L L E T I N
Portals in Libraries: Library Technology and Planning for Change
Krisellen Maloney is team leader, Library Information Systems Team, University of Arizona, Tucson, Arizona. She can be reached at email@example.com.
Libraries are rapidly expanding the Web-based delivery of content and related access services in order to meet the changing needs and expectations of their users. In the short 10-year period that the Web has existed, libraries have made great advances in their ability to provide Web-based access to a wide variety of information access services that were formerly only available within the walls of the library. Regardless of these advances, many library websites continue to replicate the physical and functional organization of the traditional library. Web-based access to services has evolved as a thin veneer over library technical infrastructures that were designed to support traditional library services. As such, library websites are typically organized around library functions (interlibrary loan, circulation, reference) or existing information stores (the card catalog, print indexes). Web-savvy users who are not familiar with traditional library organization methods do not view our websites as transparent or able to meet their information-seeking requirements.
The common task of finding an article provides a useful example of the special knowledge of library organization and practices that is required to navigate a library website. The process begins with selecting a resource to search. Many users are overwhelmed when faced with deciding which information resource will best suit their current needs and may select a resource for less than optimal reasons. Once an information resource is selected, the user must often master another complex interface to search for appropriate material. Upon identification of an item of interest the user must determine if it is owned or licensed by the library, possibly remember that interlibrary loan is available and in the worst case report an item as missing from the shelf. Each of these steps may require a few mouse clicks into a different area of the library website and, in a typical scenario, reentering similar information multiple times. We might ask ourselves questions such as: How many users are deterred by each barrier or how many users make the optimal selection?
To support the changing user demands within the Web-based service paradigm, technical infrastructures must be made available to users in a manner that supports their tasks. The library portal is one approach to organizing information resources and services in a way that supports the users' needs. However, the library portal will not be the only starting point for access to the library. Other systems, such as course management systems and enterprise portals, may also serve as primary access points for users engaged in a variety of different information gathering tasks. The library portal, along with other application-level interfaces that provide consolidated access to multiple underlying systems, must have integrated connections to every system and information resource.
It is no longer possible to think of library technical infrastructures as a group of separate systems that will be accessed and used from the same starting point and in a similar manner by all users. Successful library systems will need to be designed as a set of core functional components that can be repurposed to suit the requirements of all user-level service needs and made available to a variety of application-level interfaces. The following sections will discuss concepts of the library portal and the underlying technical requirements required to move to a deeper level of integration.
The Library Portal
A portal is a Web-based tool that provides a customizable interface to information aggregated from a variety of sources. Portals are implemented as application-level interfaces and based on software suites that provide integrated access to information resources and related services. Portal implementations vary regarding many technical (application framework, middleware, programming language) and functional (level of customization and personalization) details. The focus of this discussion will be on the different approaches to integration and aggregation of content and related services.
Lorcan Dempsey provides a useful distinction that identifies the end-points of a continuum of technical strategies for integration and aggregation employed within portals. Dempsey identifies two major types of portal technologies: p-portals, which provide integrated access to content at the user-interface level, and m-portals, which integrate information at the programming level.
A well-known example of a p-portal is the MyLibrary portal at North Carolina State University (http://my.lib.ncsu.edu/). P-portals allow users to organize links to searchable collections of frequently used information resources such as citation indexes and publishers' websites. They may also integrate other information resources at the user-interface level by incorporating appropriate display tools such as rss-readers and pdf viewers (for newsfeeds and Adobe's Portable Document Format). P-portals ease some of the burden on the user by providing a place to organize important information resources. The resources may have been selected as useful by the user or by a librarian on behalf of the user, but the integration is done at the surface level. Returning to the example in the introduction, the p-portal facilitates the first step of the user's path to finding an article by reducing the burden of selecting and remembering an appropriate information resource. However, the user must still navigate the remaining steps of the process, such as learning multiple user interfaces and identifying the correct information resource or service provided by the library to find the article.
On the other end of the continuum m-portals offer the promise of simplifying the entire multi-step user process by providing integrated and aggregated access to information at a programming level. Programming-level interfaces may allow many steps of the users' tasks to be automated even when the steps span multiple systems. Metasearching, single-search access to multiple resources, is the most common user-level process currently automated within library m-portal architectures. Instead of learning multiple complex user interfaces, within the metasearching environment, the user can enter a search once that is executed across multiple information resources. This type of access is familiar to users who have grown accustomed to Web searching techniques such as Google. As a result, it provides the promise of reducing the number of interfaces the user must learn, relieving an additional information-seeking burden.
Metasearching provides a useful example regarding the difficulty in developing m-portals in a cost-effective and sustainable manner. The underlying technical infrastructure required to integrate a single task requires that multiple information resources be accessed and searched via the programming-level interface. It would be impossible for each library to develop and maintain a program to access each of their information resource providers. To be successful, libraries and information resource providers must agree on standards that specify how systems can be integrated at this basic level. The NISO (National Information Standards Organization) Z39.50 standard is central to many metasearching architectures found within library portals. It provides a protocol for clients (in this case the portal) to communicate with servers (in this case the information resource provider). The Z39.50 server provided by each information resource provider becomes a programming-level service that can be accessed in a standard and predictable way by any library with legitimate access to the content. While debate exists over the details of the effectiveness of particular aspects of Z39.50, there is no doubt that an architecture based on standard programming-level services will be necessary to move libraries into the future.
In discussing the integration of information and learning environments, McLean and Lynch note "there is a growing acceptance that simply making resources available without an additional level of services may not be very effective" (see For Further Reading). It is not in the best interest of the user to provide multiple information resources without higher-level support for the integration and aggregation of content to support common tasks. It will be necessary to move from surface-level integration at the p-portal end of the integration continuum to the programming-level integration found in m-portal architectures. To do this, it will be necessary to reconceptualize our underlying technical infrastructures to provide these additional programming-level services. This conceptual shift will involve understanding users and the tasks they are trying to complete and identifying the programming-level services, or basic building blocks of function and information, required to complete the task. The first challenge will be to understand the users' tasks in the terms of the users rather than within the framework of our current systems and to reorganize or redevelop our systems to support the users' work. The next challenge will be to agree upon standards so that progress will be sustainable and implementations can be cost-effective.
How do we begin this enormous undertaking? Developing an understanding of changing user demands and the basic building blocks of a new architecture will be a challenge in our current technical environment where systems are organized around data (e.g., the catalog, vendor-based indexes and publishers) or services (e.g., interlibrary loan, circulation and reference). One approach is to design multi-tiered architectures that include an integration layer providing programming-level services for user-level applications such as a portal. These services extract information from existing systems and provide adapters that can be used to develop new applications.
The Shift to Service-Oriented Architectures
The shift to service-oriented architectures that will provide an additional layer of programming-level services is becoming common practice in industries outside libraries as they attempt to save the value of their existing technical infrastructures while repurposing and repackaging their services to meet changing needs. The shift is also evident within libraries and related areas. The National Science Foundation has identified and recommended a cyberinfrastructure that will be necessary to suit the needs of scientists in the future. The cyberinfrastructure specifies a service layer that includes several components relevant for libraries that support scientific research and education. They are calling for the specification of services that will provide access to data, information and knowledge management services. In addition, they are requesting the specification of collaboration services that will support the emerging collaborative processes necessary within the sciences. As mentioned above, McLean and Lynch have made recommendations to consider the definition of services to facilitate the integration of learning and information environments.
Within libraries several national and international projects are underway to specify services. An important example is the NISO Metasearch Initiative that is currently exploring the range of standards and best practices that will be required for single-search access to multiple resources at the application layer. Although library-based metasearching services replicate many of the features of their Web-search counterparts, they exist in a more complex information environment. Issues related to licensing of content and the lack of uniform support for open standards (such as Z39.50) creates a difficult environment for the development and maintenance of single search interfaces. NISO's Metasearch Initiative will develop standards and document current practices that will enable services to be developed to support the needs of information resource providers, libraries and their users. Working groups have been created to explore the issues related to the component parts of the user-level process of single-search access to multiple databases. These working groups include access management, collection description, and search and retrieval. Each working group is collecting information about current practices with a goal to integrating metasearching within existing frameworks. An overview of the work being done by the committees demonstrates the range of issues involved in these specifications of services to support the user-level task of metasearching. It is interesting to see the range of topics that must be explored and defined to solve what appears on the surface to be a single user task.
The access management group will inventory the existing processes and standards that are now in place within libraries and information providers and develop scenarios to describe current practices. A quick overview of standards includes the Circulation Interchange Protocol (NCIP), 3M's Standard Interchange Protocol (SIP) on which NCIP is based, the Lightweight Directory Applications Protocol (LDAP), Shibboleth (an authentication standard) and others, but the most common practice is the use of proprietary interfaces. The goal is to situate access management for metasearching within the current technical infrastructure and to specify best practices for the use of existing standards rather than to develop a new set of standards.
The collection description group is developing element sets to define collection and service descriptions that can be accessed directly by metasearching applications for configuration or user-related purposes. These descriptions are being created in the context of the metasearching applications so that information about collections and services will be framed in a manner appropriate for the searching context. Services provide an interesting problem for this group because many library services are not necessary within the metasearching environment. Because of this, services that may be valuable in general to a library portal may not be described (for example, virtual reference), and additional work will be required to expand these definitions to other contexts.
Although many search and retrieval protocols exist, interoperability remains a problem. The search and retrieval group will explore the existing search protocols and, among other outcomes, recommend necessary element sets for articles and result sets.
Issues for Planning
The library portal and other application-layer services that integrate and aggregate information and services across disparate systems are just the tip of the iceberg. These systems are based on infrastructures that have been designed to support higher-level user tasks. To protect our investments in current systems but at the same time make them accessible to multiple and changing needs we will have to develop an integration layer that provides adapters that can interface with existing systems.
The challenge is to identify user-level services (applications) that are important to users and to identify generalizations of common programming-level services that can be developed to support those applications. At the same time it is necessary to conceptualize library technology as an integrated whole so that we can leverage information from one system for use in many applications. Programming-level services may be used to simplify access for users and when permissible use information about users to make decisions about presentation.
Work is underway, such as the NISO Metasearch Initiative and the NSF cyberinfrastructure effort, to identify the key programming-level services. Establishment of these definitions will take time, but without the presence of robust standards, application integration for library portals, course management systems, enterprise portals and similar projects will become expensive and time consuming.
For Further Reading
Copyright © 2004, American Society for Information Science and Technology