1. Link Resolvers & Knowledge Base in Discovery Services By
Soma Sundaresh, Head of Product Development
& Jothy S, Product Manager, J-Gate, Informatics Publishing Limited, Bangalore

2. Can we perfect it?
Perfect linking is more of an ideal than an achievable goal.
One user's definition of a LR: ‘The link resolver makes the user’s path from citation to full-text slightly less tortuous. ‘

3. Trends
Increasingly the link resolver will come as components of broader systems. Transition away from Standalone Products
Support for multiple formats
Open API’s for interoperability and extensibility
KB and LR are mostly tied to discovery strategies.

4. Picture imperfect! Full Text Sources Discovery Point
Print Copy in Library:

5. Common complaints
#3 - Links Fail More Often Than You Think (Broken links as high as 30%)
“Close” might be good enough for horseshoes and hand-grenades, as the saying goes, but anything less than hitting the bull’s-eye is a failure when it comes to linking.
#2 - User experience matters
link resolvers often present multiple links to end users from multiple sources, making what and where to click a mystery.
#1 – Users Do Need Help to Succeed
Bad links are not often reported... Link resolvers can be more than simply links to content; they can serve as first gateways to library services, including access to real-time help, contextual recommendations, and available delivery services. If the link resolver is neither intuitive to navigate nor highly customizable, the library will lose out on opportunities to guide or get users where they need to go.

6. Common complaints
OpenURLs fail frequently - problems were due to variation in ISSNs, incorrect volume, issue, and page number information, and incorrect dates.
While metadata quality is important, the logical rules for creating OpenURL links using that metadata also play an important role in successful full-text linking.

8. DOI
Persistence
Accurate tracking
Easier automation
Creative collaboration (Interoperability)
ISO Standard compliance
Enhance the value of content when it has DOI

9. DOI = Digital Object Identifier Digital Identifier of an Object

10. DOI = Digital Object Identifier Digital Identifier of an Object

11. DOI
A DOI identifies an object permanently,
even if the object changes its location,
its owner or other characteristics
The DOI does this through redirection,
using the Handle system

12. DOI
The object has metadata associated with it. This metadata describes the object in whatever detail is appropriate for a service. The minimum set is kernel data and more can be added.
DOIs are supported by the DOI Community to offer persistence.

13. DOI Registration
REGISTRATION
AGENCY
DOIs are allocated via Registration Agencies who manage the DOI records, maintain the databases of metadata, and participate in the DOI Community

14. Persistence
A DOI continues to point to the object it identifies even if the object is moved, updated or modified

15. Interoperability
A common standard and agreed syntactical and semantic approach to metadata makes sharing data from different sources easier and more productive

16. Extensibility
Users can build on the standard to develop applications suited to their industry or sector

17. Platform independence
Allows data from different sources to be easily shared and facilitates multiple output formats

18. Dynamic updating
As DOIs are persistent, metadata can be added to or updated over time to further describe the object identified

19. OpenURL
OpenURL is a standardized format for encoding a description of a resource within a Uniform Resource Locator (URL), intended to help Internet users to find a copy of the resource that they are allowed to access.

20. OpenURL
The National Information Standards Organization has developed standards for OpenURL and its data container (the ContextObject) as American National Standards Institute standard ANSI/NISO Z
This Standard defines the OpenURL Framework Registry and the rules that govern the usage of this Registry.

21. OpenURL
An information ecosystem based on OpenURL has emerged to manage linking in a scalable and efficient manner.
In this ecosystem, sources present links from citations formulated as an OpenURL, following a standardized structure that presents metadata or identifiers describing the resource instead of a hard-coded link.
The link resolver then interprets the OpenURL and calculates the best link to connect the user with the full text of article, or presents other relevant services if the full text is not available.
Link resolvers rely on a knowledge base that describes the universe of available content resources, scoped to the resources made available by their library.

22. OpenURL Format
An OpenURL consists of a base URL, which contains the address of the user's institutional link-server, followed by a query string, consisting of key-value pairs serializing a ContextObject.
The ContextObject is most often bibliographic data, but as of version 1.0 OpenURL can also include information about the requester, the resource containing the hyperlink, the type of service required, and so forth.

23. OpenURL Example Version 0.1
baseurl.institution.edu/content? genre=article&title=eye&volume=23&issue=1&spage=161
version 1.0
&rft.genre=article&rft.title=eye&rft.volume=23&rft.issue=1&rft.spage=161

24. Link Resolver in J-Gate
Being well aware of the problem, we here at J-Gate wanted to remove the barrier between the discovery services and Link Resolver.
With this motto we have integrated LR with J-Gate seamlessly.

25. General LR architecture
Discovery Services
Library catalog
OpenURL 1
ILL
Link Resolver
HTML page with link options
OpenURL 2
FT Option 1
FT Option 2
KnowledgeBase
Service Y Text in translation
Others?

26. J-Gate Implementation
Search Term
PRINT
FT -DOI
Search Results
ILL
FT – Dynamic Links 1
URL Baking Unit
Cite Article FL
KB Interface
OpenURL
FT – Static Links `
Recipe dB KB
Service Y Text in translation
Others?

27. URL Baking Unit Architecture

28. Recipe Sample - 1
Publisher-1 #set ($a=$!{rft.issue}) #set ($b="temp-$!{rft.issue}") #set ($array=$a.split("-")) #set ($last=$b.split("-")) #if($a.indexOf("-") > 0) #else #end

29. Recipe Sample - 2
Publisher-2 #if($!{rft.doi} !="") #else #if($!{rft.eissn} == " ") #set($paddedSpage="00000$!rft.spage") #set($spg=$paddedSpage.length()- 4) #set($paddedvolume="00$!rft.volume") #set($vol=$paddedvolume.length()- 2) #else #if($!{rft.issn} == " ") #set($paddedissue="00$!rft.issue") #set($iss=$paddedissue.length()- 2) #set($paddedSpage="00000$!rft.spage") #set($spg=$paddedSpage.length()- 3) #else #end #end #end

30. Challenges galore Identifier Inconsistencies Title Inconsistencies
Incorrect Date Coverage
Inconsistent Date Formatting
Inconsistencies in Content Coverage Description
Embargo Inconsistencies
Data Format and Exchange
Out-dated Holdings Data

31. Our Eco System
J-Gate provides cross-database access to all resources of a library and therefore possibly eliminating the need for the any additional KB or link resolver, which will be a burden for any library to maintain one or buy separately.
We inform the user that there is print copy in the institutional library, (but without catalog information currently)
When the targeted article is not available in the library’s collection, it is available for consortia clients; For non-consortia clients, we display ‘Find in a Library’ based on union catalog.

32. Process
In-flow of Resources: KB targets are regularly maintained by J-Gate KB team to ensure the accuracy of both content and linking. Our team works with publishers to receive regular data feeds. Data received goes through rigorous quality assurance process.
Reporting errors by libraries/users: Users can report errors and omissions via the J-Gate customer support team by a workflow integrated in J-Gate (An incident gets opened automatically for the KB team to resolve the issue). We close any broken links issue within 2 workings days on an average.
Direct linking - when we know that user has only one link available, then the user is directly taken to the target destination, else the menu page is presented. Menu includes link (dynamic and static) to FT from publisher; in case of aggregators only dynamic link.

33. Strengths
J-Gate is non-biased (not driven by specific content providers).
We keep adding the packages from all major databases which are aligned with our clients’ priorities.
We add approximately 15,000+ article information to KB everyday. These are updated almost at the same rate at which source is getting updated.
The J-Gate KB maintains information about relations between objects.

34. Strengths
We have the largest collection of journals. Link Resolutions are done with various options : Dynamic OpenURL, Link by DOI and Static links as backup.
We continuously refresh the KB with FT sources information and J-Gate Resource Management Service module enables librarians to upload their holding details in a simple Excel format. This will help our link engine (LR) generate appropriate copy when a particular article is cited by patrons.
Updating holding details in J-Gate RMS is quite straightforward using excel with some easy steps.

35. Future Directions
Expand scope to incorporate higher coverage of data types, multimedia content, library catalogue etc.
Contribute to standards development initiatives
API for metadata citations
API for Link Resolutions which libraries may integrate into 3rd party applications.
Automatically update link resolver knowledge base from the client’s electronic holdings in their ERMS

36. Thank You