Information Architecture and Navigation, cont’d. from earlier post
Jacquelyn Erdman of East Carolina U. talked about her library’s web site redesign.
- No manual
- Roving librarian
- As much as possible, automated
Navigation goals (done with an MS Access db and ColdFusion):
- Facetted (e.g., publication types, peer-reviewed sources, etc.)
One-search (using Summon) and browse by topics.
Best bets (recommended site for topics), tags
Sidebar: recommendations, tutorials, contact info.
Search Discovery Tools, Greg Notess, Montana State U.
“Discovery Tools” are the hot new thing vendors are selling to libraries. But, according to Notess, Discover = metasearch. It’s the newest variation on the premise of one [Google-like] search box for all library content:
- Library catalog records
- Citation and abstract metadata (some or all)
- Full-text indexing (some?)
- Local digital collections
Some vendors with library web pages where examples can be seen:
- Ex Libris Primo (Vanderbilt)
- Worldcat Local (University of Washington, search box at the top of the page)
- Serials Solutions’ Summon (Arizona State U., U. of Michigan)
- Ebsco Discovery (U. North Carolina Greensboro, search box at the top of the page)
The difference between discovery and federated search is that discovery puts everything in one index and searches it, but federated search goes through each database separately. So, discovery is faster.
Problems with discovery:
- Search multiple databases through discovery is not the same as searching each one individually. About 90% of the data gets into the discovery index. (Is that OK? It means you’re paying for data that people might not see if they only use the discovery tool.)
- There’s also overlapping metadata between databases; duplicates are not always caught.
Montana State had problem with Worldcat Local (old WLN records didn’t show up) and Summon:
- Items deleted from the catalog didn’t get deleted from the discovery index
- Upgrades had issues
- Thousands of results from simple searches
- Problems searching for known items, such as database names, book titles, journal titles
Other problems with discovery tools:
- We don’t know when it’s searching full text
- Relevance ranking is not as good as it should be (for example, in the known item searches above, the known item didn’t always show up first)