Creating new policies and automating standardization for serials holdings records

The University of Illinois at Urbana-Champaign (UIUC) Library has a large print serials collection consisting of 317,066 bibliographic titles with 474,190 holdings and 3,638,833 items. At UIUC, serials holdings records had two problems potentially causing hindrance to discovery and data interoperability: a lack of consistency in the local notes in the MARC Format for Holdings Data (MFHD) 852 field, and missing standard holdings statements in the MFHD 866 field. Assessment of local notes revealed that there were 58,723 unique notes in the 474,190 holdings records. This was caused by inconsistent policies and incorrect data entry. After categorizing notes into 22 types, the library came up with standardized local notes policies based on three principles: 1) notes do not duplicate information in the bibliographic record, 2) notes do not refer to information in a different holding record, and 3) notes are not created to conform to discovery systems. The policies also reiterated the need for each MFHD to have a summary written to ANSI/NISO Z39.71-2006 (R2011) Holdings Statements standard. The data clean-up work was performed in two ways. First, using Python and OpenRefine the local notes contained in the MFHD 852 field, subfields x and z were batch edited. The updated MFHD 852 fields were batch replaced with notes that conform to the new policies. Second, summary of holdings statements were written to ANSI/NISO standards using Python. The statements were batch added to 105,404 holdings records as the MFHD 866 field. This cut the number of print serials records with missing summary of holdings statements from 277,282 to 171,878. This paper will share how the library created the policies on local serials notes and automated data clean-up work, and will discuss lessons learned from the project.Ope