Information Retrieval and the ASIS Standards Committee

by Mark Needleman

As many people are aware, the 1994 version of Z39.50, the ANSI/NISO protocol for information retrieval, has been under development by the Z39.50 Implementors Group (ZIG) for several years. This new version adds many features and enhancements to the 1992 version of the protocol while keeping functionality backwardly compatible with it. Most of the new features added to the standard were the result of two factors: the experience gained by implementors building real-world systems using the earlier version and the need to meet requirements of the larger community using Z39.50 for additional functionality to handle the larger set of services those real-world systems support. Among the new services added are a SCAN facility, a SORT facility, an EXPLAIN facility that provides a standardized mechanism within the protocol for clients to learn about the characteristics of servers they are connecting to, and an EXTENDED SERVICES facility that allows the initiation of services within the Z39.50 session that will then be accomplished outside of the protocol itself.

Some of the extended services include requesting document delivery, mailing and printing of result sets, initiating and scheduling periodic queries, and database update. Other functionality that has been added in this new version includes support for larger record sizes and additional record transfer syntaxes.

The new version of Z39.50-1994 was recently sent to NISO's voting members for ballot. Thus, ASIS had the opportunity to comment and vote on the revised version. Below are the comments that the ASIS Standards Committee, acting on behalf of the ASIS membership, sent into NISO on Z39.50-1994.

ASIS Comments on Z39.50-1994

Having seen in draft format the ballot comments entered by the Library of Congress (LC), ASIS wishes to support all of the technical and editorial points made by LC and assumes that since those comments were an attempt to systematically go through the standard and correct both editorial errors and detailed technical errors and confusions, those comments will be incorporated into the final version of the standard.

ASIS also wishes to express serious concern about the complexities and interoperability problems that have been introduced into version 3. These comments echo those made by other voting members. The standard has become unnecessarily complex and confusing, and it is difficult for potential implementors to understand which portions of the document represent real required implementation and which do not.

The document has confused protocol with the data formats used by the protocol. What should be in the standard is the protocol. Many of the things, especially in the appendices, such as data formats, the defined extended services, element specification formats, variant sets and tag sets, do not need to be documented in the base standard. Z39.50 uses a registration model. Those things should be registered outside the standard, and the standard should describe how information on registered formats can be obtained, which it does not currently do.

If the data formats are of sufficiently broad utility, they should be considered for standardization through NISO in its own right, much as Z39.2 has been. MARC is not defined in Z39.50; it is not clear why GRS is. In addition some of the formats, especially GRS and the variant and tag specifications, are not sufficiently mature that they should be registered at all, let alone directly in the base document. At this point, they would be better left at an experimental level until further experience can be obtained with them and there is more of an understanding of the potential interoperability problems they may have. The immaturity of these features is evident in the particularly opaque descriptions of these features and the debates among current implementors of these features themselves about how to interpret material in the standard and use these features.

Another example is the Extended Services facility. The description of the facility does belong in the base document; nor should the definitions of the extended services be included until there is more of an understanding of which of them really represent a core service that will be widely implemented. Due to historical considerations and probably because of its special nature, the BIB-1 and EXPLAIN material is probably an exception to the above comments and needs to remain in the base document.

Evidently, one of the basic design decisions made in developing version 3 was that it be entirely back compatible and interoperable with version 2. The complexity and cost of this decision are painfully evident in the version 3 standard. ASIS suggests that this not be a design assumption for future versions and that efforts be made wherever possible to streamline and simplify things in future versions of the protocol.

Having made the above comments, ASIS also recognizes that a lot of work went into developing version 3 and that many of the new features bring the standard a higher level of functionality than was possible in the 1992 version. In addition the community is waiting for version 3 to be stabilized. But more important perhaps are those in other communities who are evaluating Z39.50 as a potential usable standard who may decide to bypass it for other protocol solutions if they feel that the process by which the standard gets stabilized is too cumbersome and time consuming.

Thus, ASIS, in spite of the above technical comments, reluctantly votes Yes with the hope that many, if not all, of the issues detailed here (and perhaps in other ballot comments) will be addressed either in ancillary documents or in the next version of the standard which is expected to start development once version 3 is approved.

Mark Needleman, chair of the ASIS Standards Committee, works in the University of California Department of Library Automation.

Voting on NISO Z39.14

The National Information Standards Organization (NISO) is currently seeking comments on a draft revision to the American National Standard for Writing Abstracts (ANSI/NISO Z39.14). The proposed revised standard gives guidance to authors or editors of abstracts that represent the content of texts reporting on the results of experimental or investigative work (informative abstracts) or descriptive or discursive studies (indicative or combined indicative-informative abstracts). The guidance is applicable to all conventional forms of abstracts and such unconventional ones as structured abstracts.

The ASIS Standards Committee is currently reviewing the proposed revisions and will cast its ballot on behalf of the ASIS membership before the March 15 deadline. Any ASIS member who would like to provide comments on the basis of a careful review of the proposal is invited to contact Mark Needleman, chair of the ASIS Standards Committee, by e-mail: or through ASIS headquarters.