Metadata migration to Islandora: Is there an easy way?

This presentation will introduce UCF’s digital collection migration from DigiTool to Islandora, the new content management system for the state universities in Florida. It discusses the issues in DublinCore (DC) to MODS transformation, explores the possible options, the approach adopted and the tool used for MODS metadata editing. As part of the state-wide Islandora implementation, UCF has been migrating its collections in DigiTool to MODS records for Islandora to ingest. In migrating from a less granular metadata schema to a more granular one, many issues are involved such as data ambiguity, overly generic data representation, the markup inadequacy in describing sub-elements and element relationships, and a less intricate data structure. Two options were explored: a. revamp the Library of Congress’s DC-MODS stylesheet to produce more desirable MODS metadata, b. edit the MODS records generated from a more generic stylesheet conversion. Due to the fact that the consensus needs to be reached among the state universities for any change, only some adjustments such as adding local subjects and online thesauri were made to the LC stylesheet state-wide, and the major work of MODS metadata editing fell on the individual university libraries. At UCF Libraries, Notepad++ was used to edit the MODS records, such as the first set of 847 records in the Political & Rights Issues & Social Movements (PRISM) collection. In batch editing the MODS metadata and dealing with the DC-MODS transformation problems, data patterns in author year, author role terms, publication places, corporate and conference names were identified, data normalization and cleaning was executed, and several types of mark up and editing were performed: authors were marked up to distinguish author year and role from name; personal, corporate and conference names were differentiated; the main entry and added entries were distinguished; subtitle was separated from the main title; publication place was distinguished from publisher; topical, temporal, geographic, genre subdivisions were marked up for subjects; and series name and other common fields were added for the collection. This presentation addresses the common issues in DC-MODS metadata mapping and transformation, discusses possible solutions of customizing the XSLT stylesheet and editing the MODS XML records, and the balance that needs to be sought in pre- and post-transformation. It also raises some interesting questions in machine vs. human labor and utilizing the computer’s analytical power. It invites audience to participate in a wider discussion