Draft 2/17/98
NEH Final Report
Project CHIO:
CIMI's Exploration of Z39.50 for
Distributed Search and Retrieval of Cultural Heritage Information
1. Introduction
One of the world's great treasures is the vast repositories of information that document and reflect humanity's cultural development and accomplishments. Included in these repositories is a wide variety of data, information, objects, and publications in multiple formats. These repositories are spread over the globe and many are associated with museums, libraries, and archives. The networked environment comprising digital data networks, computers, databases, and information resources provides a new context in which these organizations can make their information available to many groups of users. Further, the technologies raise expectations among users that they will be able to identify, search, and retrieve information from these repositories without regard to geographic proximity, hours of operation, and ownership.
Project CHIO has been a successful two-year effort to demonstrate the proof-of-concept of distributed search and retrieval of cultural heritage information. In the process of this demonstration, CIMI led an international collaborative process that identified stakeholders and their requirements, selected a set of these requirements for which technical specifications could be developed, motivated the adoption of a standards-based approach for information search and retrieval, and seeded technology in the cultural heritage information community as a foundation for achieving a goal of distributed, interoperable search and retrieval.
This report summarizes this two-year effort by identifying the activities undertaken, processes used, participants involved, outcomes achieved, problems solved or clarified, new problems identified, and the products developed. This document serves as the final report to the National Endowment for the Humanities whose generous funding of Project CHIO enabled CIMI and the broader cultural heritage information community to take important steps towards the ultimate goals of serving users world wide to access information.
2. Project CHIO--Goal and Objectives
Project CHIO's overall goal was to demonstrate how the use of standards facilitated online access to information held in textual databases and imagebases independent of the hardware and software used to store the data or search for it. Specifically, the Project proposal identified three tangible products that would result:
Projects such as CHIO that address complex and sometimes ill-defined problems face particular challenges. When CIMI submitted the proposal to NEH (November 1, 1994), the evolving networked environment in which CHIO activities occurred looked much different than today. For example, the World Wide Web (WWW) has affected major changes in how museums and others viewed the opportunities of the networked environment since that time. Innovations such as JAVA and the efforts for standardizing metadata for resource description and discovery have taken center stage in discussions relating to the networked environment. Through this very dynamic period, CIMI has pushed the envelop of integrated access to distributed resources by its efforts in Project CHIO. As a consequence, the shape and focus of CIMI's efforts in the past three years has evolved to address emerging issues, problems, and potential solutions for providing meaningful networked access to cultural heritage information.
The tangible results of Project CHIO were:
Less-tangible but just as significant results from this experience have been the increased visibility of a standards-based framework for searching and retrieving cultural heritage information. This heightened awareness was a product of CIMI's outreach and dissemination of Project CHIO's developments and outcomes by presentations at conferences and publications in journal's such as the Museum Computer Network's SPECTRA and the Association for Computing Machinery's Communications of the ACM.
3. Activities to Accomplish Project Objectives
Project CHIO included three major thrusts to address the original goal and objectives. The first thrust was to deploy the CHIO demonstrator that allowed a single Web-based interface to disparate resources. The second thrust was the exploration and specification of ANSI/NISO Z39.50, the American National Standard protocol for information retrieval, and resulted in the CIMI Profile. The third thrust was actual implementations of the CIMI Profile in an interoperability testbed as proof-of-concept for Z39.50 specifications to support distributed search and retrieval. The second and third thrust intertwined as the interoperability testbed provided implementation experience that informed major revisions to the CIMI Profile.
3.1. The Project CHIO Demonstrator
[John to provide details]
3.2. The CIMI Profile: Specifying the Use of Z39.50
A major focus of Project CHIO was to explore the use of Z39.50 in applications for search and retrieval of cultural heritage information. Z39.50 is represented as an American National Standard, Information Retrieval (Z39.50): Application Service Definition and Protocol Specification, and also as an international standard, ISO 23950. Project CHIO's objective was to specify how Z39.50 could be used to support interoperable search and retrieval transactions. To accomplish this specification, CIMI produced the CIMI Z39.50 Application Profile for Cultural Heritage Information. An application profile is an ancillary standards mechanism that specifies the use of one or more standards to achieve a set of functions and provides guidance in applying the standards interoperably in a specific limited context.The CIMI Profile details specifications for the use of Z39.50 and other standards by museums and other organizations offering cultural heritage information.
The profiling effort included the following groups of activities:
This section reports on the people, processes, and products of the profiling work.
3.2.1. Participants Involved in Developing the CIMI Profile
CIMI established a Z39.50 Working Group comprising experts in museum information systems and Z39.50, software developers, and commercial vendors to explore and specify how Z39.50 would be used. An important consideration in developing community-based standards is to ensure that stakeholder organizations have input into deliberations. Z39.50 Working Group membership changed over the years of the project, and Attachment A identifies over 35 individuals who attended two or more working group meetings. Some of the individuals were involved primarily in the first year of the profiling effort; while others participated throughout the project; still others were involved as part of the CIMI Z39.50 Interoperability Testbed (see below). It is important to highlight the range of organizations represented and the internationalization of CIMI's Z39.50 work.
3.2.2. Developing the CIMI Profile
The development of the CIMI Profile occurred in two major phases. The first phase covered the period of September 1995 through completion of a draft CIMI Profile in June 1996 (see Attachment C). The second phase covered the period of July 1996 through February 1998. This second phase overlapped with the interoperability testbed (March 1997 through February 1998), and the final revised CIMI Profile, Release 1.0 (see Attachment D) reflected the experience gained through the testbed.
CIMI Profile development was an exercise in defining problems and issues and building consensus within the Z39.50 Working Group on appropriate solutions or approaches. To achieve meaningful distributed search and retrieval, the CIMI Z39.50 Working Group addressed three basic problems:
The resulting application profile specifications provide a workable solution to these problems.
The bulk of the work occurred in meetings of the Z39.50 Working Group (see Attachment B), but the work of the group was supported by online discussions through the Working Group's listserv. Electronic mail, electronic distribution of documents, and the listserv assisted in moving the work forward between meetings. After each meeting, the Z39.50 project manager produced a detailed summary of discussions, agreements, open issues, and other information that was distributed to the Working Group members and served as a record of activities of the Working Group.
During the first phase (September 1995 through June 1996), the Working Group met four times. In a profiling effort, the initial focus is to identify requirements for the application, and then move to specify how Z39.50 or other standards can support those requirements. The result of these meetings was a draft CIMI Profile released for public comment in June 1996.
A primary early requirement was that the use of Z39.50 would need to support the search and retrieval of data cast in the Standard Generalized Markup Language (SGML) Document Type Definition (DTD). (The DTD was being developed by another, separately managed and funded, aspect of Project CHIO.) The assumption for the Z39.50 Profile was that SGML data (e.g., representing full text exhibition catalogues, wall texts, etc.) would be available for Z39.50 search and retrieval. To a certain extent, the progress on the CIMI Profile was dependent on stabilized SGML specifications (in the form of the DTD). For example, the DTD built upon CIMI's work to define a set of CHIO Access Points. Although relatively stable, the CHIO Access Points as implemented in the DTD continued to be refined and modified. A more important aspect of the requirement to address SGML data in the CIMI Profile was the modeling of complex query and retrieval structures that Z39.50 would need to support.
Another aspect of the CIMI profiling work was an early decision to align the CIMI Profile as a companion profile to the Z39.50 Profile for Access to Digital Collections. The Digital Collections Profile development work had started shortly before the CIMI profiling began. It, too, was neither stable nor tested in practice.
Beginning with the January 1996 meeting, the Z39.50 project manager developed drafts of the CIMI Profile for discussions at each meeting of the Working Group. The group discussions informed subsequent redrafts of the Profile. The process was one of gaining agreement on the functionality that needed to be supported by the Profile, clarifying the problems and issues related to the functionality, and then determining exact Z39.50 specifications to include in the Profile. This was an iterative process lasting from September 1995 through May 1996 when the Z39.50 project manager produced a stable draft CIMI Profile for final comment by the Working Group.
The first year of profiling effort did produce a draft set of specifications, a notable accomplishment. The specifications in that draft, however, can be viewed as the result of committee discussions (theoretical and abstract) without necessary implementation experience to check either the practicality or utility of the specifications. Ironically, at the early meetings the CIMI Z39.50 Working Group had agreed to two basic principles for developing the Profile:
The draft CIMI Profile contained specifications that moved beyond these two principles, and ended up pushing the envelope of both technology and available data.
The second phase of profiling work began subsequent to the June 1996 release of the draft CIMI Profile. To accomplish the objectives of Project CHIO, CIMI needed to ensure that, following the release of the draft Profile, organizations would begin implementation. CIMI called a meeting for September 1996 to discuss implementation strategies and follow-up profiling efforts. The Z39.50 implementor who participated in the CHIO Demonstrator reported at that meeting his concern with the implementability of the draft Profile specifications. Another issue that needed to be addressed was the fact that the SGML data from the first component of Project CHIO had not been created to the extent that expected.
Project CHIO Z39.50 staff spent the period of September 1996 through January 1997 strategizing how best to move forward to implementations. Implementation experience was a key factor to revise and refine the CIMI Profile. The result of these discussions was a background paper developed for a February 1997 Working Group meeting (see Attachment E). The document outlined possible redirection of the CIMI Z39.50 effort so that implementations could proceed quickly. The attendees at that meeting endorsed a plan whereby a subset of the draft CIMI Profile would be used to guide near-term implementation for interoperability testing and to focus on a limited set of data including museum object records, images, and bibliographic records. The result of this meeting was the decision conduct the CIMI Z39.50 Interoperability Testbed. The testbed provided the implementation experience to revise the CIMI Profile to reflect a tested set of specifications for distributed search and retrieval of cultural heritage information. Section 3.3 discusses in detail the processes and products of the testbedl.
At the conclusion of the testbed in October 1996, and based on its results and experience, the CIMI Profile underwent a major revision. An initial revised draft was circulated prior to the January 1998 Working Group meeting. Discussions at that meeting informed a revised draft issued on February 1, 1998 for final comment by the Working Group. The CIMI Profile: Z39.50 Application Profile for Cultural Heritage Information, Release 1.0, is the final outcome of the two and a half year profiling effort.
3.3. CIMI's Z39.50 Interoperability Testbed: International Testing of Z39.50 for Distributed Search and Retrieval of Cultural Heritage Information
Subsequent to the February 1997 meeting, CIMI Z39.50 staff determined that a CIMI-sponsored Z39.50 interoperability testbed could provide a useful vehicle for involving a broad range of potential stakeholders in implementing the CIMI Profile. A Call for Participation for the testbed resulted in over 40 responses (see Attachment F). A follow-up questionnaire solicited detailed information about participants' capabilities to carry out successfully the requirements of the testbed. CIMI selected six organizations (two of which submitted a joint application) and two alternates to participate. In addition, a number of CIMI member organizations participated in the testbed. Attachment G briefly describes the testbed and lists the participants. The involvement of international participants in the testbed was very important in creating international visibility for CIMI's Project CHIO and its associated Z39.50 work.
The organizational meeting of testbed participants occurred in May. The Z39.50 project manager developed a draft implementors agreement for discussion. The strategy resulting from the February 1997 meeting was to use a subset of specifications from the June 1996 draft CIMI Profile for the testbed. The implementors agreement comprised that subset. The implementors agreement went through numerous drafts over the course of the testbed and served as the guiding specifications for implementations deployed during the testbed.
CIMI contracted with two software developers to build Z39.50 tools for use in the testbed. The strategy was to jump-start implementations of the CIMI specifications by lowering the barriers to participation. System Simulation developed a CIMI Z39.50 server toolkit for that purpose. Blue Angel Technologies developed a Z39.50 client for use in the interoperability testing.
By the end of the interoperability testbed in October 1997, the participants had implemented eight Z39.50 servers and three Z39.50 clients that supported CIMI specifications for search and retrieval of cultural heritage information. Interoperability testing between the clients and servers in the last month of the testbed demonstrated that the implementors agreement's specifications enabled distributed searching and retrieval of databases of museum object records, images, and bibliographic records. The specifications in the implementors agreement served as the primary basis for the revision of the June 1996 draft CIMI Profile, and the resulting CIMI Profile: Z39.50 Application Profile for Cultural Heritage Information, Release 1.0, reflects a set of specifications tested in large part through the interoperability testbed.
As part of the reporting requirements in the testbed, participants each developed final reports that summarized their experience in the testbed. Attachment H is a summary of the participants' reports. Attachment I is the full text of the reports. One item to highlight is that the implementors each identified additional work on the CIMI Profile they would be interested in pursuing in an ongoing CIMI interoperability testbed.
The CIMI Z39.50 Interoperability Testbed reached important first milestones for distributed search and retrieval. As proof of concept, CIMI demonstrated Z39.50's utility in the networked cultural heritage information environment. As a collaborative international effort, the testbed brought together implementors interested in sharing experiences and expertise related to Z39.50 and access to digital information. As a testbed, the six-month experience raised new and exciting questions and problems related to the complexity of interoperability and distributed search and retrieval. Attachment J contains a draft article submitted and accepted for publication in Communications of the ACM that provides an overview of the interoperability testbed.
3.4. Summary of Project Activities
Project CHIO evolved during the two and a half years of effort. Reviewing the November 1994 proposal to NEH, one sees the seeds of all that was actually accomplished during the project. The original focus on a "resource of approximately 10,000 records on the them of folk art" gave way to the challenge of distributed search and retrieval across databases and servers spread around the world. Project CHIO reflected--as well as participated in--the dynamically evolving networked environment of 1995-1998, a period marked by dramatic changes in technologies, user expectations, opportunities, and challenges.
As a result of the effort, Project CHIO has produced the following:
The effects of this project are being seen already. Implementors are committed to evolving the CIMI Profile to provide additional functionality. Organizations have referenced the CIMI Profile in Request For Proposals or are considering its use in large applications for distributed search and retrieval. The project has enabled CIMI to work effectively in several standards arenas, notably, the Z39.50 Implementors Group and the Dublin Core Metadata effort. CIMI will leverage the experience gained through Project CHIO to promote a standards-based approach to solving problems related to access of the rich cultural heritage information stored in museums and libraries around the world. The work, however, is far from complete, and the following sections identify challenges and issues that Project CHIO clarified.
4. Key Issues and Lessons Learned
The original Project CHIO proposal to NEH outlined an ambitious undertaking. In many ways, Project CHIO was an expedition into a digital landscape where there were few maps or familiar landmarks to guide the work. Through Project CHIO, CIMI blazed new trails in this digital landscape. In the process, CIMI discovered problems that had no right solution, took some paths that were less-than-fruitful, and developed approaches to distributed search and retrieval that demonstrated the utility of Z39.50 for access to a range of cultural heritage information. This section highlights selected issues and lessons learned.
4.1. Incremental Solutions to Complex Requirements
In the early stages of Project CHIO, the Working Group identified and discussed a set of complex requirements for functionality in distributed search and retrieval of cultural heritage information. The first nine months of activities responded to these requirements and specified the use of Z39.50 to support the requirements. Yet, a number of assumptions on which the specifications were based turned out to be problematical. First, there was an assumption that a significant amount of richly encoded SGML data would be available for implementations to use. Second, there was an assumption that existing search engines and database management systems could accommodate the sophisticated searching of that SGML data. In both cases, the Working Group's assumptions did not reflect existing reality. The June 1996 draft CIMI Profile emerged from a "committee room/theoretical" approach. That effort was useful and necessary to gain understanding and agreement on the shape and texture of many of the problems. Although the Working Group modeled the search and retrieval of complex data using Z39.50, the resulting specifications pushed beyond currently available technology and data.
It was that situation that shaped CIMI's decision to scale back the Z39.50 specifications to address existing data and systems, and pursue a more incremental approach to the complex requirements and challenges. The success of the interoperability testbed as measured in interoperable implementations suggests that project that try to solve Grand Challenges should do so by incrementally solving smaller pieces of the Grand Challenge. As a result, the importance of constantly revisiting the scope of a project is essential, and identifying the appropriate scope is itself a challenge.
4.2. Evolutionary Response to Dynamic Environment
As noted above, the networked environment during Project CHIO was volatile and dynamic. By the end of the Project, technologies that were merely nascent at the Project's early stages had been widely deployed. In 1994, the Web was emerging as the primary interface for Internet-based applications. By 1998, platform independent software such as JAVA had been deployed. The integration of Z39.50 and the Web became a focal activity. The Dublin Core Metadata initiative began in Spring 1995 and momentum built around that solution for networked information discovery and retrieval. To be successful, Project CHIO could not focus simply on the requirements outlined in the proposal but had to address this constantly changing environment.
What is striking about Project CHIO has been its ability to be open to the changes and incorporate new approaches that served the goals of the Project. Projects such as CHIO need to remain agile and flexible, all the while keeping on track towards achievable goals and objectives. Thus, another lesson learned through this Project is that one should assume an evolutionary perspective. The Working Group's understanding evolved over time. Sometimes it would take several meetings before the specific problems the Working Group was trying to solve could even be defined.
4.3. Refining the Scope
Project CHIO's overall goal was to demonstrate how the use of standards facilitated online access to information held in textual databases and imagebases. One approach outlined in the proposal was to create a resource of records, but that was based on the assumption that records would be created as part of another project separate from Project CHIO. A modification of project activities was to focus on existing data rather than assuming new data would be created. The second year of the project was guided by priority need for implementing Z39.50, and it was inappropriate to assume a broad range of new data would be created and available for the implementations.
A lesson learned from this experience is to keep in mind the overall goal of a project, be flexible in interpreting how best to achieve the goal, and continually refine the scope to gain maximum results within the constraints of resources and time of participants. As a result, CIMI continued to refine what was most important to accomplish, and that turned out to be captured in the phrase "integrated access to distributed cultural heritage information."
4.4. Limitations of Time
Because CIMI wanted to demonstrate the utility of a standards-based approach, the interoperability testing component was a paramount consideration. Yet, to get to that phase of the Project, the Working Group had to first agree on a set of specifications that could guide implementations. Social processes that require building consensus on complex technical and operational solutions require time. There was an early expectation that once the Working Group agreed to a set of specifications, implementations would quickly become available. The actual testbed phase suggested that the project planning did not anticipate the time and resources required to get implementations up and running, to get data prepared, and to carry out interoperability testing. CIMI requested two no-cost extensions for Project CHIO to accommodate the time requirements for these activities.
In the future, it will be important to identify incremental steps that are achievable in the time frame and resources allocated. The dedication and commitment of the CIMI Interoperability Testbed participants enabled the overall success of Project CHIO. Yet, as witnessed in the participants' reports on their experience with the testbed, allowing adequate time for testing was a primary issue. Future projects should acknowledge the time and resource constraints of participants and scope the project activities to accomplish objectives and goals.
5. Successes and Accomplishments
Projects such as CHIO should be viewed as evolutionary in that they need to be responsive to the changing environment in which they are being carried out. The original project proposal guided the work and the successes of Project CHIO can be tied directly back to the goal and objectives outlined in the proposal. In addition, the Project had successes and accomplishments that went beyond the expectations of project planners. This section highlights selected some of those.
5.1. Leadership and Collaboration
Through Project CHIO, CIMI demonstrated its ability to involve numerous stakeholders in conversations about the utility of a standards-based approach to solving problems related to access and interchange of museum information. CIMI has focused on developing community consensus on standards and their use. In the networked environment, these agreements are essential to achieve goals of distributed and integrated access to a wide range of information no matter where it is located. Project CHIO provided the impetus for bringing these different stakeholders to the table and discussing appropriate approaches for solving complex problems.
In the networked environment, and especially in the museum community, no organization is in a position to dictate adherence to a standard. CIMI's approach has been to nurture an open process in which interested parties can collaborate and build consensus. The outcomes of Project CHIO speaks positively to CIMI's approach. In particular, the CIMI Interoperability Testbed provided an environment where organizations and individuals could develop new partnerships and working relationships that will extend far beyond the end of the project.
5.2. Process as a Basis for the Product
The process for carrying out Project CHIO was collaborative from the outset. As noted in the attendance at Working Group meetings, many individuals representing a diversity of organizations participated. CIMI's working assumption has been that standards are agreements among people to do certain things in certain ways to achieve predictable outcomes. Stakeholder communities need to be involved in consensus building around a standards effort if the resulting approach is to be acceptable and accepted.
Project CHIO enthused participants from around the world, and through the Working Group's open process for deliberations, CIMI provided a forum for developing consensus on the use of Z39.50 for distributed retrieval. Yet, as a social process with many stakeholders, the Project CHIO work was necessarily slow as participants moved through stages of problem identification, problem definition, and problem solution. As noted above, leading edge efforts such as Project CHIO do not have the luxury that they will define the right or perfect solution. In the case of the interoperability testbed, Project CHIO participants defined workable solutions to problems such as searching distributed databases and retrieving multiple images. Through the testbed experience, Working Group members learned to make decisions about what could be achieved with limited resources and time.
CIMI used a number of mechanisms in addition to regular Working Group meetings to support the deliberative process. These included an online discussion listserv, electronic mail, and a website. All of these provided ways to disseminate information, gather input, and move forward on the Project between face-to-face meetings.
5.3. International Influence
The networked environment is global, and appropriately, Project CHIO is influencing projects and activities in the international arena. As an example, the European Project Aquarelle adopted the CIMI Profile as a basis for its work in distributed search and retrieval of cultural heritage information. Participants from Aquarelle attended Working Group meetings and also participated in CIMI's interoperability testbed. Project CHIO's collaboration with the Aquarelle Project was essential to ensure common approaches and standards to realize common goals. As discussed below in Section 6, CIMI plans for ongoing collaboration with Aquarelle participants in 1998 to arrive at a harmonization of Aquarelle and CIMI Profiles.
Another important influence of Project CHIO outcomes can be seen in the Request for Proposal that reference the CIMI Profile. To date, CIMI is aware of at least two projects that have referenced the Profile.
5.4. Identification and Clarification of Problems
Project CHIO provided an opportunity to explore issues and problems in providing meaningful access to cultural heritage information resources. The CIMI Profile resulting from this work indicates a set of solutions that address the original challenges:
The CIMI-1 Attribute Set provides a way for Z39.50 clients to express search queries to multiple servers and their associated databases. This is a basis for semantic interoperability across distributed and heterogeneous information resources. In addition the Profile outlines a standard way to identify elements of database records for retrieval purposes. The Interoperability Testbed demonstrated the utility of these specifications.
Another set of project outcomes, however, emphasized a clarification of problems rather than their solution. This is in keeping with the understanding that initiatives such as Project CHIO are by necessity evolutionary: finding a solution to one problem highlights and clarifies others. For example, the CIMI-1 Attribute Set provides a solid basis for representing appropriate access points, yet individual implementors may not have search engines that support one or more of those access points. Further, given the heterogeneous nature of the resources, all databases will likely not contain the same data available for searching. Again, the practical experience of the interoperability testbed was valuable for both solving some problems and clarifying others.
5.5. Interaction with Various Standards Efforts
The Working Group for Project CHIO included a number of Z39.50 experts. From the beginning, CIMI assumed that the experience from Project CHIO could feed into the standards work of the Z39.50 Implementors Group (ZIG) as well as other standards efforts. Several of the Working Group meetings were held conjunction with the ZIG, and the presence of Project CHIO participants at the ZIG raised the visibility of CIMI's Z39.50 efforts within the larger Z39.50 community. Further, when the Working Group identified approaches for solving distributed search and retrieval problems or identified changes to or clarifications of the standard's specifications, presentations were made to the ZIG. The Z39.50 Maintenance Agency represented by Ray Denenberg was a regular participant in the Working Group which also assisted in providing input into the Z39.50 standards work
As it turned out, however, Z39.50 was not the only standards effort which related to or affected Project CHIO. The Dublin Core Metadata initiative that began in 1995 turned out to have an important influence on the CIMI Profile. Project staff monitored and took part in that initiative because of perceived opportunities for using Dublin Core to support wider interoperability. The final Profile incorporates the use of Dublin Core elements for both searching and retrieval.
Community-based standards efforts gained visibility during the last half of Project CHIO. For example, the Art Museum Image Consortium (AMICO) developed a set of core record elements for use in building a library of digital documentation of art available under educational license. The Record Export for Art and Cultural Heritage (REACH) initiative of the Research Libraries Group (RLG) is an effort to create a testbed database of museum object records. The goal is to export existing machine-readable data from heterogeneous museum collection management systems and analyze the research value of the resulting database when researchers use a single interface to search the database in conjunction with RLG's other resources. It was imperative that CIMI's work was cognizant of these related projects, and as a result, the CIMI Profile specifications accommodate record elements from AMICO, REACH, and other community-based standards (e.g., the Museum Documentation Association's SPECTRUM standard).
6. Open Issues and Future Work
Project CHIO accomplished a first, and foundational, step towards integrated and distributed access to cultural heritage information. A primary accomplishment was the interoperability of separately developed implementations supporting the CIMI Profile. Most of the effort in the testbed, however, was to assess the specifications for interoperable search and retrieve of records. In the digital environment, there will be a need to navigate between collections and object records.
CIMI's proposal (July 1997 ) to NEH, CHIO-II: Practical Access to Cultural Heritage Information Online, highlighted the vital importance of identifying appropriate resources and navigating collections to extend the type of access demonstrated in Project CHIO. The CIMI Interoperability Testbed addressed "known" resources (i.e., a selected set of databases mounted by the interoperability testbed participants). The next step is to develop mechanisms, as proposed in the CHIO-II proposal, for users to identify potentially useful collections with which to begin their information gathering activities. Subsequent to the identification of those resources, users must be able to navigate seamlessly through and among those collections. Z39.50 has an important role in navigating digital collections, and future work must explore and specify how this can be done. CHIO-II provides a strategy for accomplishing those goals. And, as noted below, many of the participants in the current interoperability testing identified navigating digital collections and the further integration of metadata as high priority work items for CIMI.
For the CIMI Profile, navigation of digital collections is one of the next major challenges. In addition, CIMI views the work around the Dublin Core Metadata initiative and the World Wide Web Consortium's specifications for the Resource Description Framework (RDF) and Extended HyperText Markup (XML) as important points of reference for developing the next stage in the CIMI Profile.
Beyond specific profiling and testing activities, there are other important activities CIMI plans to address:
Participants in CIMI's Z39.50 Interoperability Testbed indicated strong interest in continuing with Z39.50 profiling activities. Determined in part by CIMI's strategic needs and programs of work (e.g., the Metadata Project and the Integrated Information Management Case Study) and in part by the needs of the participants, the CIMI Z39.50 Working Group will identify and prioritize a set of profiling activities that can be successfully concluded by January 31, 1999. This will include interoperability testing of specifications that can be rolled into a new release of the CIMI Profile in the first quarter of 1999. Some of the new work items include:
Available CIMI resources requires that the highest priority items addressed first. Yet, this list shows that CIMI has a full Z39.50 work agenda extending into the foreseeable future.
7. Dissemination of Information
Throughout Project CHIO, the project director and the Z39.50 project manager pursued opportunities to communicate the ongoing work and accomplishments of the Project. In addition, participants in the Working Group also participated in presentations and publications related to the work. Attachment K lists a selection of presentations and publications related to Project CHIO. CIMI project staff will continue to publicize the results of the project.
The CIMI Website is a central point of information dissemination. Many project documents have been available on the website for the past year. This has allowed interested parties not affiliated with Project CHIO to monitor the ongoing developments of CIMI's Z39.50 work.
With the completion of the CIMI Profile: Z39.50 Application Profile for Cultural Heritage Information, Release 1.0, CIMI will make it widely available in both electronic and paper formats. CIMI has discussed with the National Information Standards Organization (NISO) distribution of the CIMI Profile through NISO Press. (NISO is the standards developing organization responsible for Z39.50).
CIMI also plans for more extensive user documentation on the CIMI Profile, the CIMI Access Points, and the CIMI Element Set. There is a need for documents that can convey the utility of the CIMI Profile but do so in a non-technical language. As part of CIMI's Z39.50 work, such user documentation is an extremely high priority.
8. Concluding Assessment
By way summary, NEH support for Project CHIO has allowed CIMI to demonstrate to communities such as museums, libraries, and others the potential for integrated and distributed access to cultural heritage information -- no matter where it is located, no matter the local structure it is stored in, and no matter the format of the resource. Building globally distributed digital collections and providing meaningful search and retrieval of objects in those collections is an incremental process. CIMI's Project CHIO has provided a foundational step towards reaching shared visions of information access.