As the United States’ oldest federal cultural institution and the largest library in the world, the Library of Congress (LC; loc.gov) has millions of books, recordings, photographs, maps, and manuscripts in its collections. LC also provides leadership to libraries worldwide. As part of its mission to provide services to the global library community, LC information professionals developed Cataloger’s Desktop (“Desktop”), a searchable cataloging, metadata, and library automation documentation system consisting of 300-plus resources (loc.gov/cds/desktop).
Librarians use Desktop in their daily work to find the instructions they need to create metadata to bibliographically control library resources. More than 10,000 librarians at approximately 1,000 subscribing institutions use it. Since its initial release in 1994, Desktop has evolved into a widely used and authoritative web-based service that allows professional catalogers to work more efficiently with the most up-to-date, authoritative cataloging information at their fingertips.
Search Technologies (searchtechnologies.com) provides software development and managed services to support Cataloger’s Desktop. LC and Search Technologies recently redesigned Desktop as part of a system migration project. Critical success factors for this project included the following:
- Engagement with subscribers during design and implementation. Collecting user insights and acting on that feedback was essential to ensuring the updated service met end-user needs.
- Iterative software development allowing for continuous review and testing throughout the project
- Selecting current open source and licensed technologies appropriate to meet LC’s needs for search, content preparation, and the user experience
This article highlights the work involved in implementing the best-in-class Cataloger’s Desktop 4.0, a custom search application.
Overview of Development History
Like many documentation systems, Cataloger’s Desktop began as a collection of print publications from a variety of sources. Each publication used its own editorial practice, and little thought was given to making them easy to use with one another.
Pre-1994: In the early 1990s, online documentation was in its infancy. The Library of Congress began considering how to make it easier for librarians to find answers to technical cataloging questions. Print publications of the time were published in a variety of forms and frequencies. Some publications appeared in interfiled loose-leaf form, while others had multiple supplements that had to be searched independently to answer basic questions. Much time was lost searching and manually updating resources that were limited by the print nature of their distribution. This led LC to explore technologies to allow users to search and navigate among several electronic resources on their desktop computers.
1994–2005: A goal for the new application was to allow librarians to use online documentation resources while working concurrently in an online cataloging application. A Windows-based application appeared to be the best solution. Accordingly, the Library of Congress settled on Folio-Views software for distributing about a dozen resources on CD-ROM. Librarians used Cataloger’s Desktop to search and browse content in the electronic product, which was a radical technology leap forward at the time. (Remember that Windows was new in the mid-1990s and had not achieved the level of acceptance that it enjoys today.)
However, as good as the FolioViews-based Desktop was, it had limitations. The production process limited how frequently resources could be updated. Subscribing institutions received new issues via regular postal mail, which has obvious limitations on timeliness. Subscribers also faced technology and security concerns related to providing networked access to a CD-ROM publication as well.
2004–2014: By the late 1990s, industry experts were predicting the end of the CD-ROM era. This prediction became very apparent as web transmission of information became quite common after the turn of the century. Efforts to move Desktop from its well-established CD-ROM base to the web began in earnest in 2002, and by 2004, Desktop 2.0 appeared.
This release used the FAST ESP search engine with a ProPublish-based user interface, allowing LC to incorporate crawled web resources into a federated search experience. Users also benefited from a wider range of customization options. Moving to the web freed customers’ IT staffs from the need to support Desktop as a networked CD-ROM. The FolioViews CD-ROM product was retired at the end of the 2005 subscription year, and the Desktop 2.0 user interface was replaced with a much improved user interface in 2009; behind the scenes, it was nicknamed “Desktop 3.0.”
For all of its strengths and high user acceptance, FAST ESP and ProPublish were technology dead ends that were difficult to support and too labor-intensive to enhance. Given the proprietary nature of the underlying software, LC began looking for a solution built on largely open source technology.
Present: LC launched the current Desktop 4.0 in September 2014, and it continues to evolve with new features and enhancements. Desktop now includes more than 300 resources in more than a dozen languages. Goals include eliminating dependence on unsupported technologies and providing users with the tools to help them in their daily cataloging tasks.
The Feedback Loop
By 2013, it was apparent that Desktop 3.0’s user interface and underlying technology needed to be replaced. LC staff interviewed several dozen subscribers to help determine how the updated service should function, asking what they liked about the service, what seemed to be missing, and what didn’t work as anticipated. People from both LC and Search Technologies reviewed this user feedback and developed a preliminary plan for addressing user needs.
Throughout the project, team members frequently asked catalogers further questions to get a better understanding of user needs. Although the previous Desktop incarnation wasn’t perfect, subscribers had grown accustomed to using it. A goal for the new system was to provide a much better user experience that would not cause nostalgia for the legacy system.
The development team employed several avenues for collecting user feedback:
Expert user interviews: Power users with diverse job responsibilities were interviewed to develop an understanding of the range of product expectations. The interviews focused on how the application was used, what features were desired, and how the overall system should be improved.
Wireframes: The team developed 25 wireframes for the new user interface that combined the most heavily used Desktop 3.0 features with a backlog of requested features and expert user interview feedback. These wireframes captured the vision and desired functionality for the new product, providing a guide for software development.
Focus groups: The team conducted several subscriber focus group interviews to review the wireframes and obtain their insights. Feedback gathered during these sessions informed software development and design throughout the project.
Polls: The team conducted focused polls of subscribers to gather feedback and to ask questions on specific topics, such as these: What would you consider to be the core resources in the product? Would you like the ability to customize your own resource list? Would you use Desktop on a tablet like the iPad or Surface Pro?
Conference presentations: LC staff conducted live demonstrations of the new Desktop interface at the American Library Association annual meeting, allowing conference attendees to ask questions and provide feedback. This also kept customers engaged in the project.
Main Features and Benefits of Today’s Desktop
The custom search application developed for Desktop reflects how librarians approach locating task-based information. Here are some of the key features that are included:
Searching: Keyword, wildcard, and phrase search; Boolean operators; “did you mean?” spell correction; multi-term synonym support; faceted search; and hit-highlighting. There is extensive support to account for differences in American-, Australian-, British-, and Canadian-English spellings, as well as very specific language support of library science terminology in more than a score of languages.
Browsing contents: Subscribers can browse a table of contents to navigate to specific resources and sections. The table of contents can be browsed by title, library material type, or cataloging activity. (See Figure 1 below.)
Figure 1: Browsing the contents of Cataloger’s Desktop
Viewing a resource: Subscribers can view documents in the application. Search terms are highlighted to provide context. Tables of contents, inter-document links, and next/previous links allow users to easily and intuitively navigate in context. (See Figure 2 below.)
Figure 2: Viewing a document in Cataloger’s Desktop
Subscribers can personalize Desktop:
- My resources: Users can define a personalized resource set that they use most often.
- Internationalization: Users can select one of four languages for their user-interface: English, French, German, or Spanish.
- Bookmarks: Subscribers can save bookmarks for specific documents and share them with colleagues.
- Shortcuts: Users can create shortcuts to be saved in a browser or inserted into an external web document.
- Embedded help: Desktop features context-sensitive help and tool tips that provide instructions and guides to using the application.
Iterative Software Development and Technologies for Desktop 4.0
Search Technologies architects and engineers worked with LC throughout 2014 to develop the new Desktop. Using a “code a little, test a little” approach, Search Technologies provided LC with new versions of the service for evaluation every couple of weeks. Frequent testing and feedback cycles allowed the team to adjust Desktop’s design and development schedule.
Desktop employs the following technologies:
- Solr search engine: The open source Solr search engine (lucene.apache.org/solr) is the core of the new system. Solr indexes content and supports all search, browse, and navigation actions.
- Content processing using Aspire: Desktop includes content in a variety of formats including Folio infobases, SGML, XML, HTML, PDF, MS Office documents, and remote websites. The Aspire Content Processing Framework (searchtechnologies.com/aspire) prepares Desktop content by doing the following:
Harvesting website content
Generating metadata to support Desktop’s search relevancy model
Creating hierarchical and navigation metadata to support browse and document linking
Providing file repository services to support content rendering
- Query Processing Language (QPL): QPL (searchtechnologies.com/query-parsing-language) parses user queries, implements the relevancy model, and supports custom features such as multi-term synonyms.
- Search service layer: A Java-based servlet application provides interfaces for querying Solr using SolrJ, the Java API for Solr, and interacting with the personalization data store (MySQL).
- User experience: The user experience supports search, browse, view, and personalization and is built on HTML 5 and Bootstrap (getbootstrap.com) technologies. The application is responsive, supporting adaptive display through web browsers and tablet computers.
- Hosted in the cloud: Desktop is deployed on Amazon EC2 (aws.amazon.com/ec2), leveraging cloud services for high availability, backup, and extremely fast system response.
Launch of Cataloger’s Desktop 4.0
Desktop was ready by late summer 2014 for its last rounds of testing and a launch in concert with a new content release. LC worked with Desktop customers to ensure a smooth release.
Expert users were recruited from the customer base and given early access to Desktop 4.0. These users provided valuable feedback about adjusting to the revised application and were available as go-to resources for other subscribers.
To further facilitate the release, LC hosted webinars attended by hundreds of subscribers around the world to introduce the new system and address questions. These webinars were recorded and subsequently posted online for subscribers who were unable to attend the webinars due to time-zone challenges. Emails sent to the Desktop discussion list kept customers informed.
Desktop 4.0 launched on Sept. 10, 2014. Very few support questions were received, as customers found the new updated service intuitive, easy-to-use, and requiring little to no training.
BENEFITS TO USERS AND LC AS A PUBLISHER
Additionally, the application supports work tasks and provides multiple, intuitive pathways for information discovery. Desktop meets the needs and information-seeking styles of a wide range of subscribers. It is responsive to multiple screen dimensions; it can be used side-by-side with other applications on a desktop computer or on tablet computers. The improved Desktop allows users to find needed information efficiently, saving time in their work.
Anna M. Ferris, associate professor, University of Colorado–Boulder, a longtime user of Cataloger’s Desktop, told LC, “As an original cataloger, I have witnessed Cataloger’s Desktop’s evolution as the working cataloger’s most indispensable tool. Each development phase has resulted in a state-of-the-art resource that has evolved with the proliferation of resources we must routinely consult. What’s more, the new and improved Desktop interface provides end-users with a remarkably intuitive and user-friendly system.” She added, “As my institution’s Desktop ‘expert user,’ my role as trainer for new catalogers has nearly been superseded by the excellent enhancements made to the training and support modules within the resource. In short, the Cataloger’s Desktop team should be commended for being so forward-looking and for providing us with the tools that help us be more efficient catalogers.”
As a publisher, LC is no longer dependent on unsupported, legacy technologies. The application uses open source and licensed components designed to support services such as Desktop. Using modern technologies allows the team to continue to improve the service to meet user needs. Deploying Desktop on Amazon EC2 results in high availability, reliable system management tools, and an extremely fast response time. Subscribers made the switch to the new user interface with little difficulty and have not looked back.
According to Beacher J. Wiggins, director for acquisitions and bibliographic access, Library of Congress: “To recap the evolution of Cataloger’s Desktop is to realize the enormous strides that have been made to enable greater cataloging efficiency. It is very gratifying to acknowledge this progress and service to LC and the larger library community. Where was Cataloger’s Desktop when I was still cataloging!?”
LC and Search Technologies continue to enhance Cataloger’s Desktop to meet subscribers’ needs, building on the successful migration. Engaging with Desktop users throughout the migration project and incorporating feedback into the design were essential to its success. Iterative software development practices allowed LC to review the application as it was developed. Finally, the choice of current technologies for Desktop 4.0 provides LC with a platform to build on for years to come.
[This article is based upon a presentation given by Bruce Johnson and Derek Rodriguez at the Enterprise Search & Discovery conference held in Washington, D.C., in November 2015 (enterprisesearchanddiscovery.com). –Ed.]