New possibilities for metadata creation in an institutional repository context

Abstract

PurposeThe purpose of this paper is to develop automated methods for creating metadata for documents in an institutional repository.Design/methodology/approachTwo methods are examined for automatically building metadata in an institutional repository context. Text mining techniques are employed to discover relationships among documents with similar content, from which are inferred possible values for missing or incomplete metadata elements. Machine learning techniques are used to identify and extract specific metadata element values from document content.FindingsText mining techniques can be used to cluster documents with similar content. This allows values for metadata elements, like keyword, to be projected from documents with established metadata to related documents. Machine learning techniques are found to be reasonably accurate for extracting from documents values for metadata elements, such as, title, author, and abstract. Results show sufficient promise to support the next phase of the project: the development of assistive tools for use by metadata specialists to create or edit document metadata.Originality/valueThis paper focuses on the use of automated metadata extraction techniques to assist metadata creation, lessening the time and effort required to add documents to institutional repositories.