Date: January 3, 2006 12:11:55 PM PST
Cc: email@example.com, firstname.lastname@example.org
This talk will discuss the core development activities of the "Quality
Metrics" project at Emory's Woodruff Library. This project is being
conducted under an IMLS grant to research requirements for and build
a working prototype digital library search system.
What this project is doing that is new is truly generalizing and
integrating explicit and latent quality indicators which allow
users to ascertain the fitness of digital library resources. Most
search engine components have only one indicator: content-query
similarity ("relevance"). Google only has two, adding PageRank to the
latter. Our system, QM-search, will have an unlimited number of these,
which will be customizable by the digital librarian for the target
community and collections, and even customizeable from user to user or
search to search.
Some basic examples of quality indicators that digital libraries might
be able to exploit would be activations (views online or check-outs in
circulation), selection (compilation in "bookmark" lists online or
additions to course reserves lists), extent of review (from a peer-
reviewed journal, conference, or not?), or citation-based metrics.
The ouput of QM-search will be in a completely generalized XML format,
with the search results represented as a structure based on the
structure specified in the input "organization spec". This XML output
can be transformed into presentation HTML resembling anything from a
"linear" Google-like search results list to an A9-like column display to
more exotic groupings and breakdowns.
Requirements for QM-search are being gleaned from focus groups being
conducted at Emory (preliminary results will be shared), and development
is being conducted as a high-level layer atop the excellent Lucene open
source search engine project.
Head of Digital Library Research
Emory University General Libraries
President and Founder, PlanetMath.org