How does the CDI work with the Unexpanded Search in Primo (AKA full-text search or filtered search)?
Answer
Alma publishes your electronic holdings to the Central Discovery Index (CDI) through a publishing profile (Resources>Publishing>Publishing Profiles) called "Publish electronic records to Central Discovery Index," which runs daily. This happens so that you can connect your electronic titles with all the relevant search records for those titles in the CDI and make them show up as available in Primo VE. Without this publishing job, only your Alma title-level records would display in Primo VE.
You can see the time that it runs if you click the three dots to the right of the job.
As soon as you activate or deactivate an electronic portfolio, the information about that title is added to the list of content that's sent to the CDI through the daily publishing job. This happens automatically, and has nothing to do with whether your electronic collection's CDI Search Settings are set to Active.
In general, here's what's included in your CDI publishing file:
- Portfolios that you have active for full-text, such as ebooks or ejournals, including the ISBN/ISSN, coverage dates, and embargo information
- Collection level availability if it's a link in record collection and you are either leaving the collection bib record unsuppressed or you've activated the CDI-only full-text option for the collection
There's a way that you can see the list of content in your holdings file, as described by Ex Libris: The Output of the Publishing Process.
If you want to prevent content from being included in the daily publishing file, such as ebook titles for collections that do not offer chapter level linking, you may check the "Do not show as Full Text available in CDI even if active in Alma" checkbox on the CDI tab in Edit Collection.
Once it's published, how does the content get searched in Primo VE?
The holdings file matches the content with the ****entire CDI universe of records**** and adds anything that matches by identifier and coverage date to your full-text search. What you have active or inactive for CDI search has NO impact on the full-text search's search records. There is no way to deactivate any package from being used for the full-text search. There is also no way to only have the titles match on content that you get from your specific vendor. For example, there are blog posts and other online content from newspapers in the CDI because maybe Proquest provides access to that content on their platform, but you may subscribe to Gale and not Proquest, and Gale may not include that content on their platform.
The record that is used for the search could be from ANY package in the CDI. You might get lucky and end up with the search record that's the same as the collection that you have active in Alma, like Gale OneFile: News, but there's no guarantee that that's going to happen. CDI uses a merged search record, where it clumps together all of the search records from all of the vendors for that specific article and title, but it only displays one search record out of the clump of records. It's not designed to pick the best record to display/use.
It's possible that the merged record doesn't always display the same record each time you search for an article. Meaning, if there are 5 records included in the group of records, it's possible that the first time you search, you get record #1, but the next time you search the same thing, you get record #4. This can cause significant linking problems because record #1 might have the correct article or document id, or the title that matches how the title is indexed by your subscription database, but record #4 from a different vendor for the same article may not.
It's also possible that the search record may be enhanced by the other records in the merged set, or that you might get search results based on other records in the set but that content doesn't display on the full record in Primo VE.
Best Practices for an accurate holdings file:
- Never use the CDI-only full-text tag for anything except collections that are link in record
- Make sure that all of your selective packages, like ebook packages where you only buy some titles, are marked on the Edit Collection, CDI tab "We subscribe to only some titles in this collection" = yes. If you don't do this, the entire collection will be published as available in Primo VE.
- Mark any of your aggregator collections where Full Text Linking in CDI = Linkresolver in the same way, with "We subscribe to only some titles in this collection" = yes. Setting the value to No may result in linking failures, if the collection in CDI contains additional content that is not reflected in the object portfolio list in Alma." This additional content could be things like blog posts or other content that you cannot get on the platform. This is counterintuitive since you subscribe to the whole collection, but it's best practice.
False availability: you get a search result in your full-text scope that says it's available but when you open it and the live Alma fulfillment look-up happens, the availability changes to unavailable.
- Why does this happen?
- Something in your holdings publishing file is telling Primo VE that you own this item.
- The CDI analysis Tool can help you identify the collection that's causing this search item to appear in Primo VE, but note that its finding are not always 100% accurate.
- What can you do about it?
- Try to figure out where the holdings might be coming from. First, look up the journal or ebook title in your Alma to see if it has any active portfolios.
- Use the Display CTO tool to see if there's a problem with the metadata in the search record. Sometimes you do get access to the article that initially displays as available, but the search record metadata isn't accurate enough to make the live Alma fulfillment lookup match correctly. If you scroll to the bottom of the CTO display you and see if the lookup tool found matching portfolios but hid them for some reason. Pay special attention to the ISSN/ISBN and dates in the search record. If this data is incorrect or missing from the portfolio bib record in Alma, it won't match correctly. The only way to have inaccurate search record data fixed is to submit a Salesforce case and ask Ex Libris to fix it.
- Try to find the electronic collection that might be causing this record to display in Primo VE. Unfortunately, since the merged record search record's CDI collection name doesn't correspond to what you have active in Alma, this can be challenging. You could look up the ISSN or ISBN in the Community Zone and try to figure out which of your active collections might be publishing this content as available even though you can't find the title in your IZ. If you see any active collections on the CZ title record, make sure that those collections are marked "We subscribe to only some titles in this collection" = yes
Missing content: You expect to find articles or titles in Primo VE's full-text search but they're not there.
Why does this happen?:
- It could be a timing issue. It takes approximated 48 hours for CDI changes to go into effect, so you won't see article results from activated journals until 2-3 days after the activation. If the search record has already been activated through another collection, you might be able to find it in your expanded, non-full-text search. If you've activated it correctly, it'll start out as unavailable but the live Alma look-up should change it to available.
- If it's a link in record collection like Films on Demand, you need to make sure that you've both activated the collection for search AND you've activated the CDI-only full text option. It will take 2-3 days for this content to appear in Primo VE
- Content is not available in the CDI: some vendors have proprietary search records that they do not share with Ex Libris. It's possible that you may not be able to get full article coverage for titles in CZ collections. This is especially true for some EBSCOhost collections. The only way to determine if the content does not exist in the CDI is to submit a Salesforce case.
Additional SUNY Troubleshooting Documentation: