Cataloger acceptance and use of semiautomated subject recommendations for web scale linked data systems

As catalogers begin to integrate linked data descriptions into large-scale discovery graphs through RDF editors, interventions such as semi-automated subject description (http://lcsh.annif.info) are extending and supporting their professional expertise. A large corpus of 9.3 million (9,304,455) title and subject pairs from the IvyPlus Platform for Open Data (POD), along with SVDE bibliographic data, were used for training a semi-automated subject indexing tool for use in BIBFRAME linked data editors. Thereafter, catalogers evaluated the automated subject outputs for inclusion in their descriptions of BIBFRAME resources and the general usefulness of semi-automated subject suggestions. This paper presents the findings of a mixed-methods inquiry to better understand catalogers’ preferences for incorporating machine learning outputs into their work.