Setting

The Eberly Family Special Collections Library is located on the University Park campus of Pennsylvania State University. Housing over 200,000 printed volumes, the Special Collections Library serves a range of researchers, including undergraduate and graduate students, professors, and community members.

In the past, the Special Collections Library was three distinct units: the Rare Books Room, Historical Collections and Labor Archives, and the Penn State Room (later called University Archives) (Penn State University Libraries, n.d.). The three units were brought together administratively in the 1970s, and moved into a shared physical space in 1999. Although all materials are delivered to patrons through one service point, behind the scenes, materials remain organized in these three historic units.

Problem

Legacy practices for assigning home locations have led to retrieval problems. The Special Collections Library uses nearly 100 home locations. For example, within Rare Books, artists' books are shelved together in a "Fine Printing" home location, while books in the utopia collection are assigned the "Utopia" home location. Some of these are then further subdivided into sub-locations. "Fine Printing," for example, is divided by publisher, so that all books published by Bird & Bull are shelved in one location, books published by Compagnie Typographique in another, and so forth. In total, 19 publishers had established Fine Printing sub-locations. To add to this confusion, some sub-locations are actually located in different physical areas. The Allison-Shelley Collection, named for a donor, is shelved partially in the Special Collections stacks, and partially in a named room on a different floor. Both are assigned the "Allison-Shelley" home location. Only sub-location indicates which item is where.

This arrangement allows curators, instruction librarians, and exhibition planners to quickly locate materials, but is not intuitive for the reference staff who retrieve and shelve items. As a result, item retrieval was frequently time consuming, causing patrons to wait while staff looked for their item.

Prior to this project, cataloguers recorded sub-location in a public note in the item record, which presented several problems. First, while public notes display in the online catalogue, they do not print on retrieval slips or call number labels. In addition, text in this field is not searchable, making it impossible to generate accurate shelflists for sub-locations. Finally, if Penn State were to migrate to a new library system, there is no guarantee that these notes would transfer.

We needed to devise a new approach for recording sub-location information. We needed the new approach to allow printing on retrieval slips to make the item's location clear to staff and decrease retrieval time. Staff needed to be able to search sub-location to generate accurate shelflists. Finally, it needed to be protected in the event of future migration.

While the easiest approach would have been to create separate home locations, due to the large number already established, our systems librarians preferred we find another option. Instead, we elected to implement a new, locally-defined MARC field, MARC field 799, to capture sub-location information. Adding the field to the catalogue was a simple matter of defining a new policy in our ILS. Populating the field with sub-location information, however, was more involved. In total, we identified 5 home locations with sub-location information we needed to record in the 799 field, totaling 63 sub-locations and over 6,500 items. Adding this information by hand would have been time-consuming and risked introducing human error.

Evidence

To gather sub-location information, we decided to use analytics software. Ben Showers defines "analytics" as the "discovery and communication of meaningful patterns in data" and "analyzing data to uncover information and knowledge (discovery) and using these insights to make recommendations (communication) for specific actions or interventions" (p. xxx, emphasis in original). Analytics reports would allow us to generate lists of all public and internal notes, find patterns, and spot variations.

Penn State University uses BLUEcloud Analytics from SirsiDynix. We generated a report to retrieve public and internal notes from item records for the five collections with sub-locations. The report output included: title control number, title, author, barcode, call number, home location, internal notes, and public notes.

Figure 1

Sub-location recorded in a public note field.

Figure 2

Bibliographic record with MARC field 799 inserted.

After running the analytics report for each home location, we exported it to a CSV file. A few problems became immediately apparent. First, while we had expected to see sub-location information recorded in public notes, we learned that this information was also recorded in internal notes. The Fine Printing collection, for example, contained 2,743 items. Of these, 379 items had a public note, where 1,018 had an internal note, showing that the internal note was actually used more frequently than the public note. Second, while many of these notes recorded sub-location, some recorded other information, such as limitation statements or binding notes. Finally, we found numerous variations in name form for some sub-locations. For example, "Children's Literature" was recorded variously as "C.L.," "Child. Lit.," "Children's Lit.," and so forth, totaling over 20 variations.

Using OpenRefine (http://openrefine.org/), an open source tool for cleaning data, we separated this information into different columns, isolating the sub-location information. Following this, we used OpenRefine again to normalize location names. Using OpenRefine, we were able to edit all identical cells, so variants were quickly updated to the full name form for each sub-location.

Implementation

After successfully isolating sub-location information and normalizing name forms, we needed to push this information into bibliographic records. Using the item information from the analytics report, our Digital Access Team successfully pushed MARC 799 fields into the appropriate bibliographic records, successfully updating all 6,500 records across 5 home locations. Moving forward, cataloguers will add this information directly to the 799 field rather than using the note fields. In addition, since we had discovered all the variations in names for sub-location, we were able to normalize and document name forms, ensuring that cataloguers will enter the correct form in the future.

Outcome

Implementing the MARC 799 field for sub-location had some immediate impacts. First, we were able to map the MARC 799 field to our Aeon retrieval system. Sub-location information now prints on retrieval slips, which enables faster and more accurate retrieval and re-shelving of these items.

Adding sub-locations in the MARC 799 also allows us to generate shelflist reports reflecting actual shelving order. Now, we can simply search for records with a given sub-location name in the 799 field and sort the results in call number order. Staff can perform shelf-reading more easily, which in turn improves collection maintenance and security.

In addition, as sub-location data is now in the bibliographic record rather than the item record, it is more visible and protected in the event of future migration. This has become an even more pressing issue as Penn State University is preparing to implement a new catalogue discovery layer, in which public notes will no longer be visible.

Reflection

Our chief obstacle in this process was gathering data from BLUEcloud Analytics. BLUEcloud relies heavily on pre-packaged reports, and none of the reports available provided the information we needed. We worked closely with our local BLUEcloud Analytics expert team to write and test the report, making changes as needed to ensure we captured all of the note fields, along with item information to update records later.

The rest of the process was relatively straightforward. In addition, since the report has already been written, it's now available for use to other local BLUEcloud Analytics users, and we won't have to repeat creating this report in the future.

However, while the addition of the MARC field 799 fulfilled the immediate project goals, the larger problem of having 100 home locations remains. Moving forward, we hope to address this, potentially condensing home locations to a smaller number. When (and if) we do this, the 799 fields may be obviated, but it could be several years before we take this step. In the meantime, the sub-location information in the 799 field will play a valuable role in retrieval and collection maintenance. If we do later decide to condense our home locations, the shelflist reports made using the 799 fields will be invaluable for ensuring accurate interfiling of materials.

Conclusion

Analytics are a powerful tool for finding patterns in catalogue data and targeting records to edit. Using analytics allowed us to fulfill our project goals by getting a list of every sub-location recorded in a note, normalizing sub-location name, and pushing the MARC 799 into targeted records. We completed this work quickly and accurately, and in a fraction of the time that we would have required to do this work manually. Running an analytics report has become standard anytime we need to update catalogue information on a large scale. Subsequently, we have used analytics reports for updating call numbers, maintaining genre headings, and updating home locations following collection moves.

References

A short history of Penn State Special Collections. (n.d.). Retrieved from https://libraries.psu.edu/about/libraries/special-collections-library/short-history-penn-state-special-collections.

Showers, B. (Ed.). (2015). Library analytics and metrics: Using data to drive decisions and services. London: Facet Publishing.