Linked Open Data in Practice: Emblematica Online
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Title of Series | ||
| Number of Parts | 16 | |
| Author | ||
| License | CC Attribution - ShareAlike 3.0 Unported: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal and non-commercial purpose as long as the work is attributed to the author in the manner specified by the author or licensor and the work or content is shared also in adapted form only under the conditions of this | |
| Identifiers | 10.5446/47574 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
|
6
11
12
00:00
Lecture/Conference
03:17
Computer animation
10:24
Computer animation
13:35
Program flowchart
21:10
Lecture/Conference
Transcript: English(auto-generated)
00:12
Our next speaker is Myoung-Ya Han from the University of Illinois at Urbana-Champaign.
00:25
And her presentation is about linked open data in practice, so Emblematica Online. Alright. Thank you so much for the program and committee for having me here today.
00:42
Emblematica Online has been supporting humanities scholars for their teaching and research for the last 25 years. And most of the services that are available in the Emblematica Online is possible because of the metadata that we are using for describing emblem resources.
01:02
So my colleagues in Germany and in the States are very excited to share our project and our experimentation on linked data today with you. So, my talk today is in four parts. First, introducing background, what are emblems and Emblematica Online.
01:21
And descriptive metadata that is used in the Emblematica Online called Spine. So what is Spine and how do we use it? And the linked data in Emblematica Online. How we publish emblem data as HTML and IDFA and how we use linked data sources to enhance user experience.
01:41
And lastly, lessons learned from our experimentation. So first, what is an emblem? Do you know what emblem is about? The early modern emblem is flourished in Europe as a popular literary genre from 1531 until about 1750.
02:02
And its form is compound, a combination of text and images. So sometimes one emblem is just one page long. But sometimes an emblem can be spined into 20 to 50 pages because of the texture content is that long associated with the images.
02:21
And the emblem book is called emblem book because it is a collection of emblems. And it is a very highly- emblem is very highly contextual influenced by contemporaneous events such as reformation and the Thirty Years War. And emblem creators drew their inspirations from such diverse sources.
02:43
This is not me. Example, the bible, fables, mythology, science, and medicine. And as you can see from these images, all those images come from so many different sources. And as object of research, emblem straddle multiple scholarly domains such as art and cultural history,
03:05
literature, semiotics, religious studies, and political science. We can only guess what kind of emblems could be created after that election. But that is everyone's guess. So what is Emblematica Online?
03:20
Emblematica Online is a portal for a key genre of Renaissance text and images, which is emblems. And this is actually studied in 1990s in University of Illinois at Urbana-Champaign. We studied digitizing emblem books in our collection. Started with four books, then 10 books, and then 19 books.
03:42
And then in 2002, we joined forces with the Herzog August Bibliothek in Wolfenbüttel, Germany. So we tried to find ways to digitize each library's emblem collections and to share those collections together. In 2000, we got the grant from JFK and NEH to digitize in massive manner.
04:08
And then in 2013, University of Illinois got another grant from NEH and shaped our emblematica online in current form. And now it provides access to digitized emblem resources from six different institutions around the world,
04:25
including Getty Museum and Glasgow University, Utrecht University, and so on. So now we have more than 1400 digitized emblem books and 28,000 emblems included in a portion of those books.
04:43
And we provide granular levels of access to digitized emblem resources by employing the metadata standard called SPINE. And what is the SPINE then? From early on, the emblematic community knew that we need a new metadata standard to describe emblem books and emblem resources together.
05:05
So in 2004, Stephen Rose published a spine of information headings for emblem-related electronic resources that provides descriptive metadata structure that allows describing emblem books and emblems included in the book together in one record.
05:25
And in 2007, as library metadata moved toward XML environment, Thomas Stecher developed a SPINE XML schema. And that became a metadata standard for the emblem community that is used for description and dissemination
05:44
of digitized emblem resources at a granular level. This is the reason why we can get the emblem books and emblem metadata from six different institutions. So then what kind of that granular level entails? What does that mean?
06:00
So SPINE works as a container of different levels of descriptions. For example, we can add a book level description, which is traditionally added in marker format. We can add those book-level bibliographic metadata in SPINE in either MARS or TI header. And also we can add copy-specific description about the digitized book information in SPINE as well.
06:26
And then lastly, we can add emblem-specific descriptions in SPINE. And the emblem-level description include the title of the emblem, called MATU, and the URL of the emblem spread 50 pages long or whatever.
06:40
And so on. And all those transcriptions and subscriptions for the textured materials information. And finally, the image information and the icon class headings that describes emblems and images included in the emblem.
07:00
And because we have these granular levels of metadata, we can provide multiple granularities for discovery, access, and citation services in emblematic online to meet scholarly needs and expectations. By using bibliographic metadata, we can provide book-level discovery and access service.
07:23
And by using transcriptions and MATU, we can provide emblem-level discovery and access service. And then, by using imaginary topoid themes and motifs, we provide a pictural-level display and access service. And in addition to do that, we allow scholars to choose arbitrary segments of the images.
07:46
And we provide URLs for that selection of the image, so scholars can use those URLs for their publication. And now, we all want to work with some kind of linked data things. And emblematic is not different.
08:03
So, we would like to approach linked data in two different aspects. First, as a producer, we'd like to make digitized emblems resources discoverable on the web. They don't need to come to emblematic online. Emblem resources should be discoverable on the web.
08:21
And as a producer, we would like to provide related web resources that are out there somewhere to bring into emblematic online portal, so users can see other contextual information from our site instead of going out and search by themselves again.
08:40
And we have a little bit different approaches for publishing emblem metadata, emblem data to linked data. But first, we have two considerations, the identifier and vocabulary. We want something that provide persistent, measurable URIs at each level of
09:00
granularity and the different entities, for example, name and subject and etc. And we want to have something that have multilingual vocabulary for describing themes and motifs of the emblem images and the text. And then we have these two things in mind. We thought we decided to use VIF and icon class as an object of our emblem data.
09:29
And when publishing spine metadata to linked data, we have a very unique approach, which is HTML and IDFA by using schema.org semantics.
09:43
So why schema.org semantics? I think everyone already know because it is used as an encoding standard by web search engines, Google, Yahoo, Bing all use schema.org semantics for their resources. And it is really easy for us to represent emblem resources in the web using IDFA.
10:04
So for example, the pictura level metadata looks like this in spine metadata. And then we transform this into HTML and IDFA page. So pictura is a type of emblem and emblem is a new schema.org semantics that we are going to propose.
10:24
And then icon class is represented as a schema.org concept code. And it has several properties such as coding system, name, same as, and coding values. As you can see, not every spine metadata element has matching schema.org semantics.
10:46
So because emblem is such a unique resources. But instead of creating the new ontology that only works for emblems, we try to work away as schema.org community recommends. We try, we create an emblem extensions for emblem specific elements.
11:06
So this is what we have been working on it. We almost, we created this one for two years ago, but has been delaying the publishing this emblem extension. So we hope that after we publish this, the emblem extension for schema
11:22
.org, then other emblem users can publish their data by using schema.org extension. So how, then how Google understand our HTML page with IDFA embedded in. For if you look at the name, name has a type, a role and type and ID included in our HTML page.
11:44
And then Google grabs all those information as we hoped. So they grab the name with roles and type that is a person. And if there is a VIF link, then they grab those IDs. And we really hope that this will allow emblem medical resources discoverable and usable with more people in the web.
12:10
And as a consumer of the metadata, the linked open data, we try to use VIF web services because it provide additional information about the name, including nationality, gender and others.
12:24
And it connected to other resources, including WorldCat identities and Wikipedia, excuse me, that provide a lot of contextual information as first presenter talk about. So in order to do this kind of, in order to exploit VIF web services,
12:44
we did the metadata reconciliation work for all the names appeared in the spine metadata first. So if the name has the matching VIF link, we store those link on our local server, not the string values, those personal names.
13:01
And all additional services are dynamically performed just to go out to VIF and so on. So this is a book level page, book level display page in emblem medical online. So if the name appeared in the spine metadata, usually a creator or a printer or somebody who create those emblems.
13:25
If the name has the matching VIF link, we added those little box with more info. And if users wants to find out more info and clicks that button, then our system script goes out to VIF and then grab this information and display to users.
13:44
And you can see that there are two National Libraries authority data link as well as VIF and WorldCat and no Wikipedia. Sometimes if it's a Wikipedia entry, you can have a Wikipedia entry in here as well.
14:01
So users doesn't need to go out to the web and search their own for specific National Libraries authority data or Wikipedia and so on. And for the icon class headings, it is the controlled vocabularies for describing iconographic images and used for emblem community widely.
14:21
And it is developed and maintained by the company called archives in Netherlands. And this is the same. We only keeps the icon class annotation in our own server and provide services. And icon class web service provide multilingual service and it supports hierarchical browsing service as well.
14:42
So this is the example of their web service for one of the icon class headings available in their server. You can see that icon class is available in two different forms. One is called notation in numeric form. The other one is called preferred label in string format in four different languages.
15:02
So we can actually use this for the multilingual services for our users. It also have narrower and broader terms available in their web service so we can use this for icon class browsing service. This is an example of emblem label display page in emblematic online.
15:24
Every emblem page displays the title of the emblem on top and then emblem images. And icon class that describes that emblems and text. And those are actually assigned by special scholars who really understand the emblems and the history of emblems and so on.
15:47
And from emblem label display page, users can select their browsing option languages from here. Those are four languages available in icon class heading web service. And then when you choose those languages, all the display of icon class headings automatically change into the language of user's choice.
16:10
And for each icon class heading, if you click it, then four options appears for additional service. So users can find more emblems with the same icon class headings in the emblematic online
16:22
or find images in other sites such as festival culture and virtual print room or browse icon class headings in different narrow or broader terms. So if user choose browsing icon class headings, then the new window pops up and shows all related terms of that specific icon class headings.
16:48
And from here, users can also again change languages and have four other services. So this is an emblem site, but you can also find images from available in other web pages such as virtual print room and festival culture.
17:06
So if user can click one of those two sites, then they can find images from other available web collections right away. Because we know how the other web collections work with their URLs for search algorithms.
17:26
And this is not exactly linked data, but it shows the power of controlled vocabularies. And I believe that controlled vocabulary is one of the important foundations of linked data. So this is possible because virtual print room and festival culture
17:42
collections are also using icon class as their descriptive controlled vocabularies. So what kind of lessons learned from our experimentation? I think it is almost the same as the previous presenter and the last year's Europeana's presentation about reclamation work.
18:03
Metadata quality really matters. If your metadata is using controlled vocabularies and created consistent manner, then publishing your metadata to linked data or design new services by exploiting linked data services can be easier if than not.
18:27
And automatic metadata re-consolation work requires human intervention. And we have to really improve those matching algorithms regularly. And you have to run those re-consolation programs regularly.
18:42
And you have to also check manually, the spot check, whether the matching is really right or not. And then not all linked data services, especially authority data, are not the same. Some authority data has really rich contextual information. But some authority data has just the names or place names, sometimes with the wrong spelling.
19:05
And not all linked data services are not open. Some are closed, some are partially open, and they have such a limited service as well. So you really need to know what kind of linked data you are going to work with to provide additional services to your users.
19:25
And I have to say that Emblematica Online is a really unique case study, a case for how one specific metadata standard is developed, used, and supported in a community, the Emblem community. And I think the linked data and the semantic web is exactly the same.
19:43
It really requires community effort that we have to, together, we have to identify needs and areas of work, and try to develop the new workflows together. Thank you so much, and please visit us at emblematica online and let us know what kind of things we can do better and more. Thank you.
20:30
First of all, thank you. I'm just curious because this is always a question of disputes in the community, so your experience.
20:44
The same schema.org data can be expressed three different ways. RDFA, JSON, LD, microdata. If you like, from a semantic web point of view, they are identical. Have you experimented with the others, and have you chosen RDFA for very specific reasons, or you just took whatever worked?
21:10
Well, we chose RDFA because it is quick and easy, I have to say. Our web page has two different approaches, so those RDFA embedded attributes can be represented as JSON as well, so it is not about XML.
21:30
The main reason that we are trying to read RDFA is that it is quick and easy to make our metadata available on the web.
21:41
Okay, next. Hello, my name is Daniel Santi from the Library of the Representative of Indonesia. I want to know how to identify between one emblem to the other emblem, and then why do you choose the period for the emblem test.
22:14
Thank you. Do you... Okay, how to identify authority data to make sure that there is link data between one emblem and the other emblem.
22:43
Yeah, to get that, yeah. So we use icon class headings as a control vocabulary that describes emblems, and usually one emblem has more than... sometimes it has one icon class headings, but it has emblems and it has more than ten icon class headings.
23:01
And emblem cannot mean so many different things. So if we just try to match emblems with the same icon class headings, so they cannot be exactly the same emblems, but they have the same representation of images or motifs. So if you find the trees, then you have really a lot of emblems from our site, but that actually means very different things.
23:29
Each emblem means different things. The only connection is the tree is appeared in each emblem. Thank you.
23:41
Okay, and let me ask, now it's great that your data is available as link data. What's next? What uses of this data do you envisage? What applications do you want people to build? Well, we have been thinking about putting our data in RDF and trying to build RDF store since three years ago.
24:07
And we identify the right triple store and the way to transform all RDF data to RDF. But we haven't done that work yet.
24:20
So at this point, for now, we can just provide access and discovery. But we hope that sometime in the future, scholars can come to our site and build their own collections with the same elements of their emblems, with their choice, so they can publish those collections for their own purpose.
24:46
But that is still in the future in plan. And you can't know all the uses, how people will use your data. Mobile applications as well. I hope so. Alright, thanks a lot. Thank you.
Recommendations
Series of 36 media