9.09.2010: Reference team will meet to talk about what market basket libraries are doing with multiple database search.


Library tech committee member attendance optional.

Here is what Katie compiled and Fran sent out to the reference team:
=======
Before the Sept. 9 reference team meeting, check out what a few of our market basket libraries are doing with multiple database search. What do you like or not like? Fran suggests checking out RIT, Georgia Tech, and MIT. I thought Rice and Dartmouth were interesting, too. You also might want to compare different products or look at libraries that are using the same product to see how they're handling them differently.

Here's a list of links to our market basket libraries (add others if you want)
http://rpi-lib-tech.wikispaces.com/Market+basket+libraries

For each library, I noted what product they're using and added a link path to their multi-database search (sometimes there's more than one). Hopefully this will help save your time! Feel free to add your notes to the Market basket libraries wiki page if you want!

Katie
========

Discussion: Thoughts on market basket libraries?
  • Some libraries have a product for which they created a custom interface. The customized ones look the best, but how much work does this involve? The libraries with well designed interfaces have much larger IT support staff or departments.
    • But, RIT, Georgia Tech are using an open-source interface (Xerxes, developed at Cal State) over Ex Libris' Metalib. Probably that would be less work than to develop an interface from scratch? Investigate.
    • We really like how RIT is doing things. Georgia Tech is good, too, but RIT's page design makes more sense. (They are both using Xerxes interface over MetaLib software.)
    • MIT is nice, too - custom programmed interface over Metalib that probably involved a very large committee and great IT support.
  • How are people who are simultaneously searching a bundle of databases and retrieving significant hits going to get meaningful results?
  • How do the different multidatabase search tools rank results?
  • How does continued cancellation of resources impact on this investment?
    • It is possible endowment money might pay for this - money that is earmarked for technology improvements and enhancements.
    • It's good to have formalized our examinations and to have recommendations in the event funding becomes available.
  • Retrieving a giant result of articles and other materials, like the "Summon" search engine at Dartmouth College, doesn't appeal to us because of the diversity of our user groups. Different people need different things. Our group also didn't like that "Summon" uses a mega-aggregate index instead of actually searching source databases.
    • KD noted that Summon seems more popular generally (based on our initial fed search exploration) than it is within our market basket libraries. Maybe it's not as good a fit for engineering/tech schools? Of our group, only Dartmouth is using it - and they are a 4-yr liberal arts college.
  • We think it's important to be able to tell what you're searching and what you're not searching.
    • RIT does a nice job of this. Federated and un-federated databases are presented together on the subject database pages, and it's pretty clear which ones you can search together and which you have to search seperately.
    • Georgia Tech's article search pages mirrors our current structure: ("Most useful"--like our "Best Bets"--are grouped together and searched by default, although you can select others, too.
    • MIT had a page explaining limitations of the search. Useful.
  • Also, it is really important for us to be able to search different groupings of databases for different subjects. It would be nice for users to be able to select databases on the fly, too (RIT, etc.)
  • Maybe a group of us could go visit RIT - talk to them about how it's working, how much work it was to set up.
  • What do we want to be able to do that we can't do now? Can we already search multiple databases from same vendor at same time, like CSA, Scitopia, others?
  • We should keep in mind the question from Gary Schwartz (Director, CMT) at one of the discussions: What are we losing if we don't do this? Some of the possibilities are noted below:
    • Are we not meeting user expectations? Are our users unaware of the depth of our resources? Do people avoid using our resources because of our interfaces?
    • Maybe federated search would give a more promotional or educational entry point to our databases. And encourage those who might be tentative or overwhelmed?
    • (KD after-meeting note: would be good to look at stats pre- and post- implementation, to see what effects are on things we could actually measure. I think I saw a webinar where a library did this. I will look for the information)
    • Maybe federated search could help students meet information literacy goals for school or university accreditation.
  • The group is not sure how to approach the dual tasks of improving catalog search vs. improving database search. We have not seen these combined in a useful, non-confusing way. The tools that work well for one may not work well for the other. It might be hard to move forward on both at the same time, though.
    • Can Xerxes sit on top of catalog as well as databases?
  • Encore - is III's discovery interface. Their representative, E. McDowell was here today.
    • ConnectNY is in the process of implementing Encore for the ConnectNY catalog
    • Encore is primarily a catalog front end right now. The federated search component of Encore (Symphony) is just coming out now, and doesn't seem like it has the kind of functionality that we're excited about (tabbed results? don't we want one list of results?) We'd still like a demo, though.


What do we want to know? (link here)
Action items are in green
Followup to action items are in purple
  • At this point, it seems like we're primarily interested in finding out more about:
    • Ex Libris MetaLib with the Xerxes interface (because of RIT and Georgia Tech's implementations)
    • Encore (because it's III and ConnectNY is in the process of implementing it for the ConnectNY catalog)
    • (KD note: these two products have really different functionality right now. Encore is a catalog front end, just coming out w/ a fed search component for articles. MetaLib is an article search primarily and may or may not be able to search the catalog as well.)
  • We want to talk to people who've used these (Georgia Tech, RIT) - how are they working? How much work were they to develop/maintain? There is interest in visiting RIT. Fran/George/Bob have contacts there.
    • How is MetaLib/Xerxes working out for people who have implemented it?
      • Informal gathering of impressions from professional contacts at RIT.
        • Fran will be talking w/ her counterpart at RIT soon about something else. She will also ask about her impressions of MetaLib/Xerxes
        • Bob and Pat will be at the ConnectNY meeting Thurs/Fri Sept. 23-24 - if there is time, they'll ask the RIT attendees about MetaLib/Xerxes.
          • Notes are available from an October 11 conference call with RIT. Attending from RIT: Shirley Bower (Director of RIT Libraries), Jonathan Jiras (Technology Librarian), and a co-op student who works on Xerxes 10 hours/week.
      • Georgia Tech May 2010 presentation on MetaLib / Ex Libris here: http://hdl.handle.net/1853/33098
    • How difficult was it to get the Xerxes interface working over MetaLib? What customizations did they make to Xerxes? How much work was all this?
      • Katie and George will contact someone from RIT and/or Georgia Tech to get a general sense of how much effort it was to get running (out of the box? with customizations? how much customization did they do?), how much on-going maintenance it takes. George will check out the Xerxes webpage for background information and George and Katie will come up with a short list of questions before contacting people.
        • Georgia Tech May 2010 presentation on MetaLib / Ex Libris here: http://hdl.handle.net/1853/33098 - these would be good people to get in touch with.
        • George looked at the Xerxes documentation online and didn't see any big red flags as far as setup. He is wondering how authentication would work for us, since we use a combination of LDAP and the patron authentication database in Millennium.
        • Notes are available from an October 11 conference call with RIT. Attending from RIT: Shirley Bower (Director of RIT Libraries), Jonathan Jiras (Technology Librarian), and a co-op student who works on Xerxes 10 hours/week.
    • We could also talk to RIT about their involvement with the Extensible Catalog project (U Rochester). Is this something we should get involved in?
      • Katie will investigate status of eXtensible Catalog project. Functionality? Have they rolled it out already? If so, who's using it?
        • The eXtensible Catalog is still in development; nobody has it rolled out as a live public catalog interface yet. In our RIT conference call, Jon Jiras estimated it may be about 18 months before the public can use it. Based on our conversation with RIT, it seems like involvement in the project at this stage consists of committing developer time from your institution to develop specific functionality in the software.
    • Potentially visiting RIT to talk about MetaLib/Xerxes and eXtensible Catalog: Katie will ask Bob whether he might be able to fund a trip if it would be useful. We would need to be clear on what we would like to accomplish with an in-person visit.
      • I (KD) think we got a lot of the value of an in-person visit from the conference call. I don't think that a in-person visit is necessary at this point in our process. We can keep the idea in mind for the future, but we should also consider conference calls as a viable alternative.
  • We would like to see demos from III (Encore) and Ex Libris (MetaLib). III customer sales person (E. McDowell) was here today and may be able to come sometime this Oct./Nov.
    • Katie will check with Bob at beginning of October to see if Ericka has a more specific idea of when she might be in this area again.
      • Ericka McDonald, III Customer Sales Consultant will give a presentation on Innovative Interfaces' Encore discovery solution Wednesday, Nov. 11, from 10-11:30 in the Fischbach Room.
  • Functionality
    • What do results pages look like? What are your options for working with results? This is hard to tell by looking at other websites, because of our non-authenticated status.
    • How do the tools rank results? How is the ranking determined?
      • Katie will look for documentation.
        • "Xerxes use Metalib's ranking algorithm. You may be able to tweak the ranking algorithm on MetaLib, but I am not sure.""--email from Jon Jiras at RIT.
        • We would have to ask Ex Libris about what factors go into MetaLib's relevance ranking algorithm. We may not be able to get much information out of them, though - the specific algorithms are almost certainly proprietary. What they are able to tell us might not help us much. We might be able to get a sense of the quality relevance ranking by searching the Xerxes demo site.
  • How much does MetaLib cost? Initial? Ongoing?
    • Katie asked Pat to get a quote on MetaLib from Ex Libris.
  • Can Xerxes serve as the front end to a catalog?
    • Katie will look for documentation.
      • There is a Xerxes demo site for books (link to it from the intranet), but it doesn't seem to be as well-implemented as the article search.
        • It doesn't allow much flexibility in searching, compared to a typical library catalog - fewer options.
        • It searches a local library catalog, a consortial library catalog, and WorldCat, but not at the same time. The default search is for the local catalog. If there are no matches in the local catalog, it prompts you to search the regional catalog. If it doesn't find anything in the regional catalog, it suggests you search libraries worldwide. There is no obvious way for the user to choose whether to search the local, regional, or worldwide catalogs, or to switch back once you're in one..
        • It's quite slow (even for federated search!) and it timed out frequently on me.
        • Try searching for Palmer Ricketts Baker to see how it directs you to larger resources after a failed search.
      • I looked at five libraries that use Xerxes: Georgia Tech, RIT, and CSU Northridge and Fresno. None of them use Xerxes as the interface to their catalog. Perhaps it doesn't have enough search options available to do true catalog searching? Or maybe it seems too slow to be worthwhile.

How to find this stuff out / what to do next:
See above.
Action items are in green
Followup to action items are in purple