Contact Us | Members Only | Site Map

Association of Research Libraries (ARL®)

  Resources Contact:
Lee Anne George
Publications, Reports, Presentations
Portal Applications in Research Libraries

ARL Directors Forum on Portal Applications Exploratory Discussion Meeting, January 14, 2005

Share Share   Print

Omni Hotel, Boston, Massachusetts
10:00 a.m.–12:00 noon

The Association of Research Libraries (ARL) hosted an invitational meeting of representatives from libraries using different metasearch and portal products. The purposes of the meeting were to conduct an initial environmental scan across a few selected ARL libraries with functioning representative portal/federated search systems and determine if there might be some common interests and concerns that would benefit from being brought to the attention of ARL. The meeting was held on January 14, 2005 just prior to the ALA Midwinter conference held in Boston, Massachusetts..

Attendees included Olivia Madison, Iowa State University, facilitator; Robert Gerrity, Boston College; Kris Maloney, University of Arizona; Michelle Newberry, Florida Center for Library Automation; Andrew Pace, North Carolina State University and NISO Metasearch Initiative; Edward Weissman, Cornell University; and Mary Jackson, Association of Research Libraries.

I. Environmental Scans

Institutional reports, which included descriptions of software tools used, vendor support and role with linking to full-text content, content, and partnerships

A. NISO Metasearch Initiative and North Carolina State. Andrew Pace provided an update on the NISO Metasearch Initiative. There are three task forces working on different aspects of review and the need for standards in this area. The Collection Description Task Group is focused on Dublin Core (with IFLA interest) and will recommend ZREX for a service description. The Access Management Task Group is developing use cases on authentication and access needs. The Search and Retrieval Task Group has had the most difficult and complex issues to work on.

Andrew also provided background on North Carolina State's investigation of portal products. They "started shopping" six months ago and found that all four products they reviewed had drawbacks in different areas, and some in the same areas. All had different pricing models and no product could be seen as out-of-the box.

B. Boston College. Bob Gerrity noted that Boston College (BC) has used MetaLib in a production mode for over two and a half years. BC was seeking a product that would control their e-resources and e-content, so they went beyond simply a metasearch tool. BC is implementing DigiTool and will use MetaLib as a front end to digital content. BC is developing an electronic resource management system. BC has about 350 databases in MetaLib, of which 90 are cross searchable via Z39.50, screen-scraping, or accessible via an XML gateway. The majority (80 of 90) of the databases are accessible via Z39.50.

C. University of Arizona. Kris Maloney stated that the main problem at Arizona, and with several others libraries partnering together in the Scholars Portal (SP) Project, was lack of direct access to commercial or vendor databases. Arizona and other SP participants have access to approximately the same number of databases as Boston College. Arizona uses Innovative Interface's ILS and Serials Solution. They are hoping for a 70% success rate for linking. The project participants are focusing on relevancy ranking, errors and sorting, a more robust open linking product, and a portlet builder.

D. Cornell University. Cornell University implemented ENCompass in summer 2003 and it serves as the underlying system for "Find Articles," Cornell's federated search service, according to Ed Weissman. Cornell uses Link Finder Plus for its resolver and knowledge base. They have configured 35 databases, of which 8 can be searched at one time. The majority of databases are accessible via Z39.50, but others are accessible via XML. Cornell has given up on screen scraping. Ed noted that Cornell does not include the library catalog as an option in its federated search service. Tests showed that the size of the catalog made search results unwieldy and we were concerned about granularity issues. BC separates the results into books and articles.

E. Iowa State University. Olivia Madison provided an overview of Iowa State's "Find It." ISU, a Scholars Portal Project participant, soft launched Find It about two years ago. Find It provides access to about 80 databases with six to seven subject-based sub-profiles. Almost all of their resources are accessible via Z39.50, with a few accessed through a screen scraping system.

Library staff have concerns about the relatively "slow" speed of searching and research results, and Find It's current difficulties in providing open linking non-Z39.50 resources. Catalogers have expressed concern that the catalog may loose its identity through a fully developed federated search system, a view Bob Gerrity noted is also held by BC catalogers. Michelle Newberry noted that the concerns are similar to 70 years ago when librarians were asking why a catalog was needed. Matching expectations is key - both for users and for librarians. If one wants to find out why portal products are not good, just ask a reference librarian. Roy Tennant's talking points document, which he developed for the California Digital Library, is a useful resource for those interested in these areas of concern.

F. Florida Center for Library Automation (FCLA). Michelle Newberry described FCLA's implementation of SFX and MetaLib for its members. FCLA includes eleven universities, and migrated from NOTIS to Olive (ExLibris). Two libraries use MetaLib version 2, with the others waiting for version 3. Each library maintains its own instance of MetaLib and builds a local knowledge base, and FCLA maintains the server. FCLA is looking to MetaLib as a tool to provide a union catalog among their university libraries with interest in creating subsets of regional catalogs.

II. Discussion on State of the Industry, Standards and Partners

Andrew mentioned Ockham, a registry of sources, being developed by Martin Halbert and others. Index Data, a firm that supports open source software, is working with one university to develop connectors that would be made freely available. One estimated a cost $300,000 to $400,000 to produce all of the Z39.50 and screen scraping connectors librarians currently want.

What have libraries learned with their experiences to-date?

A. Regardless of the systems represented, all the representatives expressed and discussed similar concerns and experiences. There is enough commonality that an organization like ARL might consider exploring the influence it could yield upon standards development and best practices.

B. There doesn't exist one standard or method for searching across domains and currently libraries face using a variety of methods, which collectively are problematic.

  • There is still heavy reliance on Z39.50 as the standard to use for searching against structured bibliographic or metadata databases. This standard is not going away soon and is integral to the current generation of federated searching systems. For some libraries, their federated search systems only draw upon Z39.50-enabled resources.

  • Screen scraping technologies, regards of how "primitive" or "elegant" they might be, are being used extensively where Z39.50 is not possible. A central difficulty is the lack of a standard for the citations being searched. There are numerous disciplinary citation practices that require potentially hundreds of technical scripts to written and maintained. One possible standards development could be in the area of disciplinary citation practices explored through partnerships with scholarly societies and commercial publishers.

  • There are a very few instances of direct access to publisher/vendor databases - whether full text or indexing/abstracting - and they are largely represented by agreements between library system vendors and commercial publishers or vendors. There was agreement among the group that ARL should consider expanding the discussions on the issues librarians are confronting in implementing metasearch and portal technologies.

The group strongly recommended that Andrew Pace encourage NISO to continue its work on standards that would support efficient and non-proprietary software development for federated searching systems.

Olivia will report back to the ARL Collections and Access Issues Committee in May. She thanked everyone for attending and sharing their experiences.

Olivia Madison and Mary Jackson, May 24, 2005