NZ Testing - IZ linking pilot study

Objective:

To test Alma workflow for linking IZ records that lack a Sirsi 035 to the Network Zone (NZ). This study will help to establish a clear set of instructions for future IZ linking and identify related decisions for Bib and Met to consider when matching or moving IZ records to the Network Zone.

Duration:

One week

Study Group:

Humber, Sheridan, Lambton, La Cite

Plan:

Analyze records in your Institution Zone to determine how many records did not match upon migration due to lack of Sirsi 035.  Keeping in mind that some IZ records should remain local (e.g., brief records), search IZ to determine availability of alternate unique identifiers in bibliographic 035$a and 020$a, etc. You will use these match points to configure and test the job to ‘Link a Set of Records to the Network Zone.’ See NZ Testing - IZ linking pilot study -- Background Notes for additional rationale, context and links to Ex Libris resources.

Steps:

1. Roles: ensure you have the roles to do a manual job. You must be able to access the Run a Job. Further details are listed in the EXL documentation on Manual Jobs.

2. Analyze your collection: use search queries that produce sets of unlinked items.  Once you have determined which queries are most helpful in showing unlinked items, you will use them to create itemized sets and define the linking parameters (e.g., choice of unique ID) needed to run a linking job.  

Open

Open

Note: the above searches can be further restricted to library location, example:

3. Create and Save Two Itemized Sets:  Starting with examples 2 b & 2 c above, create two itemized sets. Identify your sets clearly. Note that query 2a (above) does not need to be saved for linking purposes – it simply provides you with a total picture of unlinked content in your ILS.

Steps to create an itemized set  and select members for testing purposes:

• Go to: Admin > Manage Jobs and Sets > Manage Sets > Add Set > Itemized (Set content type: Physical titles)

• Name and describe your set(s)

• Then click ‘add members to set’ > do a Repository Search (choose or modify a search from above)

• Select members from the results list to add to set, then ‘Add selected’ and ‘Done.’ Choose to start with a manageable set size, e.g., 50 - 100 records.

• Keep Excel lists of each itemized set that you are planning to link. You will use these lists to document linking and comment on the the accuracy of the linking.

4. For each set, ‘Run a job’:  Go to Admin > Manage jobs and sets > Run a job.

• Select ‘Link a set of records to the Network Zone’ 

• Choose the set(s) that you saved:

• Enter task parameters and match method that fit the unique ID prevalent in the set you saved.  For the purposes of this study, we will restrict our task parameters to either OCLC number (for Set 1) or ISBN (for Set 2), however further information on match methods is available at Managing Import Profiles. We recommend that you leave blank ‘ignore resource type during matching’ and ‘contribute IZ records’ until further study is done.

• ‘Review and Confirm’ then Submit

• All linked titles will be identified by a NZ icon:

5. Document results: For each linking job, go to: Monitor Jobs > select your job > Report to see how many records linked

• How many linked? How many remain unlinked?

• Do a title check of your itemized set against the Excel spreadsheet. Go to Admin > Manage Sets > Your itemized set > Members. Review items that show a NZ linking icon: mark linked titles on your spreadsheet; check the bibliographic record; create a column for comments regarding accuracy of linking.

• Spot check remaining unlinked records to determine type of records, locations, etc. Comment on whether there are other unmatched unique identifiers in the records that may be worth exploring (e.g., 024); if remaining unlinked items have no unique identifiers; or if remaining unlinked items should simply be left in IZ

6. Scale up: once you have reviewed, are happy with linking results from a small set, consider scaling up to a larger set or a particular location that needs to be linked to the NZ (it’s possible that not all of your library locations will contain records that need to be linked). 

Experiences:

• Humber: started with leisure reading (titles purchased directly) - 487 titles - linked 94 on ISBN, but 1 ISBN not enough; can share MMS

  • no SIRSI but with OCLC no.: 706 records 375 linked; 331 not

• Lambton - set of 60 - capitalization in source record - 21 linked; 9 were not in NZ; 27 did not match;

• Sheridan - some punctuation, e.g., pipes, might throw off matching

• La Cite - two sets;

  • OCLC 156 records; TESTED 50 from set: 17 linked; 33 did not link

  • ISBN; 3000 records; TESTED 50 from set: 33 linked 17 did not link (5 visual resources)

Questions:

1. Humber - @Dijana Kladnjakovic asked if we can run a complete file and EXCLUDE a location? Answer: rather than excluding a location, Alma’s advanced search allows you to include multiple locations in advanced search. So, “for certain drop-down lists, such as library and permanent physical location, you can select multiple matching values (see the example for Library above).” For more information see Searching in Alma, step #4.

2. Local fields (La Cite) - local fields have come across – where did they come from? Louis to check source record.

Results:

Please send results by May 6, 2022 to: Danielle Emon, with cc to Dijana and Stacey

Next steps:

Danielle, Dijana & Stacey to reconvene: 2nd week May

Whole group: later in May to share findings

Findings:

• Page 1+ IZ_NZ Linking Study Report - Presentation

Open

• NZ Pilot Analysis_Final

Open

• Project Record Files Used for in Test Environment

Further Testing and Consultation:

To follow up on recommendations made in the NZ Testing – IZ Pilot Linking Study, May 25, 2022, additional linking jobs were tested in the Alma test environment and the Sandbox.  The testing goals were:

• confirm that libraries could successfully use the Multi Step Match Method for linking

• review the impact of the parameter ‘contribute IZ records upon no match.’

1. Test environment linking with Multi Step Match Method:

Immediately before cutover – IZ linking was tested using Confederation College’s test environment, using the following parameters:

•  Multi Step Match Method

• Contribute unmatched to NZ

Results:  The job was completed with errors.  Unlinked records were not contributed to the NZ. The records in the file had a mix of ISBN and OCoLC numbers for matching.  When a subsequent job was run matching only on ISBN, linking was successful as long as the ‘contribute IZ records’ box was not selected.

Shared with Ex Libris: The above results were shared with Ex Libris along with a request for any additional configurations that might be needed to link using the Multi Step Match Method.  At the time, Ex Libris was not aware of any limitations for using the multi-step match methods.  However, the Multi Step Match Method is a newer process, and there was also subsequent confusion as to whether it could be used for IZ linking. After some consultation and some assistance from a cataloguing specialist, a follow up meeting was held to discuss IZ NZ linking and related issues.  Ex Libris supplied a ppt containing recommended configurations for using the Multi Step Match Method specifically for IZ-NZ Linking. 

2. Sandbox testing:

After recommended configurations were made to the NZ sandbox, a linking job was run in the sandbox using a test set of OCLS records.  The linking job yielded 100 error messages.  A detailed message was sent to Ex Libris as per their request.  While Ex Libris was not able to open the error log, they confirmed that the configurations and the workflow were good.  Their analyst confirmed that their tests run using the Multi Match Method were successful.

 A second sandbox test of 113 records yielded more positive results, with NZ linking and contributions to NZ.  Links created were tracked on a spreadsheet generated from the itemized set. 

Lessons Learned:

1. Successful linking with multi-step is dependent upon additional configurations in NZ as laid out by Ex Libris. OCLS staff will need configure the NZ accordingly before project starts.

Open

2. ANYTHING can be contributed to the NZ.  Therefore, libraries planning to run the IZ linking job will need to restrict their query for unlinked records to those records that have an ISBN/OCLC number and keep other records off the list. 

3. Libraries would benefit from having an Analytics report that would allow libraries to create a file of bibliographic MMSIDs that could be used to create an itemized set for linking. The report criteria should include OCoLC and ISBN numbers and filter for location. Other report criteria could include 024/028 as relevant for visual resources. This analytics report could not be tested in the sandbox.

4. Note: a review of linked test records shows that there is nothing in the IZ bibliographic record to show when a record is linked to the NZ – no change in bibliographic creation date or modification date.  However, records contributed to NZ will show a bibliographic creation and modification date that reflects the date contributed from an IZ.

The results of this additional testing and the additional configurations from Ex Libris have been applied to the IZ_NZ Workflow.