Text linking in the humanities:citing canonical works using OpenURL

CNI Spring 2009 Task Force Meeting

Eric Rebillard

Departments of

Classics and History

David Ruddy

University Library

Planning grant overview

• Mellon Foundation funding• Three components

– Canonical citation– OpenURL quality metrics and evaluation– L’Annee interface enhancements

Linking solutions

• Bilateral, direct vendor links– Costly to implement and maintain– Unstable– Does not easily allow one-to-many relationships– Does not address ‘appropriate copy’ problem

• A method with more indirection, such as OpenURL

OpenURL

Classical text resource

(e.g., JSTOR, L’Annee)

Source text in translation

Source text in translation

Library catalog

Library catalog

Inter-library loan

Inter-library loan

Source text in original language

Source text in original language

Others?Others?

HTML page with link options

HTML page with link options

User’s link resolver

OpenURL

• OpenURL: a package of metadata about a resource, encoded within a URL and sent to a user’s link resolver

• OpenURL Framework (2004) generalized and formalized OpenURL 0.1 (2000) architecture so that services can be extended and new applications can evolve

• Focus on allowing “context-sensitive” services related to a specific resource

Advantages of using OpenURL

• System/vendor independent• Uniform linking syntax• Minimizes cost of creating and maintaining

links• Easily allows one-to-many linking• Allows appropriate copy

OpenURL in practice

• Define a metadata format that can reliably reference canonical citations

• Devise a feasible implementation plan

Metadata format: Identifying the work

• Canonical citations are at the level of the FRBR work

• Very little metadata (a title, possibly an author name)

• An undefined work ID included in metadata format– Community defined?– ISTC?

Metadata format: Author names

• Existing OpenURL metadata formats expect modern, western names (aufirst, aulast)

• How to capture ancient author names?– Allowing multiple ways to encode author forms,

which may depend on era of author

aufirst auform1aulast auform2

Metadata format: Citation components

• Two possible approaches:– Try to define every component

• Book, section, canto, stanza, act, scene, etc.

– A more abstract approach, recognizing the typical hierarchical structure of works

• level_1, level_2, level_3

• Second approach more generally applicable across heterogeneous material

Implementation challenge

• Normally, resolution to resource is left to local link resolver

• And yet:– Detailed and specialized knowledge – Uncertain commercial incentive for link resolver

vendors

• Proposed solution: domain-specific, community-supported knowledge bases

OpenURL

Classical text resource

(e.g., JSTOR, L’Annee)

Service X Text in original language, Edition B

Service X Text in original language, Edition B

Service Y Text in original language

Service Y Text in original language

Service Z Text in translation

Service Z Text in translation

Service X Text in original language, Edition A

Service X Text in original language, Edition A

Others?Others?

HTML page with link options

HTML page with link options

Classical Works KnowledgeBase

(CWKB)

Problems with this approach

• Doesn’t solve appropriate copy problem• Doesn’t allow for institutional branding,

presentation, page layout• Doesn’t allow for local options (library catalog

search, etc.)

OpenURL

Classical text resource

(e.g., JSTOR, L’Annee)

Inter-library loan

Inter-library loan

Library catalog

Library catalog

Others?Others?

HTML page with link options

HTML page with link options

Classical Works KnowledgeBase

(CWKB)1

User’s link resolver

OpenURL2

Service Y Text in original language

Service Y Text in original language

Service Z Text in translation

Service Z Text in translation

Advantages of intermediate resolver / knowledge base

• Domain specific knowledge base can augment and/or normalize metadata values

• Can provide specialized linking information• Can allow direct access to targeted resources

for users without a local link resolver • Could provide other potential services to

community

OpenURL

Classical text resource

(e.g., JSTOR, L’Annee)

Inter-library loan

Inter-library loan

Library catalog

Library catalog

Service XText in original language

Service XText in original language

HTML page with link options

HTML page with link options

Classical Works KnowledgeBase

(CWKB)1

User’s link resolver

OpenURL2

Service Y Text in original language

Service Y Text in original language

Service Z Text in translation

Service Z Text in translation

CWKBURLredirectservice

CWKB: Classical Works Knowledge Base

• A relational database that stores metadata about classical works

• A link resolver software that does the lookup in the relational database

CWKB: the database

CWKB: Author’s name

• The Latin form is the international authority form for Latin and ancient Greek authors.

• For lookup purpose, CWKB will store the established form of the author in several modern languages and the abbreviations used by the main reference tools.

• Draft ISO standard 27729: International Standard Name Identifier (ISNI)

CWKB: Work’s title

• The Latin form is the international authority form for Latin and ancient Greek titles.

• For lookup purpose, CWKB will store the established form of the title in several modern languages and the abbreviations used by the main reference tools.

CWKB: work identifiers

• They may come from a range of well-established registries such as:

• They may come from a range of well-established registries such as:– The Canon of the TLG– The Clavis Patrum Latinorum– The Clavis Patrum Graecorum– The CHS Canon of Greek Literature– The Stoa Canon of Latin Literature

• CWKB will maintain a concordance between them.• One development to follow: the ISTC standard (ISO 21047).

CWKB: the link resolver

OpenURL

Classical text resource

(e.g., JSTOR, L’Annee)

HTML page with link options

HTML page with link options

Classical Works KnowledgeBase

(CWKB)1

User’s link resolver

OpenURL2

OpenURL 1

Baseurl: cwkb.org/in/r.php? OpenURL version: ctx_ver = Z39.88-2004Metadata format: rft_val_fmt = info:ofi/fmt:kev:mtx:canonical_cit Citation: rft_auform1 = Aeschylus

rft_titleform1 = Suppliants rft_slevel1 = 40 rft_elevel1 = 57

 User’s resolver: res_id =

http://resolver.library.cornell.edu/net/openurl/? Source’s id: rfr_id = info:sid/aph

OpenURL 1 (cont.)

http://cwkb.org/in/r.php?ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:canonical_cit&rft.auform1=Aeschylus&rft.titleform1=Suppliants&rft.slevel1=40&rft.elevel1=57&res_id=http://resolver.library.cornell.edu/net/openurl/?&rfr_id=info:sid/aph

OpenURL 2

Step 1

The link resolver software looks up in the database for a match on the author’s name and the work’s title.

Step 2

The link resolver software adds data to the incoming OpenURL 1:

• authority forms of the author and the work (used for the lookup in the local library catalog);

• data on the services providing full text for the citation sent as an OpenURL ServiceType identifier.

The ServiceType identifiers contain the name of the service and the full text url.

Ex.  &svc_id=info:sid/cwkb.org:perseus_eng:url:http://www.perseus.tufts.edu/hopper/text.jsp?doc=Perseus:text:1999.01.0016:line=40.

If the service does not accept GET requests, the link goes to CWKB and is resolved.

OpenURL 2 (cont.)

http://resolver.library.cornell.edu/net/openurl/?&ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:canonical_cit&rfr_id=info:sid/cwkb.org&rft.auform1=Aeschylus&rft.titleform1=Suppliants&rft.titleform2=Supplices&rft.auform2=Aeschylus&rft.slevel1=40&svc_id=info:sid/cwkb.org:tlg_demo:url:http://cwkb.org/in/tlgpost.php?package_id=tlg_demo%26local_package_id=0085001%26scheme=z%26hidden_z=40&svc_id=info:sid/cwkb.org:tlg:url:http://cwkb.org/in/tlgpost.php?package_id=tlg%26local_package_id=0085014%26scheme=z%26hidden_z=40&svc_id=info:sid/cwkb.org:perseus_eng:url:http://www.perseus.tufts.edu/hopper/text.jsp?doc=Perseus:text:1999.01.0016:line=40&svc_id=info:sid/cwkb.org:perseus_grk:url:http://www.perseus.tufts.edu/hopper/text.jsp?doc=Perseus:text:1999.01.0015:line=40

Local Link Resolver

CWKB will provide a set of instructions for:• Creating a resource profile for each canonical citation

package the library licenses or supports;• Establishing rules for displaying the full text links;• Creating rules for mapping canonical citation

metadata to non-full text services, such as the library catalog searches.

Local Link Resolver

• Recognize the format of the incoming OpenURL. Each one will contain this string: rft_val_fmt=info:ofi/fmt:kev:mtx:canonical_cit

• Parse each incoming canonical citation in its key value components: – Pull out each svc_id into two parts (name of the

service; full text url);– Use auform2 and title2 for the display of the citation

requested and for the search in the local catalog.

Local Link Resolver

The OpenURL 2 is thus interpreted by the local link resolver to produce a page such as this:

Wider application

• Potential value to any discipline that cites works independent of specific editions or translations

• Potential model for chaining link resolvers / knowledge bases together to provide enhanced services to users

• Potential for providing other information services to community