Quicksearch Initial Testing Results

On September 9th, the Quicksearch Implementation team sucessfully hosted our first testing session.

Overall reception of the interface was good!

The primary goal of the testing session was to test searching and faceting in Quicksearch. Participating were 3 Public Services staff, 2 Rare book catalogers, 2 language specialists, 2 staff from Enterprise Systems, and 6 members of the Discovery Metadata group. We asked testers to evaluate the Quicksearch test interface using the Quicksearch Test Script as a guide, in addition to their favorite searches.

Results from our test clearly showed that Keyword searches in Quicksearch work much better than Keyword searches in Orbis. Relevant results are more likely to rise the top of the result list, and results in a result set are more likely to be relevant. This is good news moving forward as we begin now to focus on fine-tuning metadata display issues and on developing Patron Services like Booksmarks, Requesting, etc.

The testing session also revealed a number of bugs, from minor display problems to two major software bugs, for which we sincerely thank the testers! In addition, they requested some interesting new features, like the incorporation of the BMEC (Bass Media Equipment Catalog) in Quicksearch. Their careful testing has been invaluable in helping us identify and prioritize the issues we need to work on next.

After testing, the Discovery Metadata group engaged in an intensive 3-day sprint to go through the 65 responses, categorize them, and make recommendations for the metadata-related issues. (What is a Sprint? A sprint, or scrum, is a concept adapted from the software development world. The team identifies a list of goals, and meets for designated period of time to work on that goal. The team works on that goal and that goal only during the period of the sprint, and meets daily to report on progress and make plans for the next day’s work). The results of the testing and the three-day metadata sprint that followed are summarized in two publicly-available spreadsheets, Reported Issues, and Requested Features.

The Quicksearch Implementation team is currently hard at work resolving issues identified in testing, and doing a full extract and re-load of Orbis and Morris. The Reported Issues and Requested Features spreadsheets will be updated once a week on Friday so you can follow these issues as they are resolved. We expect to continue working on these issues through the remainder of the semester.

Next up: Our goal for this week is to release a Beta version of Quicksearch to all Yale Library Staff for review.

What is a discovery service?

According to Marshall Breeding, a discovery service is a system that searches “seamlessly across a wide range of local and remote content … [providing] relevance-ranked results.”

Marshall Breeding “Web-Scale Discovery Services“, American Library Association, January 14, 2014

With the arrival of Quicksearch beta, the Library will have two large scale discovery services in its arsenal.

Articles+, powered by ProQuest’s Summon discovery service, allows the user to search through electronic journal articles, images, newspapers, and abstract and index content, as well as our own LibGuides. Articles+ is a production service at Yale, supported by LibraryIT staff and maintained remotely by ProQuest. Along with our eJournal title list and our database A-Z list, the Articles+ discovery service is our patrons’ access point for licensed electronic resource items.

Quicksearch beta is also a large scale discovery service, but has several differences in structure, support and scope in comparison to Articles+. Quicksearch beta is a locally maintained service. It is powered by the open-source Blacklight framework, and is being developed in tandem with our digital collections repository service, findit.library.yale.edu, by LibraryIT staff. Because Quicksearch beta is a an open-source system that the library supports and maintains locally, there is more potential in presenting our data in specific and unique ways. Articles+ uses a central index of over 11.4 billion items, maintained by ProQuest, and it is ProQuest who ultimately decides how that data can be presented.

Like Articles+, the Quicksearch catalog search runs off of a central index; but unlike Articles+ this index lives on a Yale server and contains only Yale catalog records. Quicksearch uses a “bento box” approach to execute simultaneous searches against Morris and Orbis catalog records (the Catalog search) and Articles+ itself, directly from the Quicksearch beta landing page. In this way, a search for a book title might come up with the book in the catalog search, and reviews for that book in the Articles+ search.

Quicksearch beta is just that- a beta product that is still in development. If you’ve never heard the term before, PC Magazine has a useful definition: “A pre-release of software that is given out to a large group of users to try under real conditions.” Quicksearch beta is therefore not a fully developed product like Articles+ is now, but will be eventually!

For more information on discovery services at Yale, particularly Articles+, please see this handout from August’s Rediscover Discovery forums.

What’s in Quicksearch? Orbis records

Now that we’ve described what records from Morris, the Yale Law Library’s catalog, are being indexed in Quicksearch, let’s talk about records from Orbis. The main library catalog, Orbis, holds around 12 million bibliographic records. Of these, pretty much every record that is currently live in our online catalog will be indexed into Quicksearch. This includes:

-electronic journal titles from Serials Solutions
-In Process, On Order, and UNCAT records (unsuppressed, of course)
-records from large electronic resources like the Making of Modern Law, Eighteenth Century Collection Online, Making of the Modern Economy, Making of the Modern World, and more
-all other records currently available in the Orbis online catalog

As Scott noted in his post on Morris records in Quicksearch, we are not deduplicating records between catalogs at this time.

Stay tuned for more updates on the Quicksearch project!