HOLLIS record de-duplication and configuration details

Technical overview

We are using Primo VE's de-duplication feature to minimize duplicate records in search results. Data elements used to determine which records should be de-duplicated are detailed below in ‘Dedupe criteria.’ De-duplication applies to Alma records only.

In general, the print record is preferred over the online record. In some circumstances, the online record may be preferred. This can happen if the search query matches only terms in the online record and does not match terms in the print record. If the de-duplicated records are for print and microform, we cannot control which is used as the preferred record because they are both physical formats.

Record updates will be reflected in VE in 15 minutes.

About transitivity: if a record has a matching dedupe key with another record, it is added to the same group. Once a match is found, the system does not continue searching for matches since a record can belong to one group only. If there is ever a need to force

For CDI records, de-duplication is controlled by Ex Libris. You may see ebooks that are represented through Alma as well as CDI, and these will be separate search results. We minimize these to the extent possible by using a setting that allows us to suppress ebooks from CDI if we are loading portfolios for a collection in Alma.

Troubleshooting

If records have been de-duplicated but they should be separate, correct the bibs in Alma. See the Dedupe criteria below. When viewing a full record that has been deduped, you can add this parameter to the end of the URL to see which record is preferred and which record(s) are part of the dedupe set:

Primary record and additional de-duplicated record numbers will appear as follows:

Open

You can also use the ‘MARC view’ to see information about the primary record.

To troubleshoot ReCAP record problems please visit: ReCAP partner metadata in HOLLIS

Dedupe criteria

Serials

• ISSN (022 a, 776x) + 245a

• Also: Local key 959 built from OCLC numbers in 035 and 776 $w (key is only constructed when there is no ISSN in the record)

Non-serials

• ISBN (020 a, 776z) + 245 abnp + 008 Date1

ReCAP records

• 959 SCSB dedupe key

Notes

• Field 775 is deliberately not used to cluster.

• At one time LCCN was used, but there were too many errors.