Bob Buckwalter, Marianne Burke , Rod Goins, Ted Pappadopoulos, Jon Rothman, Lynne Schmelz (chair), Hinda Sklar, Dorothy Solbrig, Judy Warnement, Robin Wendler (recording)
Absent: Bob Buckwalter, Michael Fitzgerald, Carrie Kent, Kathy Klemperer, Eva Moseley, Mark Van Baalen
Guests: Dale Flecker, MacKenzie Smith
April 19, 10:00-12:00 OIS Conference Room May 17, 10:00-12:00 OIS Conference Room June 21, 10:00-12:00 OIS Conference Room
Dale Flecker gave a presentation originally given to the Automation Planning Committee proposing a project to implement Z39.50 on a non-HOLLIS server. APC accepted the proposal and the project is now in the planning stages.
Dale started by explaining that there are two ways to provide campus-wide access to library data now. Mounting a database in HOLLIS provides wide access, a common interface, and local holdings information. Giving access to a remote database via HOLLIS Plus is much easier and cheaper, more flexible, and offers a wider choice of databases. Accessing data outside of HOLLIS is our long term preference, so we need to find ways to recreate the benefits of HOLLIS for non-HOLLIS databases.
1) The need for wide access is being solved by the growth of the network. As more and more libraries and users are getting network connections, providing access to remote resources is becoming a reasonable solution.
2) The benefit of a common interface can to some extent be gained through the use of Z39.50. In this client/server search-and-retrieval protocol, the client determines the interface. Harvard could mount a common client behind HOLLIS Plus through which users could access a variety of servers, such as those provided by OCLC, RLIN, CD-Plus, and Silverplatter, with a single public view.
3) Local holdings can be provided for use with remote indexing and abstracting databases by extracting copies of holdings from HOLLIS and adding them to a local Z39.50 server, which can be searched from the Z39.50 common client.
Ted Pappadopoulos asked whether users could buy their own Z39.50 clients. Dale explained that while such clients do exist, there is significant tuning required to make a particular client work with a particular server, and there would also be no way to handle security in that scenario.
HOLLIS II will certainly support Z39.50,and so will fit neatly into this picture. Down the line, we can expect the further enhancement of Web access, using gateways between http and Z39.50 protocols. This combination allows the presentation of a more modern GUI interface without the library having to write and distribute Web client software.
The APC approved the proposal that OIS acquire a search engine and a common Z39.50 client, explore both terminal and Web interfaces, and pursue adding local holdings access to the client. Other possible uses for a new search engine include providing a copy of HU to be used as a replacement for the Distributed Union Catalog in the event of short-to-mid term HOLLIS outages, access to finding aids, non-bibliographic databases (e.g. full text and images), and possibly as a better home for some databases currently in HOLLIS.
Judy Warnement asked whether some functionality would be lost in the move to Z39.50. Dale acknowledged that Z39.50 is to some extent a least-common-denominator solution. Few clients and servers can support the entire range of possible searches and features. We would likely run a common proprietary client for some databases with important special features.
The Seach Engine Evaluation Committee (SEEC) is expected to select a search engine by summer, and OIS hopes to have the platform up and running this calendar year.
OIS last asked library staff to prioritize outstanding HOLLIS enhancement requests 15 months ago. Many of the selected enhancements have been completed, and OIS is ready to launch the next round of prioritization. HAAC agreed that in order to reduce ongoing investment in HOLLIS, OIS should ask for fewer enhancements per category than last time.
Jon Rothman reported the results of increasing the maximum hits allowed in keyword searches. The increase from a maximum of 250 to 1000 went in on February 27. In looking at data before and after the change, OIS did not see a huge impact on resource utilization, and approximately 50% of the searches which would have hit the maximum before were now able to complete successfully. Jon did discover that we cannot raise the limit higher than 1000 without encountering ABENDs. The problem has been reported to NOTIS (this problem is occurring in the "black- box" of NOTIS keyword software to which OIS does not have access). If NOTIS can not resolve this problem in a few weeks, 1000 will be the maximum number of keyword hits we can support, and we will announce that new limit to the public.
Keyword qualification poses a much different problem. Tests show that it would have a big impact, but exactly how big can't be determined without real use, which would mean not only adding the capability to HOLLIS, but also educating users about its availability. If we do this kind of test, we may find that the impact is too great and have to take qualification back out. In any case, we would have to disallow "system-busters" such as language=English or format=books. Date qualification is also problematic. SSHUSH has suggested indexing each record by exact year and by decade, which would allow some flexibility, but not true date ranging. Keyword qualification would not include faculty or location. HAAC unanimously agreed that OIS should go ahead and try qualification, with the understanding that it may have to be removed. It also instructed OIS to work with qualification to gauge its supportability before investing time in developing a qualification prompt screen.
OIS asked for guidance on whether to produce the next Distributed Union Catalog microfiche supplement. HAAC asked that the supplement be deferred for now, and that the topic be revisited in the fall when the timeline for the new search engine (and possible online DUC) is more fully known.
MacKenzie Smith presented a project plan to follow up on some of the recommendations of the Task Force on Special Collections. One of the recommendation of the Task Force was to provide central support for the editing and searching of SGML-encoded finding aids. The pilot project will have two phases.
Phase 1: Creating Digital Finding Aids.
Form a task force which will identify finding aids to be included (on the order of a few hundred), select an SGML authoring system, create digitized versions of the finding aids (scanned or keyed), train staff, and do SGML encoding using the Berkeley-defined SGML Document Type Definition (DTD) for finding aids.
Phase 2: Making Digital Finding Aids Available.
The task force will identify relationships between kinds of archival information as defined in the original report. OIS will do technical work to bring up the new searching system, provide tools and procedures, load the finding aids for indexing, and design the display.
Search and launch capability has been added in the test system, but there difficulty of providing adequate keywords for general coverage databases means that search results are apparently random. Search and launch will not be added to the production system until these problems are resolved.
Some Law School Library finding aids will be added to HOLLIS Plus.
OIS will survey libraries on CARL usage, because the failure rate is particularly high and could be reduced by buying dedicated access. This access is fairly expensive, though, so we need to make sure that the need can't be met in other ways.
HDR now has approximately 20 users, and is still being billed at $30/account. OIS hopes to put a users manual online.
Diane Garner is profiling Harvard's depository arrangements with two vendors, to receive shipping list records for check-in and corresponding cataloging records. OIS has begun programming and is aiming for late May.
The programming is done for adding new fields to item records. These changes will not affect most people's work unless they are using macros. The addition of the new fields will cause the screen positions of other fields to change, so any item record macros may have to be modified. SSHUSH will report to HAAC soon about the possibility of adding user-initiated recalls and/or holds to the FETCHERRS project.
OIS implemented SSHUSH's recommended changes to the options area to make STORE/SEND an explicit option, and the use of these commands has increased. The COMMENT command was also added to the options area and its use has doubled. Staff and users are quite pleased with these changes.