This session was packed, so clearly this is a subject of interest to a lot of us.
Is Federated Searching Good for Sci-Tech Libraries?
Speaker: Tate Nunley, exLibris
Tate Nunley (exLibris):
- very much a gray area: whether federated
searching is good for sci-tech libraries is entirely dependent on the sci-tech
library
- metasearch: a single query form used to access a diverse number of information
sources
- metasearch uses translators/connectors to fit
the incoming query (metasearch query form) to the specific requirements of the
individual database
- metasearch systems provide the capability to
search heterogeneous resources
- for novice users there are often scope lists
- for advanced users there is control over the
search scope, and further capabilities for result manipulation
- access management: authentication (establishment of network
identity) and authorization (resources a particular authenticated user can
access)
- resource discovery: electronic collections
are often not used optimally, metasearch can assist with this problem
- resource access: linking to native interface,
searching using a unified interface, but linking to the native interface for
results, searching using a unified interface and viewing results via a unified
interface
- importance of a knowledge base: provides
descriptive collection information and functional information (how the
metasearch interacts with the separate resources)
- what does metasearching take: searching is adapted to the user's expertise
and expectations and should also be adapted to the requirements of the database
- query interface can be very simple or very
complex or somewhere in between
- building a query: each query is adaptable, configuration should
apply rules that define the flow, the interface, and the manner of searching
- the configuration should deal with: access
mode, password control, URL creation, optimization of query (syntax, field
mapping, phrasing of authors names, normalization issues), character conversion
- processing the results: all results should be
converted to a unified format, should apply rules that define the manipulation
of the retrieved data, should deal with cataloging format, script, special
characters, normalization of data for OpenURL creation
- metasearch can provide: uniform display,
ranking, comparison, OpenURL
- display of result: one merged list or lists
per resource? Or both? how do you get the good results on top?
- various approaches: dependent upon the user,
users can choose, and then change their mind
- how many resources can you search at once? It
depends on your users, but setting a limit is important.
- Searching too many: unnecessary burden on
users, unnecessary burden on info providers, unnecessary burden on lib
resources
- the solution is to focus on APPROPRIATE
resources
- how to make the good results show up first?
- users expect results by relevance, reliance
on order of arrival is not good enough, relevance ranking enables the system to
show the best hits at top, ranking is limited to the records retrieved, future
directions: further analysis of the results
- clustering may be key in future developments
(Vivisimo as model rather than Google)
- clustering allows more freeform subject
headings
- integration: linking to the metasearch system, deep-linking into the metasearch
system (embed a search box, guide users to the information they need with a
scope of resources)
- additional tools for integration: XML web services (API) to extract resource
descriptions, extract lists of resources, use the metasearch system as a
"portal"
- local integration: OPAC, institutional
repositories
- standards: the NISO Metasearch Initiative, a response to content provider/service
provider concerns
- NISO distinguish library offerings from free
offerings, allow metasearch providers to offer more effective services,
provides a way to implement a common search with a simple shared protocol
(SRU/SRW as a starting point, as well as OpenURL 1.0)
- called NISO MI XML Gateway
- only beneficial, in the end, if it increases usage of databases
