One of the hottest topics in the library automation arena is creating a vision of the next round of catalogs. I hear a drumbeat of complaint about the current offerings. While I think the criticism is often overstated, it's clear that the catalogs have lagged behind the interfaces found elsewhere on the Web. They've gained a reputation for being less than intuitive and lacking the features seen in other popular Web sites.
One of the most biting indictments of all is that I hear stories about users who find library OPACs so unfriendly that they often go to places like Amazon.com to look for books of interest, and then flip over to the catalog to see if those titles are owned by their libraries.
It's clear that today's typical library users are Web-sawy and have very high expectations. If we want to draw these folks in, we need to offer interfaces on our Web sites that match or exceed those found on the commercial Web. Paired with high-quality content that's selected and created by librarians, a state-of-the-art Web interface is a compelling destination for users.
Advancing the Interface
We're seeing a tremendous amount of effort being put toward creating new library interfaces, both by the commercial automation vendors and in the open source arena. Each of these endeavors focuses on a different set of assumptions, features, and functions related to the ideal library Web environment, but each of them advances the state of the catalog far beyond what we've seen in the previous round of offerings.
The following are some of the interfaces and products that I've seen over the last few months. They don't necessarily comprise a comprehensive list, but they're some prominent examples of library interfaces that are fairly radical departures from those of the earlier generation:
- Encore (Innovative Interfaces, Inc.)
- Primo (Ex Libris)
- AquaBrowser Library (Medialab Solutions, distributed by The Library Corp.)
- Endeca ProFind (Endeca)
- eXtensible Catalog (an open source project by the University of Rochester's River Campus Libraries)
- SirsiDynix Rooms, SchoolRooms (SirsiDynix)
- Koha and Evergreen (the latest Web OPACs created from open source ILS projects)
- SirsiDynix Enterprise Portal Solution and the latest version of Polaris from Polaris Library Systems (some of the latest Web OPACs from commercial ILS vendors)
I think it's urgent that libraries move quickly to implement more up-to-date interfaces. We just can't afford to offer stale, unattractive, and ineffective catalogs and search tools on our Web sites. I'm afraid that users will drift away to other alternatives unless they find interesting interfaces and compelling content at their libraries' Web sites.
As I look at the latest generation of interfaces, I see a lot of excellent ideas that address these issues. Each has its strengths, but I'm not going to assess each as a whole. Rather, let's think about some of the key concepts, features, and techniques that stand out as essential elements of a successful library search interface for the current era of Websavvy users.
Good Relevance Ranking
One of the most important elements of a successful search interface is the order in which the results display. Google, Yahoo!, and the other global search engines have acclimated us all to the idea of relevance ranking-the best and most interesting items rise up to the top of the list. If everything goes well, even if you get thousands of items in a result set, the best come first.
Good relevance ranking isn't magic. It comes through careful refinement of the formulas that determine the ordering of results. It's very hard to achieve good relevance ranking. When performing a query against a body of information, there will be a number of records that technically qualify as valid results. The tricky part lies in determining which of these result candidates are most likely to be important or interesting. The programs that rank these results will need to draw on a variety of factors, such as the relative position of where the terms occur in each item, the number of occurrences of the search term within an item, or the proximity of multiple search terms to each other. A clever relevancy ranking system will look beyond the items themselves and take into consideration various measures of popularity and interest. How often have other users with similar queries selected each item? How often do external resources that relate to the topic of the query link to each item?
A search interface must have exceptional relevance ranking to be successful. Only the most persistent of today's library users have the patience to pore through page after page of results. This isn't to say that relevance ranking should be the only option available for sorting results. There will be many times when users will want to choose to sort by title, author, or date.
One of the user interface conventions that's being rapidly accepted on the Web is navigation through facets. When generating a response to a query, the system culls out topics, dates, material types, or other selections that are presented to the user as shortcuts to narrowing the search results. By seeing the range of topics represented within the results, for example, a user can click on the one that seems to reflect the intent of the search in order to quickly narrow the results and to filter out unrelated items. A well-constructed faceted navigation scheme will allow users to quickly drill down from a broad set of results to a manageable group of results with just a few clicks.
The search interface selects the facets from metadata that's associated with the items in the body of information. Names, subject headings, publication dates, and other fields from a structured record provide good opportunities for generating facets. There has been considerable interest in Faceted Application of Subject Terminology, an alternative way of applying the Library of Congress Subject Headings. This system has been designed to be more amenable to interfaces that employ faceted navigation. (see www.oclc.org/research/projects/fast.)
Search Result Clustering
Result clustering, like faceted navigation, aims to give the user a fast and easy way to narrow results. While faceted navigation usually relies on metadata terms, clustering operates by analyzing the raw text of the items in the result set to create labels that represent each group or cluster. From the user's perspective, faceted navigation and clustering look much the same. With both, you click through a succession to progressively narrow the results down to a manageable number.
Collections that already have rich metadata, such as library catalogs, lend themselves to faceted navigation. Those with sparse metadata or none at all call for result clustering. Ex Libris and Serials Solutions have licensed Vivisimo Clustering Engine, one of the leading clustering technologies, for use within their federated search products.
Faceted navigation and result clustering both recognize the search paradigm that most Web users currently expect: entering a fairly general term, receiving abundant results, and drilling down to incrementally narrow those results. This approach contrasts against the structured method that previously prevailed, where the searcher had to construct an advanced Boolean-logic query at the beginning of the process.
A Trail of Breadcrumbs
Anytime users navigate through a Web site or search results, it's helpful to provide breadcrumb links that can help them take some steps back toward home when needed. You may recall that in the fairy tale, Hansel and Gretel left breadcrumbs along the trail in the woods to find their way back home. Unfortunately for that pair, the breadcrumbs were gone by the time they were needed. That's not the case with a good Web site. Anew breadcrumb link appears with each step through the site and persists until the user safely exits. Web sites provide breadcrumbs as links almost always near the top of the page and show the user the path he has traveled through the site. Clicking on a breadcrumb link takes the user back to that step.
Breadcrumbs are especially helpful in search environments that use faceted navigation. New breadcrumbs show each facet that's been selected to narrow the search. Since this drill-down approach is often an iterative process, a good presentation of breadcrumbs allows the user to easily back out of an existing facet selection and choose another.
A More Comprehensive, Speedy Search Environment
Many, if not most, users have little understanding of the ways in which librarians organize their collections. We can't necessarily expect them to know that they need to search one interface for the books on our shelves, use another set of interfaces to search for content in electronic publications, and use yet another to search specialized digital collections. It makes sense to us, but it may not make sense to people who have grown accustomed to ever larger and ever more diverse content collections.
One of the major trends in our field is to provide access to more and more content through a single search interface. Federated search and metasearch tools allow librarians to construct search interfaces that include the catalog and large portions of the electronic content to which they subscribe. Some of the new environments, such as Primo and Encore, are paving the way toward providing more immediate access to diverse content collections.
The concept of the library catalog is evolving. As librarians continue to make larger investments in electronic content, it makes some sense to reconsider the primacy of the interface that focuses on print collections. We need more comprehensive search environments that give equal footing to both the digital and print components of our collections. It's becoming less relevant to the user whether the content is print or electronic, local or remote, or whether it's text, video, animation, or audio.
Today's users have little patience for Web interfaces with slow response times. If a site doesn't respond within a few seconds, a large percentage of users will hit the back button and move on to another source. The commercial Web sets high expectations for performance and stability. With today's incredibly fast computer hardware, there's hardly any excuse for sluggish systems. I visit hundreds of library Web sites each week in maintaining the lib-web-cats library directory, and I'm continually frustrated by how long it takes many OPACs to respond. Having all the sexy features won't impress your users if the performance or response isn't snappy.
I've listed just a handful of the features that I think are important for libraries to offer in order to be current with the competing alternatives. Take a look at what's available through your library's Web site. How many of these features do you see? Is your site nextgen or is it a has-been? Don't let stale interfaces be a bottleneck in your library's ability to deliver its high-quality content to users.