The CIMI Profile
Z39.50 Application Profile
for
Cultural Heritage Information
Release 1.0
March 1, 1998
Prepared by
Consortium for the Computer
Interchange of Museum Information (CIMI)
CIMI Z39.50 Working Group
William E. Moen
CIMI Z39.50 Project Manager
<wemoen@jove.acs.unt.edu>
School of Library and Information Sciences
University of North Texas
Denton, TX 76201
940-565-3563
This
publication was made possible in part by a grant from the
|
Responsibility
and Acknowledgments
Foreword
Maintenance and
Evolution of the Profile
1. Introduction
2. Background
3.
Scope and Field of Application
4. References
5. Definitions
6.
Z39.50 Specifications
6.1.
Protocol Version
6.2.
Z39.50 Objects Supported
6.3.
Communication Services
6.4.
Z39.50 Services
6.4.1. Init
6.4.2. Search
6.4.2.1.
Attribute Sets
6.4.3.
Retrieval
6.4.3.1.
The Retrieval Record: An Overview
6.4.3.1.1.
Tag Types
6.4.3.1.2.
The CIMI TagSet
6.4.3.2.
The CIMI Schema
6.4.3.2.1.
Constructed Datatypes
6.4.3.2.2.
The CIMI Schema-Specific Abstract Record Structure
6.4.3.3.
The Abstract Record Structure for the Retrieval Record
6.4.3.4.
Element Set Name
6.4.3.4.1.
Element Set Name b
6.4.3.4.2.
Element Set Name mb
6.4.3.4.3.
Element Set Name f
6.4.3.4.4. Guidance for the Retrieval Record
6.4.3.5.
Record Syntaxes
6.4.3.5.1.
Use of GRS-1
6.4.3.6.
Retrieval of Images
6.4.4.
Close
6.5.
Diagnostic Messages
6.6.
Conformance
6.6.1.
Interoperability and USMARC
6.6.2.
Conformance Level 0
6.6.3.
Conformance Level 1
6.6.4.
Conformance Level 2
6.6.5.
Conformance Level 3
6.6.6.
Conformance Level 4
Appendix A -- CIMI-1 Attribute
Set
Appendix B -- CIMI-1 Attributes & Attribute Combinations
Appendix C -- Semantics
for Use Attributes and Schema Elements
Appendix
D -- Dublin Core Mapping to USMARC
Responsibility and Acknowledgments
The CIMI Profile is the result of a two-year effort by the Consortium for the Computer Interchange of Museum Information (CIMI) to investigate the use of ANSI/NISO Z39.50 for use in search and retrieval of cultural heritage information. The National Endowment for the Humanities funded Project CHIO (Cultural Heritage Information Online) in 1995 as a demonstration project for using Z39.50. John Perkins, CIMI executive director, was project director for CHIO. William E. Moen coordinated and managed the CIMI Z39.50 Working Group responsible for developing and testing the specifications included in this Profile. The CIMI Z39.50 Working Group consisted of a broad range of Z39.50 experts, experts in museum systems and museum information resources, software developers, and commercial vendors. See Members of the CIMI Z39.50 Working Group for a complete roster of all participants in the Working Group. The Working Group members' commitment to the vision and possibilities of distributed search and retrieval of cultural heritage information, and their willingness to contribute time and resources to make that vision concrete, made the development of the CIMI Profile possible.
For information about CIMI membership, projects, resources, and activities, visit the CIMI website www.cimi.org or contact:
John Perkins
CIMI Executive Director
<jperkins@fox.nstn.ca> or
<admin@cimi.org>
voice: +1-902-826-2824
fax: +1-902-826-1337
For information about the CIMI Profile, contact:
William E. Moen
CIMI Z39.50 Project Manager
<wemoen@jove.acs.unt.edu>
voice: +1-940-565-3563
fax: +1-940-565-3101
The CIMI Profile is a set of technical specifications for the use of ANSI/NISO Z39.50-1995 (Version 3) in the search and retrieval of museum and other information resources related to cultural heritage information. Cultural heritage broadly defined includes art, architecture, cultural history, and natural history. Z39.50 is a standard computer-to-computer protocol for information retrieval that specifies communications between a client and server for purposes of search and retrieval of information. A profile is a technical document and uses the formal grammar and vocabulary of Z39.50. Because of the technical language of the Profile specifications, the functionality supported by those specifications may not always be clear. The purpose of this Foreword is to provide an overview of that functionality in terms that do not require familiarity with Z39.50.
The Profile can be understood as a set of technical specifications that govern the interaction of clients and servers for information retrieval from one or more distributed repositories. Three basic components of the Profile address searching a database, selecting information to be retrieved from the database, and structuring and packaging of the information to transfer from the server to the client:
The following sections provide additional details about the functionality supported by the Profile for search and retrieval.
Search
A Z39.50 profile specifies the appropriate access points for a given application. The CIMI Profile's application area is cultural heritage information. Therefore, the standard list of access points includes those appropriate for searching museum object record databases, image databases, and bibliographic databases. Searches often consist of a search term and information about that term. For example, a user interested in searching for paintings by Van Gogh will want to express that search in such a way that the system knows to treat the search term "Van Gogh" as "person in the role of artist" and not as "subject of a painting." To allow a user to search multiple databases associated with one or more servers, it is necessary to standardize the expression of the search so that the client and server can communicate unambiguously. This is accomplished by defining an attribute set that identifies a list of access points and additional information used to characterize search terms and express a query in a standard way. The CIMI Profile defines a CIMI-1 Attribute Set (see Appendix A).
The CIMI-1 Attribute Set also reflects emerging agreements within the broader museum community on a set of access points systems should support for searching. By examining existing community standards and existing production systems, and conducting an analysis of questions that users asked of museums, CIMI derived a standard list of access points (in the form of the Use Attributes in the Attribute Set).
The Attribute Set provides the mechanism for the client and server to share a common understanding, or lingua franca, for purposes of searching. When a user submits a search, for example, on provenance information, a server's database may or may not have provenance as a single access point, and it is up to the server implementor to map a search on provenance to the appropriate local database fields or indexes. A server that supports the CIMI Profile can understand when it receives a search for "provenance" because the search is represented and expressed in the standard form of the CIMI-1 Attribute Set.
CIMI prepared a document that illustrates the mapping between the CIMI-1 Use Attributes and corresponding elements and concepts from lists developed by other members of the museum community. For example, the CIMI-1 Use Attribute owner corresponds to the REACH element Current Owner, the AMICO element Ownership, and Spectrum element Lender. This mapping is intended to assist members of the community who "understand" specific labels for their data to see the relationship between their data and the attributes listed in the CIMI-1 Attribute Set.
To address the need for wide-interoperability and cross-domain searching, the CIMI-1 Attribute Set also integrated the concepts represented by the Dublin Core Metadata Element Set. By defining Use attributes associated with those elements, users can express a search in terms of access points represented or characterized by the Dublin Core elements.
Selection
For meaningful and useful retrieval of information from multiple databases, two requirements exist. First, the clients and servers must be able to interchange the database records (or elements from the database records) in formats they both can understand and process. Second, clients and servers need to share an understanding of the elements in those databases and be able to label those elements unambiguously.
A Z39.50 profile defines in a schema a list of elements likely to exist in actual databases. Each database reflects the needs of a local organization in terms of naming practices for database fields and their structure. The schema provides an abstract view of these databases. In this abstract view, database fields are enumerated as schema elements. Each element has a unique name, a unique numeric label, and a definition. A schema also shows the structural organization of those elements in an abstract record structure.
Similar to the CIMI-1 Attribute Set discussed above, the CIMI Schema and associated abstract record structure serve as the lingua franca for communication between the client and server for purposes of retrieval. The CIMI Schema abstractly identifies the units of information that may be found in a database of object records, images with associated text, and cataloging records. The schema does not dictate how a field is named in a database. Instead, it provides a standard way of referencing those elements or fields. For example, the CIMI Schema defines an element dateOfOrigin. A local database might have one or more fields related to the "date an object was created." Since semantics are provided for each of the CIMI schema elements (see Appendix C), an implementor knows that when a client requests the element dateOfOrigin, the unit(s) of information related to "date an object was created" should be returned.
The client can request groups of database fields to be returned. This is done through the Z39.50 convention of Element Set Name. The CIMI Profile defines several element set names (see Section 6.4.3.4.). Each element set name lists the elements that the server should return to the client. The CIMI Profile defines an element set name that includes the pertinent elements to enable a client to create a tombstone view of a database record. The client can also request that the server return the entire database record.
The server, using the standard list of elements defined in the Schema, labels all the units of information retrieved from the database record. Upon receipt of the record, the client can then manipulate and arrange the individual units of information appropriate to users of that client system (e.g., presenting captions in local language).
One important feature of the CIMI Profile is that it specifies how to return images (e.g., digitized photographs, audio clips, etc.). Specifications in the CIMI Profile allow a server to return one or more images associated with an object record. Since a local database may hold the image in more than one resolution (e.g., thumbnail and high-resolution), the CIMI Profile introduces the notion of rendition. A rendition represents a specific version of the image. Therefore, the server can return to the client one or more images as well as one or more renditions of each image. In addition, specific descriptive information can be retrieved for each image and for each resolution. The Profile supports this level of retrieval by defining the CIMI Schema and associated abstract record structure.
Transfer
The CIMI Schema and associated abstract record structure prescribe how the database elements/fields can be labeled unambiguously by the server. Transferring the elements from the server to the client requires one more set of specifications. Z39.50 use the concept of record syntax to address how the server packages up the database elements to return to the client. The record syntax prescribes how the server will format the database elements/fields to transfer to the client. The Z39.50 Generic Record Syntax (GRS-1, see Section 6.4.3.5.) allows the server to handle arbitrarily structured data, and GRS-1 is the record syntax required by the CIMI Profile. Since there is a need to support interoperability between libraries and museums, the CIMI Profile also provides guidance for using USMARC as a record syntax.
The CIMI Schema and the abstract record structure can be used outside of Z39.50. While it is out of scope for this Profile, it is possible to construct and return database records that follow the CIMI Schema in other formats such as Extensible Markup Language (XML).
Summary
While the CIMI Profile reflects a set of specifications for the use of Z39.50 for search and retrieval of cultural heritage information, it provides two important areas of standardization that can be useful outside of the the Z39.50 application environment.
First, the CIMI-1 Attribute Set defines a large set of access points that can be used to express searches. Because this list of access points was derived by empirical investigation and discussion with members of the community, it can be viewed as representing a common set of access points useful in the cultural heritage information environment.
Second, the CIMI schema and abstract record structure provide a standard list of database elements and an organization of those elements for interchanging cultural heritage information. The standard list can be used as a translation device or metalanguage for labeling local database elements and interchanging those elements with other systems.
Z39.50, as a computer-to-computer communications protocol, uses these structures to enable interoperable search and retrieval of information. In the context of the cultural heritage information application, the CIMI Profile specifies how to use the lingua franca of attributes and schema elements for robust information retrieval through Z39.50.
Maintenance and Evolution of the Profile
CIMI has overall responsibility for the maintenance of the Profile. CIMI serves as Editor of the Profile. The CIMI Z39.50 Working Group serves in an advisory capacity to the Editor.
The CIMI Profile will evolve in response to community needs and requirements. The CIMI Profile, Release 1.0, provides a tested set of specifications. Implementors are encouraged to use the specifications defined in this Profile. They may, however, need to extend the specifications as necessary to support local requirements or to support functionality not yet addressed by the Profile. Implementors should be aware that private extensions may threaten interoperability with other implementations.
CIMI encourages experimentation with and extensions to the Profile, but implementors should use the existing specifications to their fullest extent and in the way they were intended (e.g., mapping pre-existing databases for Z39.50 access according to Profile specifications). For example, string tags (see Section 3.4.3.1.1.) may be used to label locally-defined fields in a retrieval record using GRS-1, but to encourage interoperability implementors should first exhaust the elements in the CIMI Schema before resorting to such measures. However, implementors can suggest to CIMI that commonly occurring database elements/fields need to be defined in the Schema so that string tags will not be necessary. Identifying and proposing such requirements can be accommodated by the procedures for maintaining the Profile.
Subsequent to the publication of the CIMI Profile Release 1.0, CIMI will develop a set of procedures for ongoing maintenance and management of the Profile. These will include:
These procedures provide a formal mechanism that CIMI and others can use for systematic and consistent management of the evolution of the CIMI Profile.