Exclude eBooks from CDI Results Discussion and Testing Appendix

Legend: not started IN PROGRESS STALLED decided

Discussion, thoughts, etc.

Primo Listserv Responses (September 2023)

• “The reason we chose not to Exclude CDI eBooks is that it also excludes them from the expanded search as well as the filtered search. Since we frequently tell people to use the expanded search results to initiate ILL requests for books not in our network, this significantly decreases the utility of the expanded search. “

• “We’ve explored and experimented with these settings, but decided not to Exclude CDI eBooks because we often found that the CDI records were more complete than the [CZ] Alma records. If we excluded the CDI records, discovery was compromised by the poor and limited metadata within the Alma records. We have done the opposite, and suppressed the Alma records in favor of CDI when that option is possible (i.e. Full Text link in record collections), but only when we’ve had complaints or reports of confusion because of the duplicate records. Definitely I would want to compare the quality of Alma vs CDI records for any collection I was considering Excluding CDI eBook records from.

• “We have a Network Zone and haven’t encountered any NZ implications, but most e-resources are managed at the IZ level and not NZ, so your experience may be different.”

• “I know that when the "exclude CDI ebooks" option first rolled out, there was a problem where it would prevent access to content/search records if you had a collection where the content was only available as a Link in Record collection. I think that that's been fixed, but I don't know for sure. That would be something to check, since you don't want to lose access to content with this all or nothing approach.”

Testing

Electronic collections to look at:

• Ebook Central Perpetual and DDA: Example title at UCSC The Arts of Indigenous Health and Well-Being.

• Access Medicine

• JSTOR

• Cinahl

• Gem post a list of some CDL Shared Collections electronic collections titles to look at.

• Hathitrust

Specific behavior to test:

• The documentation says, “eBook records from CDI are filtered out in your blended search scopes for collections with portfolios in Alma”. Does this mean collections that have portfolios activated in our Alma or does it include portfolios that exist as an option to activate?

• Does this setting do anything if CDI results are expanded?

• What counts are ebooks? What metadata is triggering assigning the rtype of book?

  • UC Davis testing - Articles are not included and Dissertations are not included as ebook

• What happens when CDL Shared Collections has activated an electronic collection as CDI only.

  • UC Davis - Hathitrust ebooks still seem to be showing.

11/3/2023 Testing Discussion

UCD: It did reduce dupes. Particularly around JSTOR, Access Medicine. Dissertations were still duplicating (things like scholarship). Found a couple of examples where a database or bib record was calling a resource a book but it was considered a report in CDI so there were duplicates. Overall, it looks promising.

UCM: Only thing that caught where excluded CDI weren’t retrieved and want to try again. Does reduce duplicates though.

UCLA: Yes, results are looking pretty good. Want to urge everyone to list whether you’re using filter by ability. Overall pretty positive change.

UCB: Had weird situation where book chapters were retrieved but could get to the title from that. Seemed to mostly work but found some instances where you got zero results. E-Resources mentioned the quality of CZ vs CDI. Sometimes the CZ records are bad but didn’t do a lot of metadata analysis.

UCSC: Wasn’t seeing a lot of duplicates. Didn’t really feel comfortable saying they saw much reduction. Do have filter by availability on. Once excluded CDI ebooks, did find some examples where records weren’t being retrieved. Did an initial search by DB ID, just to see reduction in overall retrieved. Some collections were very significant. Others it was minimal or non-existent.

UCI: Tested a handful of records having records on did reduce dups. Handful of cases where still saw dups on. Filter by availability is off.

NOTE: There might be a situation where the sandbox doesn’t have updated records which could account for some expected records not be retrieved.

Previous Testing and Thoughts by UCI (10/2022)

UCI Environment for excluding CDI eBooks

• Pros: Cleaner results for ebooks we have local records for

• Tradeoffs:

  • Update: lots of eBooks still show duplicate records in default searches.

    • Ex Libris continues to refine the way eBooks work, and we expect improvements over the next few quarters.

    Note: print from other UCs is hidden. Cynthia found that if a CDI eBook record matches print, only UCI's holdings show. So if a librarian searches for "After the War on Crime: Race, Democracy, and a New Reconstruction," a 2008 NYU Press book

    • doc ID alma991035382833204701 shows that print is held at 6 other UCs.

    • doc ID cdi_askewsholts_vlebook shows only print at UCI

    • link resolver https://uci.primo.exlibrisgroup.com/discovery/openurl? shows only print at UCI

    Why are there still dups? It seems like duplicate eBook records still appear in an "Articles, books, and more" search if the records come from a CDI collection that 1) is type database and 2) has link-resolver records. A few of the CDI databases like this with thousands of eBook records are:

    • Scopus

    • Web of Science

    • International Bibliography of the Social Sciences (IBSS)

  • Open

    

    There is still ongoing discussion as of 10/12/22 on the Primo listserv regarding this feature and still seeing some duplication

UCI Testing Documentation and Notes

Open