to talk about a project that is being funded by the Andrew W. Mellon Foundation called a Framework for Annotation Interoperability. I'll be giving you an overview of the project in a description of what we're trying to do and then Mark will dive into some of the technological aspects and details. Fundamentally, we've been working for many years with the Perseus Digital Library based at Tufts University. At Hopkins, we've been dealing with materials that come from the medieval and Renaissance periods, the manuscripts and early printed books, and looking primarily at images from those collections. Perseus has spent even many more years looking at texts from the classical period. There are many reasons I really appreciated Gardner's talk. One of them was he showed screenshots from some of these manuscripts and early printed books, and there are notes and annotations everywhere. I think there may be billions or trillions or however many there are. Many of them don't have anything to actually do with the content in the books. There are shopping lists, there are symbols, there are things of different orientations. I showed, we had AG assumptions, if we took this kind of approach or built this kind of technology, how would that play with what you're doing now? Would it break it? Would it be seamless? Would it be easy if you made some modifications? I think in this regard, the annotation, all knowledge, levels of interoperability is really critical. I believe the lowest level of interoperability is like they actively interfere with each other or something like that. The response may be from another team that go ahead and do that, but we can't work with that. Or it could be, this is seamless. This is great. Let's just do this. We'll keep that in mind. It's not a binary kind of view of interoperability. It's much more of this tiered or nuanced way, if you want to think about it. I couldn't do this without actually showing you some of the materials themselves because they're just amazing. What you're seeing here is one of the early printed books in something we call the Archaeology of Reading. This is by somebody named Livy. You can see that on this screen you can already see some of the types of annotations and notes that are in the documents themselves physically, how we've started to transcribe them and built a schema around them so that you can display them, you can search them, and so on. That in itself was helpful and useful to the scholars that do this particular kind of work, but even they have said, well, we know there are relevant references and resources in the Perseus Digital Library. We would like to be able to get to those in some seamless and comprehensive way. This is a document about that particular Livy book that I just showed you in our Archaeology of Reading project, but this is coming out of the Perseus Digital Library. Even beyond just the reference to knowing that the same kind of document exists, we want to be able to take people to the exact paragraphs, the particular locations of those documents and align them, and we want to be able to do that even if you can't read Latin. I mean, I can't read Latin and I don't think I will anytime soon. Many of the students, of course, that are even in these classes, even if you're a classicist in your early stages of your career, you may not have these language skills to do that on your own. A text alignment or a language translation is a different kind of annotation from what we're looking at it. To be able to do that in a more granular way, so that's a particular aspect or a particular portion of the section that you're looking at. And also even things like tree banks. So there's additional scaffolding that could be useful for people to understand what they're looking at. So in addition to seeing this, the translation, the alignment, you might see how the grammar is sort of laid out, how it's broken out. How do you work with those as annotations is one of the things we'll be looking at. We're looking at doing this kind of work through October of this year, because in November, one of the protocols that Mark will talk about is the triple IF for the international image interoperability framework. I think that's right. They have their working group meeting in November, so we're hoping to be able to show them some of the work we've done at that meeting. But then in the second year, the grant, which will basically run from November through next September, we'd like to start looking at these other kinds of annotations, for example, geospatial content. So there are a lot of references to places, names of cities or historical monuments, things of that nature. And Perseus has very nice tools. And actually, I think Palagios is in the room, maybe. Oh, there we go. So Palagios is one of the groups who we're hoping to work with as well to make this kind of thing happen. And as you might imagine, again, this is basically providing people much more scaffolding and understanding about what you're looking at in addition to just reading it on the screen. I will turn it over to Mark who'll give you a little bit over here with the technology and the approach that we're taking. Hi. I'm going to talk briefly about the design approach we're taking. This is still sort of tentative. We're just getting started with the design. So bear that in mind. As Said mentioned, we have two main partners on the project, JHU and Perseus. JHU has a lot of familiarity with sort of image-based objects, manuscripts, printed books online, folios presented together with metadata about them, etc. Perseus really has this text focus, as I just mentioned, with these very sophisticated reading environments. So one of the things we're trying to get out of this project is trying to bring them together and see what we can do in terms of annotation. So these two partners have their own infrastructure, they have their own collections, they have their own different focuses, but fundamentally they're trying to do the same thing. They have a reading environment and a user interacts with a collection with it. And then you can view the data in these collections as annotations. As Said mentioned, all the different types. You can look at a folio, you have a transcription of the folio, the transcription's an annotation, the translation of the transcription's an annotation, the part of speech tagging's an annotation, the alignment to some textual variance can be an annotation, etc. etc. So annotation is an interesting lens to think about bringing everything together and unifying it. So how can we think about doing that? Well, we want to leverage the W3C's web annotation data model. This is a nice fit because these resources we're talking about are already on the web. They have URIs. We can use the web annotation vocabulary and we can make new statements about them in a new context. It's extensible if we find some things don't quite meet our needs. For example, with text alignment we can extend it and add more information as needed. We're also probably going to look at the web annotation protocol, which is based on the linked data platform, which JHU at least is very familiar with. It may or may not be useful. We'll have to see. The one sort of big missing piece that we haven't really, we've just started talking about is thinking about discovery, is in this reading environment, how do you discover these annotations that we've put together in this shared infrastructure? How do we know what's relevant? And that's a crucial piece we still have to really think about and people have ideas we'd like to talk to you. So I think I will spare you a detailed description of all these different acronyms on the page, but I think we are happy to talk about them and do any demonstrations if you come and talk to us later. So just to end with some acknowledgments, obviously the Foundation's generous funding is really important. The Christine de Pizan Digital Scriptorium actually has given some of the content that we put into what we now call the Digital Library of Medieval Manuscripts and there's a link there. And then the projects that came to the workshop and we're very grateful for their time. And also Rob Sanderson, who wasn't representing a project per se, but he's just a really smart and nice person. So thanks to Rob. And if we have time we'd be happy to take questions. Thank you. No questions? Wait, you want to step up to the mic? Thanks for the presentation, guys. So obviously an annotation encountering that changes the reading experience with a piece of text. One of the first things in main utilities that come to mind is adding meaning or context to facilitate understanding. So one thing that I think about is when you're reading a Shakespeare book in middle school or high school and those footnotes can be really, really important to clarify words or phrases that we wouldn't understand. So with the impact it has on the reading experience, how would you consider or distinguish between annotations that provide additional context and understanding versus annotations that provide analysis and go deeper into something beyond just maybe what the author themselves intended and what is an appropriate place or time to expose that annotation to a reader, if that makes sense? Yeah, no, I think that makes a lot of sense. Those are good questions and we don't have really well-formed answers to be honest. I think, again, that has implications on discovery too, right? Are you discovering something that's contextual or something that's a comment particularly about the author? From a technical perspective they may not be terribly different but I think from a user experience they're very different. We also I think this is fair to say Hopkins does not have a lot of tremendous UI expertise so if anyone would love to work with us on that we'd be very thrilled because I think there's lots of user interface implications for this too, right? How those get presented or how they're labeled or how they get exposed those are really important questions I think. Yeah, you know that's the sort of thing that we need. We have to talk to scholars and we just have to get detailed use cases about and because we don't understand all of that we need to work with someone who can give us the background, tell us what the infrastructure needs to accommodate. My Elton Barker Palagios, in fact you probably just answered my question which was you've talked about users. I was wondering which users are you talking about and what role are they playing within this project because I understand you're kind of a planning stage and you're thinking about the design things. I was just wondering what kind of user communities will you be drawing upon and to what extent will they be helping you in this design process? Yeah and I think one thing we should mention is with previous projects like this where we've in fact gotten funding for Foundation, we've taken a similar approach where hopefully if we get the design good enough and the prototype good enough that it's not rigid and it's fixed we can make changes you know once we get feedback. Mark is a computer scientist and I'm an engineer so we know nothing about this content which is kind of exhilarating in some way but also very nerve racking in another. We're very fortunate at Hopkins that the people who've been driving these projects are scholars so we have a medievalist who led the Romantal Rose. We have actually not at Hopkins, Tony Graf to Princeton is the lead for our casual reading so we were very much working with a scholarship day one. Another thing I will say is that there's an educational component to this grant so there's someone at Furman named you may know her Chiara Paladino who is teaching courses already using some of these kinds of capability tools and plagiosis tools so we'll want that feedback throughout the course of the project but because it is a demonstration we assume at the end of this people are going to pop calls on it and and that's great we actually want that. Additional user communities would be would be very good for helping us get more requirements on the data models and the technologies. Hey a round of applause.