CUIPID 4: Building a faceted searching and browsing interface for your library catalog

(note: the preconference material referred to the software as CUIPID 3, but a new version has since been completed)

CUIPID (pronounced “cupid”) is the University of Rochester’s Catalog User Interface Platform for Iterative Development. Built by their Digital Initiatives Unit, it serves as an experimental base for library catalog enhancements.

David Lindahl and Jeff Suszczynski were on hand to walk us through what CUIPID is, as well as some insight into the development process.

First we learned a bit about just what the Digital Initiatives Unit is. The staff of 8 (including a wide variety of non-librarian disciplines, such as an anthropologist) performs constant research into user needs via work practice studies and other methods. They just finished a 2 year comprehensive study of how undergrads write papers. We saw a video clip from interviews conducted at night in a UofR dorm, which was both informative and quite funny. Some interesting facts learned: Freshmen don’t stop at just the first three results of a search, and are not afraid of the reference desk. And most are less capable with technology than expected.

On to CUIPID, which has gone through a lot of changes. Version 1 used a small subset of records in MARCXML format, and solved 80% of UI issues they had previously identified. It collapsed similar editions via text matching, used the Google spellcheck API, etc. But unfortunately it wouldn’t scale up well – the license for the Verity indexing tool they used would be prohibitively expensive to use for the full sized catalog.

Next came SARA (I didn’t catch what the acronym stands for). It was a home made metasearch engine, covering all types of material the library holds – books, web sites, video, subject guides, databases, etc. It ran multiple concurrent queries to the catalog – no need to select author, title, etc. before the search. Users would narrow results by type later. Unfortunately, SARA had extremely debilitating performance issues.

CUIPID 2 was built on a trial version of TextML, an XML database product. It’s interface and features were similar to SARA’s, and also faced scalability issues. In addition, TextML would not have been free for the full version.

CUIPID 3 indexed more than 3 million records, using MS SQL server and ColdFusion for the interface. It was pretty similar to the current CUIPID 4, which was covered in more detail:

CUIPID 4′s features are going to be slowly integrated into the U of R’s existing Voyager catalog. The name has been informally changed to simply C4, “because it’s fun” to say. It follows the previously described faceted method of searching, letting the user drill down to the correct categories of results. Their inspiration for this system came from web sites like Sears’ and Home Depot’s.

It interfaces fully with the current list of student login information, allowing services like placing holds and recalls. There’s a number of relatively small features that are nice touches – displaying contextually appropriate metadata, for example. So for a movie result the director gets displayed, instead of the author for a book. The system makes extensive use of various APIs, pulling in external data like Amazon’s book covers and reviews, recent blog posts about a title via Technorati, etc.

Describing CUIPID 4 admittedly sounds sort of dry. But we got to see a live demo of the system, and it really blew me away. The interface is very intuitive, response time is fast, and it seems to be a pretty polished product even now.

Features to be added in the future include replacing the Amazon images with local copies, imiproved acceptance of unicode in catalog records, holdings records, and FRBR functionality (either homegrown or via OCLC’s system).

A separate project of the Digital Initiatives Unit was mentioned briefly – the eXtensible Catalog (XC). While still in early pre-planning, ultimately they hope to make the XC an open source catalog to hold all types of collections. It will be designed to be experimented with, and be compatible with your existing ILS and any form of scripting (PHP, ASP, CF, etc). Sounds like a very exciting project to me – more information is at www.extensiblecatalog.info.

This presentation had a huge amount of data for me to take in, but I’m glad I went. It was really interesting to see some of these catalog innovations in practice.

One thought on “CUIPID 4: Building a faceted searching and browsing interface for your library catalog”

  1. Well, isn’t this an interesting counterpoint to the faceted search based on a vendor product, Endeca, as used at NCSU and described in earlier sessions at LITA Forum.

    What this has in common with their solution is a lot of technical staff, compared to my library (we have one person responsible for the public web site and the staff site — that would be me; and one person responsible for the ILS — that would be my cubemate Mike). By public library standards, we already have a huge systems staff. Granted, it’s way smaller than Seattle’s or even San Jose’s, but most public libraries have even less to work with — both in terms of staff, and in terms of budget.

    So I’m glad to see this move towards making advanced features available via open source! And beyond that, I think we need to start considering how to make these solutions not only available but install-out-of-the-box easy for many of our colleagues at smaller libraries. How can we take the next step to make every library patron have as easy a search experience as the undergrads at these academic libraries?

Comments are closed.