Library Hub Compare results: FAQs

There are several reasons why a batch number search in Compare returns a smaller number of results than you upload:

  • The file you uploaded may contain duplicate search values
  • If you ran your report to identify items to check in Library Hub Compare some time ago, it is possible that your holdings in the NBK have been updated in between running your report in your library management system and uploading it into Library Hub Compare
  • Your NBK upload configuration may be set to drop specific locations from your library collection on upload, but you may have items from these locations in your upload file. You can check whether your library has holdings drops set up by contacting the Library Hub team nbk@jisc.ac.uk
  • Library Hub Compare uses the data from the NBK, which is a deduplicated A match will only be reported where both your library and at least one other library have items attached to the same deduplicated record. There may be other records for the same item in the NBK, but these will not be reported as a match. The NBK match process is deliberately conservative, to avoid false matches. Matching is particularly affected by poor metadata, especially the lack of standard numbers (e.g. ISBNs). This means that older publications that were catalogued prior to the wide adoption of ISBNs, MARC and library management systems are much less likely to be matched.
  • The NBK is format agnostic. This means that records for the same edition of a title will be merged to create a deduplicated record regardless of whether they are print or electronic, and the ISBNs for both formats will be included in the final deduplicated record. In your institution, you may have one record for each format (so one record for the print version, one for the e-book, and one for a microfilm), but we would merge these all into one.
  • Your upload file may include multi-volume works which have individual records in your library management system. This local decision to create a separate record for each volume may not be reflected in the NBK, which may have one record for the entire multi-volume set.
  • If you used a spreadsheet as your upload file, it is important that standard numbers are defined as ‘text’. If they are not defined as text then leading zeros will be dropped from the standard numbers, invalidating them and affecting your search results.

Testing carried out for the UK Print Book Collection project confirmed that an acceptable discrepancy rate for searching in Library Hub Compare is around 4%.

ISBNs are often thought to be unique identifiers, but this is not the case and they are sometimes reused by publishers. They were also not created until the mid-1960s, so will not be present in the MARC records for books published before this time. Finally, the same item with the same ISBN may have multiple records in the NBK that haven’t merged into a deduplicated record. This is often due to poor metadata that doesn’t enable confident matching to take place.

This is likely to be down to poor metadata not enabling the matching and deduplication of records in the NBK. This particularly affects older, pre-ISBN, publications as there are fewer data points to confidently match against.

The UK PBC project is using your local record ID number (also known as an MMS ID, TCN, or 001) to check for duplicate holdings. It is possible to upload a subset list of your “unique” items to Compare and enable greater deduplication by including ISBN matches. This feature is towards the bottom of the ‘batch search’ tab and is labelled ‘Deduplicate’. You need to be aware that this option comes with an increased risk of mismatches and mis-consolidations, and for this reason it is not being used in the UK PBC reports. However, you could use these results to make improvements to your MARC records in order to increase the chances of a match in the NBK, perhaps by using the record uploaded to the database by another library from Library Hub Cataloguing, and then run your UK PBC report again after your next NBK update files have been processed.

Often, lists of title phrases do not return the number of results you expected. This may be because the list generated by your library management system includes title phrases that may match with dozens, hundreds, or even thousands of book titles (e.g., if you upload 500 titles including one-word titles that appear on many books such as “Biosciences” or “The Oxford Handbook of”). Title phase searches may also be affected by diacritics, punctuation, and variations in local practices for the cataloguing of multi-volume works. Title phrase searching is much less accurate than searching by your local item ID or by ISBN but can be useful for small sets of older publications.