Linked Data and Libraries Caitlin Tillman, Juliya Borie and Elizabeth Ridolfo Town Hall Nov. 26, 2015 http://www.duskyswondersite.com/tag/millau-bridge-in-france/

The Canadian Linked Data Initiative (a working title) Goal: Leverage the existing collaboration between the Technical Services departments of Canada’s top 5 research libraries and work with LAC and BNaQ to form a Canadian Linked Data Initiative.  Working cooperatively, this group, along other units in each library, will articulate, plan, and find funding for a suite of projects that will coordinate the metadata production activities of our library's technical services units, special collections, and digital libraries in both English and French.  Working together and with partners in Canada, US, and Europe our goal is to provide a path to linked data readiness for our institutions and leadership for the adoption of linked data by libraries across Canada. 

for library users What is Linked Data Why Linked Data What are the benefits Linked Data initiatives and how you can learn more for library users

ENG353Y1Y Canadian Fiction Scenario: a student taking 3rd year English course on Canadian fiction for which he/she need to read Margaret Atwood’s novel “Surfacing”

(our collections) on the web? Where are libraries (our collections) on the web? A search for Margaret Atwood “Surfacing” in Google. Where are libraries (our collections) on the web? Google Knowledge Graph brings together information from various sources and shows how different entities are connected – Linked data in action!

There is a way to find libraries from the first page of search results There is a way to find libraries from the first page of search results. If I click on a link for Goodreads, which is a popular website that provides reviews and recommendations for books, there is a link that says Libraries. It takes me to Worldcat, and I can then go into our catalogue to find a copy of the novel.

If users are not aware of this way of findings library resources and are coming through Google, they are not aware of libraries’ holdings and their access to libraries is quite limited.

Traditional Web – Web of Documents resource resource link link resource resource link link link resource link link resource link link link resource resource resource link link link link Difference between traditional web (web of documents) and semantic web (web of data) resource resource resource resource resource

Semantic Web – Web of Data link data data data link data data link data data data link link link link data data link link link data e data data link link link link data data data data data resource data data

Why Linked Data for Libraries? Opportunity to make library resources more findable on the Web. Only library community uses MARC. Information recorded in records, hiding potential context within and across collections. http://www.slideshare.net/corylampert/exposing-hidden-relationships-practical-work-in-linked-data-using-digital-collections

Inventor of the Internet …of Linked Data A set of best practices for publishing and connecting structured data on the web Builds upon standard Web technologies such as HTTP, RDF and URIs It is about making data on the web: machine-readable linked to and from other data sets https://it.wikipedia.org/wiki/Tim_Berners-Lee#/media/File:Sir_Tim_Berners-Lee.jpg Tim Berners-Lee Inventor of the Internet

Use of Linked Data D.J. Wild, et al., Systems chemical biology and the Semantic Web: what they mean for the future of drug discovery research, Drug Discov Today (2012), doi:10.1016/j.drudis.2011.12.019 https://commons.wikimedia.org/wiki/DNA#/media/File:Ssvsds.jpg Use of semantic technologies in in different domains: medical research, historical data, use of linked data in media and journalism.

Resource Description Framework RDF Resource Description Framework dc: creator Surfacing Atwood, Margaret Subject Predicate Object Each triple is a statement about a resource

Resource Description Framework RDF Resource Description Framework http://purl.org/dc/elements/1.1/creator http://www.worldcat.org/title/surfacing/oclc/697192 http://viaf.org/viaf/109322990 Subject Predicate Object By using URIs (Uniform Resource Identifier) as names for things, this makes the statement “machine actionable”.

Linked Data city_in capital_of lives_in author_of born_in located_in born_on founder_of educated_by founder_of author_of likes likes held_at Multiple statements can be linked together in a graph Local copy

Creating a Web of Data By publishing data as Linked data, we can tie information into a global network of linked data resources If we look at the linked data cloud, you can see the large number of connections coming in and out of various datasets and the library-related datasets are in the right corner. Linking Open Data cloud diagram, by Richard Cyganiak and Anja Jentzsch. http://lod-cloud.net/

MARC as “Things”, not “Strings” Link data allows to transform resource description from records-based approach by focusing on things that represent authors, topics and events in our collections.

Authority Control finally gets awesome http://lccn.loc.gov/no2015062659 http://geonames.org... http://id.loc.gov... In a report by W3C Library Linked data incubator group, the authors have identified the first step on our path to linked data is to take our human-understandable, controlled strings of text in our authority files for authors, organizations, etc. and link them to unique, persistent identifiers.(Baker et al., 2011). This is something we are currently starting to work on.

Potential Use Cases for Library Data Enhanced authorities will not have an impact on the public display at the moment, but they can facilitate new use cases. This data could potentially facilitate Knowledge cards with biographical information for authors similar to Google knowledge cards. Enhanced subject access can provide better context for our users. By describing people, places, organizations, events, and associated concepts and relationships with linked data, we would be able to tell better stories about our collections.

Does it work? Assessment studies, for example one from Bibliotheque nationale de France, indicate that people can now find library resources on the web and majority of users (80%) are coming from search engines.

Serendipitous discovery Serendipity As Juliya mentioned, it’s not completely clear exactly what linked data in libraries might look like, but there are several examples out there of institutions and groups exploring the possibilities with their own data. Some of them are interested in the possibility of more serendipitous discovery based on the multiple relationships that can be established between items. An example of a library catalogue using Linked Data is the National Library of Spain, whose goal in the project was to create a new way to explore their bibliographic data, enriching that data with external information, and bring together the library's various resources. The BNE transformed and linked around 7 million authority and bibliographic records originally in MARC into RDF. The connections between works, people and themes allow for a different type of searching and discovery, and the catalogue is enriched with pictures, biographical information and other data elements which come from external sources. As we see on their home screen, you are able to browse by person, work and theme. A search for Margaret Atwood gives us suggestions of related authors and important works, bio with photograph, works by and about and in which she participated links out to external sources of information. The records also link to the individual works and their locations and languages National Library of Spain

Sharing records rather than copying them Evolution from cataloguing to “catalinking” Catalinking:  The BNES accomplished this change to their catalogue by transforming their existing records. Outside of transforming existing records to RDF, what might cataloguing of *new* items or *new* instances look like? Instead of creating a record with data embedded in it, unconnected to anything outside of itself, this would mean pulling data from an outside authoritative source, and then creating links between items such as works, persons and instances, and identities. This change was described by Eric Miller of Zepheira as one from cataloguing to catalinking. It would be creating web-friendly descriptions of resources which would include metadata that is machine readable and discoverable outside the library catalogue, and that has a context because it is linked to other authoritative information resources such as the VIAF. Creating metadata with links that link out as opposed to creating marc records means that data can be released/reused in different ways. Whereas right now, we focus on manifestations, in the future we will probably focus more on works, finding and linking records instead of creating them from scratch or duplicating by copying into a local catalogue that is siloed.

BIBFRAME How  might changes to how metadata works and acts take place?  In 2011 as part of a library of Congress initiative, the Bibframe metadata framework was created as a replacement for MARC standards, to use linked data principles to make bibliographic data more useful inside and outside of the library community. To serve the very specific needs of library staff and patrons while engaging and integrating more in the wider information community. Instead of bundling information neatly as a “record” and potentially duplicating information BIBFRAME relies on mapping the relationships between resources (Work-to-Work, Work-to-Instance and Work-to-Authority relationships), which it manages this by using the controlled identifiers for things that Juliya mentioned earlier. This model encourages the creation of clearly identified entities and machine-friendly identifiers which lend themselves to machine interpretation of those entities. So, information will not be embedded in the record, but link out to an authoritative source for the information, and be easily interpretable by machines, both allowing for greater connection and reuse. How might this framework be used by libraries and institutions for outreach and to add value to their collections?

Building virtual collections … Linked Jazz https://linkedjazz.org/ Out of the Trenches http://www.canadiana.ca/en/pcdhn-lod …Together Here are two examples of how content might be displayed using linked data. These collections are are experiments with using Linked Data triples to dynamically generate relationships between assets, sometimes held in different locations or institutions. For example in Linked Jazz, oral history transcripts from different institutions were analyzed, names mapped and relationships determined. Choosing a person expands their network, showing their connections, the level of relationship, and linking to the assets which describe this relationship. It also incorporates images, videos, and short biographies of the jazz musicians within the network. Canadiana.org created Out of the Trenches to try to make use of existing metadata about  resources (in this case assets about the first world war) and repurpose it without loss of context and meaning. ather than reduce the metadata to a common subset, the approach was to maximize its use by expressing the contributed metadata as Resource Discovery Framework (RDF) “triples” Different types of material from different domains and institutions were integrated in a site in order to tell the story of people and places of WWI.  Maps, correspondence and photographs were all tied together to form snapshots of that time. Both of these projects combined and transformed resources from numerous different institutions in order to turn them into collections which form dynamic connections between pieces of data. They encouraged group collaboration and the sharing of resources, and were projects that encouraged all involved to work on linked data possibilities together.

Greater interoperability RDM Sirsi T-Space Linked Data Discovery Layer Collections UofT ERM(s) Summon MyMedia If information about assets from different sources across campus incorporated linked data, a linked data discovery layer could be used to pull together UofT material that would normally be only discoverable through different searches of separate containers. It could be pulled together via searches of elements of their datasets, or connected into meaningful sets using information in their triples instead of having these items exist in relative isolation, most easily accessible through searching their individual sites. This could apply not only within UofT but with the possibility of sharing datasets outside of UofT or with partners. This gives the potential to reach out to other libraries inside and outside the system.

Re-Thinking What We Do …And How We Do It Convergence of Linked Data and traditional library metadata Constructing vocabularies Describing properties of resources Identifying and linking resources Exchanging and aggregating metadata We need to change the way we do things to respond to the needs and expectations of our users Creative. Innovative. Inspired. Libraries are experiencing a time of change in ILS platforms, cataloguing standards and user discovery avenues. Linked data proposes the possibility of getting more value out of what libraries already do and also responding to some of these changes. Linked data work would build on what libraries have been doing well for a long time,  including description of resources, linking resources and using controlled vocabularies. The tools, names and types of links may change in order to make the very valuable information we already provide to the University community more widely accessible and shareable.

Stepping stones/How to get there Collaboration as part of the Canadian Linked Data Initiative Training and education Enhancing current and planning future digital collections with Linked Data in mind Leveraging /transforming the data in our existing records Moving to a linked data module does not necessarily involve dramatically changing or overhauling existing processes immediately. But what does it involve practically? As Caitlin stated at the beginning of this presentation, the Canadian Linked Data Initiative was recently formed to investigate resources, training and projects that will prepare us for these changes. Smaller steps using existing data such as enhancing current collections and planning future digital collections with LD in mind are things that can be done as a starting point. We can also leverage by transforming the vast amount of data in our existing Authority and bibliographic records into Linked Data. All of these will be steps towards improvement for our users as we head towards improved access to our resources. If you’d like to more information about this group and about Linked Data initiatives at the University of Toronto please feel free to contact either of us.

Thank you! *Special thanks to Marlene van Ballegooie for letting us re-use her slides Thank you Marlene for the use of your slides