Canadian <Metadata> Forum

Metadata in Public Libraries

Walter Lewis
Halton Hills Public Library

Also available in [PDF 189 KB]

The Halton Hills Public Library?

• Community of 50,000 people; 25 FTE including 3 professional librarians
• Fringe of the Greater Toronto Area
• Member of HALINET (HALton Information NETwork)
• Believers in the old notion of the "one place to look" (at least for accurate referral)

Metadata types

• Descriptive metadata
  • Resource discovery ...

• Administrative metadata
  • Creation, Rights ...

• Technical/structural metadata
  • Formats, location ...

• Collectively: Collection Management

Contexts for Metadata in the Public Library

• Traditional library information systems
• Community information
• Local historical and genealogical information

Traditional Library metadata

• Collection catalogue
• Indexing services (magazines and newspapers)
• MARC-centric services
• Purchased content and data management services

Community Information

• Directories of agencies and services
• Volunteer opportunities listings
• Events listings
• Developed web-centered software to facilitate collection, management and searching (www.cioc.ca)

Local Historical and Genealogical Information

• Newspaper
• Images (part of Images Canada)
• Books (including indexes of non-digitized material)
• Census, wills, property records, military records, cemetery, business directories, maps …
• 600,000 + records, millions of searches/year

Maritime History of the Great Lakes

• Power of Partnerships: Entirely volunteer, contributed content
• Personal site and Development test bed
• Newspaper articles (TEI)
• Transcribed and imaged texts
• Images
• Enrolment and registry databases
• www.hhpl.on.ca/GreatLakes

Data models vs metadata

• Single most important concept in presentation
• Metadata schemas are largely about how you encode content as you share it with other systems
• Good data models will allow you to support multiple metadata schemas
• The last thing you want to do is to derive a MARC or METS record from a database modeled to support Dublin Core

Examples (1)

• Export routine to transform the data on its way to Images Canada (modified Dublin Core)
• Working with one of the NCSA's Open Archives Initiative routines to support OAI (Dublin Core required; other transformations optional)
• Crossnet's Zedlib templates to transform the same dataset into MARC and SUTRS to answer Z39.50 queries

Examples (2.1)

• CIOC's gateway (http://gateway.cioc.ca/) uses the Crossnet Zedlib toolset to deliver MARC 21 (Community Information) records via the DRA/SIRSI TAOS Web2 interface
• Could as easily be integrated into a broadcast Z39.50 search oriented interface for a single library
• Need to be able to pass address information off to mapping agencies (from the native web interface)

Examples (2.2)

• Requirement to export to Microsoft Access for use in local derivative works.
• Requirement to export and import from the ACICO standard format while engaged in other joint projects
• Looming need for data exchange with the CCACs

Examples (3)

• Shares with the EAD (and HTML) the notion of embedded metadata in a head area to describe the content in the body
• Embed <index>, <name> tags in the body
• XSLT to create web pages with Dublin Core meta tags
• Export other tags to database to drive other kinds of searches
• XML and databases are complementary, not mutually exclusive technologies

Description in binary objects

• Challenge of orphaned digital objects
• Separate work flows
  • Digitizers
  • Metadata specialists
  • ... often for excellent reasons
• Orphaned projects (and content)
• Projects that are working on this issue, but where the priority is rights and technical metadata

Resource discovery

• Native interfaces
• Google vs the Deep Web
  • There are strategies to expose this content to Google etc.
• Harvested content
• Broadcast search

Native interfaces

• Most projects have one
• Rarely closely linked to the metadata ... schemas ...
• ... except when loading specific content into broader content management applications
  • e.g. the Milwaukee Public Library catalogue
  • (www.mpl.org/file/hum_marine_shipfile.htm)
• Maximum number of "rabbit holes" or "silos"
• Need to expose a crawlable interface to Google et al.  -  subject to their relevance criteria

Harvested content

• Open Archives Initiative Protocol for Metadata Harvesting (OAI-PMH)
  • Unqualified Dublin Core
  • Other metadata standards as configured
• Bi-lateral agreements
  • Images Canada (www.imagescanada.ca/)
• Directed web harvesting

Broadcast and Government Info

• Demand for simplified access to board range of government information
  • Municipal
  • Provincial
  • Federal
• Range of resource discovery options
  • Local, customized search interface
  • Support metadata and searching standards that allow organizations to define the interface, navigation elements that link the data

Broadcast searching

• Just-in-time searching
• One Place to Look
• Usually associated with Z39.50
• Can be built across a set of native databases
• Commercial applications (WebFeat, Agent etc.) combine techniques from screen scraping to being metadata aware

Parting thoughts

• The same data can support a wide range of metadata standards, as appropriate
• Customized local search interfaces still matter
• Standards-based interfaces that allow broader searching matter