Uploaded byrobin fay
Linked data presentation for libraries (COMO)
The document provides an overview of linked data and libraries. It discusses basic principles of linked data such as reusing and linking data to make it reusable, easy to correct, and potentially useful to others. The document also discusses how linked data fits into the semantic web vision by allowing machines to better understand and utilize data. Finally, it discusses getting started with linked data through terminology, advantages, and modeling library data in linked data formats like RDF.
More Related Content
Similar to Linked data presentation for libraries (COMO)
Linked data presentation for libraries (COMO)
- 1. Making the DigitalConnection: Linked Data and Libraries Robin Fay, UGA Laura Akerman, Emory Doug Goans, Georgia Tech Georgia COMO 2012 October 4, 2012 Macon, GA
- 2. Head/Database Maintenance University of Georgia Libraries RobinFay @georgiawebgurl libraries social media
- 3. Making the DigitalConnection: Linked Data and Libraries Agenda: ● Gain a basic understanding of linked data ● Discover new library projects ● Gain a better understanding of how linked data will drive our future systems and how it fits into the future of the web -- the semantic web What we will talk about: ● Quick overview of semantic web and linked data (more terminology in the glossary) ● Getting started with Linked data and the Semantic Web ● Making some connections with bibliographic data - a case study
- 4. ● What islinked data and open data o Linked data is about reusing data o We already do some linked data in our library catalogs and even in our daily lives o The link in a bibliographic record (like an authority record link) is linking data o A link that we share to our friends on facebook is linked data (of sorts) ● Linked data is a link to a record/data/content that can then be utilized in some way ● Open data is data that available to be used in some way with no barriers to access (licensing, etc.) Making the Digital Connection: Linked Data and Libraries
- 5. Basic principles oflinked data ● It keeps us from having to re-enter or copy information o Making our data: reusable easy to correct (correct one record instead of multiples) efficient and potentially useful to others ● It can build relationships in different ways - allowing us to create temporary collections (a user could organize their search results in a way that makes sense to them) or more permanent (collocating ALL works by a particular author more easily; pulling together photographs more easily) Making the Digital Connection: Linked Data and Libraries
- 6. How it fitsin with the semantic web The Semantic Web is based upon more precise utilization of data and is heavily dependent upon The code The metadata and its metadata schemas (rules) The ability for machines (including devices and home appliances) to talk to each other and make sense of that communication Linking data makes this process easier since we do not have to re-enter data, we can just link to it. Linking data will make the Semantic Web happen. Making the Digital Connection: Linked Data and Libraries
- 7. ● Terminology o RDF/XML(language - post MARC world) o URIs (Uniform Resource Identifer) linking point to our data) o Serialization (store data; re-use later) o Triples (Simple semantic structure; Subject- Predicate-Object) o SPARQL (QL = Query Language) Making the Digital Connection: Linked Data and Libraries
- 8. ● Advantages (reusabledata, potential to provide and built relationships, discoverability) ● How library data fits into linked data o FRBR ( a bibliographic FRAMEWORK which is more semantic by nature) RDA ( metadata rules which are not tied to a programming language such as MARC but can work with semantic web standards like XML); IRs, and CMS like Drupal which have semantic web capabilities ● RDA expressed as RDFa Making the Digital Connection: Linked Data and Libraries
- 9. Getting Ready forLOD Doug Goans Head of Library IT and Development Georgia Institute of Technology
- 10. Readiness: What, Whyand How Linked is an approach to data. Open is a policy. Data is a technology and a set of standards. Source: http://www.scholarslab.org/digital-libraries/introduction-to- linked-open-data-at-rare-books-school/ Making the Digital Connection: Linked Data and Libraries
- 11. Readiness: What Discussions ofdata and services Catalog, archives, repositories, library use data LOD emerging services/data in general ● WorldCat ● DBpedia ● data.gov Making the Digital Connection: Linked Data and Libraries
- 12. Readiness: Data Records(MARC, etc.) SOURCE: Getting triples from records: the role of ISBD http://www.slideshare.net/scottishlibraries/isbd-record2triples Making the Digital Connection: Linked Data and Libraries
- 13. Readiness: Data Records(MARC, etc.) SOURCE: Getting triples from records: the role of ISBD http://www.slideshare.net/scottishlibraries/isbd-record2triples Making the Digital Connection: Linked Data and Libraries
- 14. Readiness: Why Discovery (GoogleSemantic Search, Facebook Open Graph) Empowerment (library development, local and global) Peripheral issues (important) ● Data Use Policies / Licensing your data ● System Readiness ● Teaching, Learning, Research (use cases and service development) Making the Digital Connection: Linked Data and Libraries
- 15. Readiness: How ● UseCases - http://obd.jisc.ac.uk/ ● Integrate LOD into projects (IR, Archives, Web Presence) ● Explore RDF, RDFa, SPARQL end-points ● Framework approach (empower users and developers first) Making the Digital Connection: Linked Data and Libraries
- 16. Readiness: How Developing UseCases http://obd.jisc.ac.uk/ Campus Competition System Readiness Making the Digital Connection: Linked Data and Libraries
- 17. Readiness: How Current Services IntegrateLOD into projects Framework and Services Approach Making the Digital Connection: Linked Data and Libraries
- 18. Making some connections Linked dataat Emory Libraries Laura Akerman
- 19. First we studied Connectionsgroup formed Nov. 2011 Core group of 8, open to anyone Sponsors: Lars Meyer, John Ellinger Learning and planning for linked data important Classes taught every other week Based on Linked Data: Evolving the Web into a Global Data Space (Heath and Bizer) High level overview Programs Jon Voss, "LOD-LAM Emory", webinars Making the Digital Connection: Linked Data and Libraries
- 20. Pilot planning ● Wewere asked to develop a 3-month pilot project. ● Goal: demonstrate value of linked data for users and the library ● Goal: get feet wet with "triples" and technology ● With: o 1 person, 20+hrs/week; 6 others, 1-3 hrs/week including a couple of "techies" o Sandbox (repurposed pc/server running Linux) Making the Digital Connection: Linked Data and Libraries
- 21. Pilot project -approved ●After many ideas, choices - ofocus on topic (Civil War) oConvert sample "silo" metadata to linked data oAdd some "new" metadata as linked data oLink to some external data oSome kind of display ●PROJECT BLOG: https://scholarblogs.emory.edu/connections/ Making the Digital Connection: Linked Data and Libraries
- 22. Results: Making the DigitalConnection: Linked Data and Libraries
- 23. Things we learned,#1 Start small Making the Digital Connection: Linked Data and Libraries
- 24. #2 Working with freesoftware is time consuming! ● Even if you're a programmer... ● Don't expect it to work the first time! ● Patience, inner peace, get enough sleep! ● Don't let fear of displaying your ignorance keep you from learning! ● Windows people: learn Linux command line! ● When to "let it go", when to keep trying? ● Tool evaluation is an important step! ● Record what you learn and share! Making the Digital Connection: Linked Data and Libraries
- 25. Things we aretrying: Sesame - useful but web client limited (no batchload...) Callimachus - interesting but not ready to put time into it, and very beta LinkSailor - need to publish your data to use it Pubby, Djubby - publish your data ??? (not working yet) Our own scripts to query id.loc.gov - got the basics... ArchivesHub stylesheet - useful as a starting point, but too complicated? Simile MARC to RDF stylesheets - too different from ArchiveHub LC MARC to RDF stylesheet - Dublin Core too simple but a good starting point Simile Welkin - limited but at least it gave us some visualization! Making the Digital Connection: Linked Data and Libraries
- 26. #3 We need linkeddata to find connections! Making the Digital Connection: Linked Data and Libraries
- 27. Learn SPARQL #4 Making theDigital Connection: Linked Data and Libraries
- 28. Find two resources,name/creator in one is subject of another SELECT DISTINCT ?resource1 ?name ?resource2 WHERE{{ {?resource1 dc:creator ?name} UNION {?resource1 duch:origination ?person. ?person foaf:name ?name.} } { {?resource2 dc:subject ?name.} UNION {?resource2 duch:associatedWith ?concept. ?concept rdfs:label ?name.}} FILTER(?resource1 != ?resource2) } Making the Digital Connection: Linked Data and Libraries
- 29. Some SPARQL resources SPARQL.PRO http://en.sparql.pro/wiki/Main_Page listof SPARQL endpoints with sample queries XQuery/SPARQL Tutorial http://en.wikibooks.org/wiki/XQuery/SPARQL_T utoria W3C SPARQL Spec http://www.w3.org/TR/rdf-sparql-query/ Making the Digital Connection: Linked Data and Libraries
- 30. #5 There are manyways to model your RDF - all different! How do you decide?
- 31. ArchivesHub handles "subjects" <associatedWith><!--Aboutthe Concept (Person)--><skos:Concept xmlns:skos="http://www.w3.org/2004/02/skos/core#" rdf:about="http://duchamp.library.emory.edu/resource/id/concept/person/lcnaf/gearyjohnwhite1819-1873"> <rdfs:label xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xml:lang="en">Geary, John White, 1819-1873.</rdfs:label> <skos:inScheme> <skos:ConceptScheme rdf:about="http://duchamp.library.emory.edu/resource/id/conceptscheme/lcnaf"> <rdfs:label xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xml:lang="en">lcnaf</rdfs:label> </skos:ConceptScheme> </skos:inScheme> <foaf:focus xmlns:foaf="http://xmlns.com/foaf/0.1/"><!--About the Person--><foaf:Person rdf:about="http://duchamp.library.emory.edu/resource/id/person/lcnaf/gearyjohnwhite1819-1873"> <rdf:type rdf:resource="http://xmlns.com/foaf/0.1/Agent"/> <rdf:type rdf:resource="http://purl.org/dc/terms/Agent"/> <rdf:type rdf:resource="http://erlangen-crm.org/current/E21_Person"/> <rdfs:label xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xml:lang="en">Geary, John White, 1819- 1873.</rdfs:label> </foaf:Person> </foaf:focus> </skos:Concept> </associatedWith> Making the Digital Connection: Linked Data and Libraries
- 32. LC's MARCXML toRDF handles subjects: dc:subject "Geary, John White, 1819-1873." Making the Digital Connection: Linked Data and Libraries
- 33. Simile MARCXML toMODS to RDF: <modsrdf:subject rdf:resource= "http://simile.mit.edu/2006/01/Entity#Geary_John_White_1 8191873"/> <rdf:Description rdf:about= "http://simile.mit.edu/2006/01/Entity#Geary_John_White_1 8191873"> <rdf:type rdf:resource= "http://simile.mit.edu/2006/01/ontologies/mods3#Person"/> <modsrdf:fullName>Geary, John White</modsrdf:fullName> <modsrdf:dates>1819-1873</modsrdf:dates </rdf:Description> Making the Digital Connection: Linked Data and Libraries
- 34. Our learning No perfectmodels out there. A common model for all our sources (with maybe some special items depending on type of content) could make things much easier for us and users of our data! This is a major undertaking.. Making the Digital Connection: Linked Data and Libraries
- 35. #6 To FRBR,or not to FRBR? http://metadataregistry.org/schemaprop/list/page/1/schema_id/4.html Making the Digital Connection: Linked Data and Libraries
- 36. #7 Matching upwith external data is tricky! ● DBPedia: not easy to programmatically match up. We did some "by hand". oIssue: If only part of LC subject matches, should you link? ● LC Vocabularies - id.loc.gov - straightforward but have to skip many final subdivisions. ● VIAF: another interesting target, but not helpful enough for this pilot. Making the Digital Connection: Linked Data and Libraries
- 37. question: "Johnson's Island(Ohio)." Q: Does this mean the island itself or is it referring to the prison that was located on the island? o Found something for Johnson Island Civil War Prison and Fort Site A: I used a SPARQL query to find the collection linked to this heading. The Scope and Content Note says letters were written while someone was imprisoned on the island. http://dbpedia.org/resource/Johnson_Island_Civil_War_Prison_a nd_Fort_Site Should we use "owl:sameAs"? Making the Digital Connection: Linked Data and Libraries
- 38. #8 To havegood linked data, you have to have... data ____________|1861_________________ No dates, no timeline (scholars could've found it useful) Making the Digital Connection: Linked Data and Libraries
- 39. #9 MORE OPEN LINKED DATA! Onceyou get started, you don't want to stop! Making the Digital Connection: Linked Data and Libraries
- 40. Resources ● LODLAM: http://lodlam.net/ ●LODAM CHALLENGE: http://summit2013.lodlam.net/ ● LODLAM Zotero Group (Webliography of good stuff): https://www.zotero.org/groups/lod-lam ● GLAMLOD: https://groups.google.com/group/glamlod ● LC Bibliographic Framework Transition Initiative: http://www.loc.gov/marc/transition/ ● LITA - library linked data interest group: http://connect.ala.org/node/142470 ● Use Case Tool: http://obd.jisc.ac.uk/navigate ● Getting triples from records: the role of ISBD http://www.slideshare.net/scottishlibraries/isbd- record2triples ● Presentation materials at http://www.delicious.com/georgiawebgurl/metadata_presentation_como Making the Digital Connection: Linked Data and Libraries
- 41. Thank You -Q&A Robin Fay fay@uga.edu Laura Akerman liblna@emory.edu Doug Goans doug.goans@library.gatech.edu
Editor's Notes
- #11 Linked is an approach to data. You need to provide context for your data; you need to point to other’s data. Open is a policy. Your data is out there for others to look at and use; you explicitly give others this permission. Data is a technology and a set of standards. Your data is available using an RDF data model (usually) so computers can easily process it.
- #13 Look at your record (MARC or other item records in library applications, repositories, etc.) There may be triples already present!
- #14 First 2 columns are URIs the last column is a value. This is what you you publish out as triples.
- #15 LOD impact on existing systems (catalog, vendor, social media). Knowing how those systems will operate with linked data.
- #20 At Emory, the Heads of the Systems and Content Divisons and my Digital Assets Strategy team leader John Wang, saw the potential impact of linked data and recognized that the library needed to make some effort to educate staff. So I have been chairing a group that has been learning about it. It's an open group. In addition to people involved with metadata and digital services, we have a programmer with our ILS system, an archivist, and a subject speacialist, and people from across the library and beyond have come to the classes led by myself and a graduate fellow. I say led, because I was learning along with everyone else!
- #23 So, this is the web client that comes with the Sesame RDF framework. Not very sexy, but something to play with We've been working on a way to temporarily publish our URLS so we could actually use a linked data browser to navigate from our data to other data, like DBPedia, but haven't gotten there yet... so this is all you can see for now. Stay tuned!
- #24 Our pilot was really ambitious. We were thinking about linking many types of content and link to many external resources, using many tools... We found we really only had time to bite off a small piece of each of our goals.
- #25 How many of you that don't have systems jobs, have tried out open source software on your own? One of the things I wish we could was just spend 3 months doing evaluation of tools - triple store databases, web frameworks and "linked data enabled" tools like Drupal, visualization tools. But, we really needed this first pass to understand what we needed to look for and have something to compare. NOW maybe we're ready for real tool analysis...
- #26 We have tried out a number of tools; this isn't a recommendation, just what we tried first. Tool evaluation
- #27 Initially we were going to go for "just a handful of each kind of record" that had strong interconnections... but what we found was, that trying to use our existing tools (Primo, Aleph, and things like keyword searching a dump from our digitization workflow application) was WAY too time-consuming. We decided to grab larger samples and go on faith... We discovered that our ability to grab sub-sets of our data according to criteria (especially multiple criteria) was limited or absent in many of our systems. For example "which finding aids have subjects (or names) in common?" Once the data can be queried with SPARQL, it's pretty easy...
- #29 Our archivist, Elizabeth Roke, wanted to know if we could find all finding aids that had subject headings in common with names. This query will find them. We couldn't figure out how to do this with a report out of our ILS system... theoretically possible but would take a special app... or multiple searches. This is a small example of what you can do with SPARQL. Would our users learn SPARQL? Probably not, so more user-friendly tools need to translate.
- #32 Having a short time and not wanting to start from scratch, we took the EAD/finding aid stylesheet developed by the UK ArchivesHub project, and modified it for our own use. This was educational. ArchivesHub has a pretty precise way of expressing the relationship between a subject and archival collection... "AssociatedWith". They also use several vocabularies including FOAF. Some of this information might seem redundant. But, it could help others who want to use our data.
- #33 Dublin Core's really simple
- #34 Simile's handling of MODS is kind of... mods-centric. Along with ArchivesHub, the transformation mints some URLs for things like people and topics. We're starting to see the wisdom of this. What stopped us from using this? When querying the LC linked data, we needed the entire heading (including dates) in one string, and this expression didn't seem to have it... so we went for the simplest route for now.
- #36 None of example stylesheets have "FRBR". (If Robin hasn't already mentioned, explain)... Some of us liked the idea of using RDA vocabularies, but we discovered - there are some properties that can only be used with a work, or a manifestation, or an item.... Using those properties could automatically classify the thing we're talking about as, for example, a work, when we aren't defining it that way. This could ultimately be a good thing, but we decided we needed to do some more detailed modeling before using them. FRBR is somewhat controversial - word is that the LC Bibframe project model sidesteps it - only using "Work" and "Instance". Anyway, we better understand what having 4 entities representing the properties of a book can mean!
- #38 One of the powerful things about linked data, is the ability to assert equivalence between resources like concepts in two different vocabularies. Then, you can discover links to lots more resources from places like Dbpedia. But, what if they don't exactly match? Are the island, and the Prison different, or the same?
- #39 This would seem to go without saying, but I'll say it. We're interested (and some others at Emory will be working on) automated extraction of things like dates, people, places, etc. from text or metadata, but sometimes, you just have to have people do it or.. no connections. One of our subject specialists gave us some scenarios for scholars, they were interested in tracking people, places, battles, dates... but if our metadata has this information, it's ususally inside a textual note.
- #40 I've just given you a taste of what we've experienced... So, this sounds like a lot of trouble, and we still don't have the magical payoff or "killer app"... But, we have appetite for it now and we have lots of ideas for next steps as we carry on our explorations. If your organization is thinking about getting into linked data, getting your feet wet and involving people outside of your IT department is the way to have them REALLY learn. It's hard, but sort of addictive. Delicious, even. I'll stop now.