Search July 15 2008

Search sub-group

July 15, 2008 at 11:00 a.m., MC 1058

1. Administration & assignment of minute taker

Present: Megan, Glenn, Lauren, Kevin, Paul, Allan (delayed)

Minutes by: Kevin

2. Clean-up/further discussion on Google search appliance

Subdomains

  • groups of multiple subdomains (i.e. Information Systems & Technology(IST) servers) are set up in "collections"
  • in fact, every space needing its own search needs its own collection, even if it is a single subdomain
  • collections can be set up in bulk when we start, then additional sites would be added one at a time on an ongoing basis
  • given the setup requirements, it was asked if we really needed individual subdomain searches
    • new sites might not care about individual search, but amongst existing sites, some may not really need it, but hesitate to lose it given they already have the capability using the current search
  • it appears we can do "grouping of groups", that is, one collection that contains other collections
  • alternatively, you can specify that multiple collections should be searched by passing the keywords through the SITE parameter
  • collection-based reporting is available
  • Paul asked if there was a way that missed searches could be distributed to the site owner(s). Unknown if the appliance supports this.
  • Megan will review other Universities' practices with respect to subdomains

Controlling access

  • in order to add access controlled results, it looks like we would have to maintain a manual list of user IDs for each protected document
  • given this, who would really need this feature? How necessary is it?
    • one thing this would be particularly useful for, if it were easy to set up, is searchability of SEW course notes, which are currently password protected
  • it was decided that it was not a dealbreaker if this feature was not usable

Indexing

  • the advanced indexing might help with public content that is currently hard to index, like ANGEL

Analytics

  • while the appliance can automatically add Google Analytics, we need to determine if this only supports a single, university-wide code, or if each "collection" could have its own Analytics code

3. Yahoo BOSS service - any thoughts/implications?

Not formally discussed, other than to note that "old Yahoo  application programming interface (API)" accourding to the BOSS website was limited to 5,000 queries, so current UW search is really limited to 15,000 (10,000 Google + 5,000 Yahoo).

4. Draft report started; review of goals/objectives, report format (backtracking a bit here!)

  • debate over whether keywords were "required" or "useful"
  • moved keywords and UWdir points out of "required" section into new "useful additions" section
  • added "documents indexed" to the "no limits" point
  • added "no ads" to the "fully integrated" point
  • added "ideally without requiring JavaScript support" to the "browsers/systems" point
  • in the new "useful additions" section, added:
    • generate Extensible Markup Language (XML)
    • individual collection reports accessible by individuals
    • decentralized maintenance for keywords
  • other things discussed than were not added to our objectives or "useful additions" were:
    • distributed access to configuration
    • per-site configuration

5. Other discussion

  • debate if we should step back and evaluate other solutions
  • "What's broken now?"
    • current search is by all appearances functional, but theoretically Google could remove access to the APIs in use at any time
    • Allan advocates that we fix the current search with the free custom search, to avoid exposing the university to risk, then look into other options
      • we could combine the custom search with AJAX to support subdomains
      • there is a significant time investment to do this
      • doing this would likely delay other solutions by at least a year
      • would know the answers to Content Manangement System (CMS) questions by then, though unsure if CMS actually has any special search needs
    • Kevin believes further discussion is necessary
    • Paul suggests we bring this to Web Advisory
      • he will arrange to get this put on the agenda to get opinions
      • do we want a short term fix or a substantial improvement?

-- KevinPaxman - 15 Jul 2008 -- MeganMcDermott - 15 Jul 2008