mZ-Interop Testbed Documents -- Phase 1

The following is an annotated list of documents that describe the logic, methods, analysis, and procedures the Z-Interop team developed to prepare the Z39.50 Interoperability Testbed.  Additional documents will be developed as interoperability testing gets underway. Most documents are available in both MS Word [.doc] and Portable Document Format [.pdf].

• Z-Interop Interoperability Testing Policies and Procedures, Phase 1 Testing
• Agreement for the Appropriate Use of Test Dataset by Participants in the Z39.50 Interoperability Testbed (Third Party Vendor Agreement)
• Indexing Guidelines to Support Z39.50 Profile Searches
• Analysis Logic and Procedures for Creating a Test Dataset of MARC 21 Records for the Z39.50 Interoperability Testbed, Phase 1 Testing
• Decomposing MARC 21 Records for Analysis
• Data Normalization Procedures on Decomposed MARC 21 Records
• SQL Data Analysis Procedures to Create Aggregate and Candidate Record Groups on a Sample of Decomposed MARC Records, Phase 1 Testing
• ZDoctor Report of SIRSI Indexing Policies for Interoperability Testing, Phase 1 Testing

Z-Interop Interoperability Testing Policies and Procedures, Phase 1 Testing (revised draft dated February 1, 2002)
[.doc] [.pdf]

This document provides an overview and the details of the policies and procedures of the Z39.50 Interoperability Testbed Project. Specifically, the document lays out the responsibilities and obligations of the Z-Interop Testbed and the organizations that participate in interoperability testing. It includes an agreement for the appropriate use of the test dataset.

Agreement for the Appropriate Use of Test Dataset by Participants in the Z39.50 Interoperability Testbed (Third Party Vendor Agreement)
[.pdf]

This document includes a Third Party Vendor Agreement that must be signed by Z-Interop participants for the use of the test dataset of 400,000 MARC 21 records from the OCLC WorldCat database.

Indexing Guidelines to Support Z39.50 Profile Searches (revised draft dated February 1, 2002)
[.doc] [.pdf]

To assure rigor in the testbed, we have developed a set of guidelines for indexing the MARC 21 records to support Z39.50 profile searches. Z-Interop staff used these guidelines to index the 400,000 MARC 21 records that comprise the Z-Interop reference implementation of the Z39.50 server and online catalog. The guidelines reflect input and revisions based on public review.  At this point, indexing guidelines are available for author, title, and subject, and any keyword searches.  The guidelines can be used by interoperability testbed participants. 

Analysis Logic and Procedures for Creating a Test Dataset of MARC 21 Records for the Z39.50 Interoperability Testbed, Phase 1 Testing (revised draft dated January 1, 2002) 
[.doc] [.pdf]

This document explains the overall logic by which records were selected for the test dataset from OCLC's WorldCat database and introduces key concepts used in the Z-Interop testbed methodology: Aggregate and Candidate Record Groups.  In addition, there is an explanation of the procedures for determining the aggregate and candidate record groups for specific test searches.  These record groups provide the foundation for Z-Interop benchmarks and for analysis of interoperability testing results.

Decomposing MARC 21 Records for Analysis (revised draft dated January 1, 2002)
[.doc] [.pdf]

The first step in creating the aggregate and candidate record groups involved the decomposition of the 400,000 MARC 21 records into subrecords based on character strings bounded by spaces (i.e., words) in fields and subfields in the MARC 21 records.  This document explains the logic and procedures for decomposing the records. The result of decomposing the 400,000 records was approximately 33,000,000 subrecords. OCLC carried out this decomposition according to guidelines prepared collaboratively by the Z-Interop team and OCLC.

Data Normalization Procedures on Decomposed MARC 21 Records (revised draft dated January 1, 2002)
[.doc] [.pdf]

No data normalization was done on the MARC 21 records or during the decomposition of those records.  Normalization was necessary to more efficiently carry out the procedures to create the aggregate and candidate record groups.  This document describes the normalization procedures carried out on the decomposed records.

SQL Data Analysis Procedures to Create Aggregate and Candidate Record Groups on a Sample of Decomposed MARC Records, Phase 1 Testing (revised draft dated January 1, 2002)
[.doc] [.pdf]

Once the decomposed records had undergone data normalization, they were now ready for the procedures to create aggregate and candidate record groups.  For Phase 1 testing, four US National and Bath Profile searches (Functional Area A, Level 0) searches are being tested.  The procedures documented here describe how aggregate and candidate record groups were created for Author Keyword, Title Keyword, Subject Keyword, and Any Keyword searches for specific search terms.  The procedures described were informed by the indexing guidelines developed as part of the Z-Interop project, the Texas Z39.50 Profile, and the Bath and US National Profiles.

ZDoctor Report of SIRSI Indexing Policies for Interoperability Testing, Phase 1 Testing (revised draft dated January 1, 2002)
[.doc] [.pdf]

This report identified the MARC fields and subfields that are indexed in the Unicorn system to support various Z39.50 Bib-1 Use Attributes. This serves as a confirmation of the actual indexing policies set up on the reference implementation. 

[Page last updated March 1, 2002]