AbstractThis paper reports results of an exploratory quantitative analysis of metadata versioning in a large-scale digital library hosted by University of North Texas. The study begins to bridge the gap in the information science research literature to address metadata change over time. The authors analyzed the entire population of 691,495 unique item-level metadata records in the digital library, with metadata records supplied from multiple institutions and by a number of metadata creators with varying levels of skills. We found that a high proportion of metadata records undergo changes, and that a substantial number of these changes result in increased completeness (the degree to which metadata records include at least one instance of each element required in the Dublin Core-based UNTL metadata scheme). Another observation of this study is that the access status of a high proportion of metadata records changes from hidden to public; at the same time the reverse process also occurs, when previously visible to the public metadata records become hidden for further editing and sometimes remain hidden. This study also reveals that while most changes -- presumably made to improve the quality of metadata records -- increase the record length, surprisingly, some changes decrease record length. Further investigation is needed into reasons for unexpected findings as well as into more granular dimensions of metadata change at the level of individual records, metadata elements, and data values. This paper suggests some research questions for future studies of metadata change in digital libraries that capture metadata versioning information.
The copyright for articles is retained by the author(s), with first publication rights granted to DCMI for publication in the electronic and print proceedings. By virtue of their appearance in this open access publication, articles are free to be used with proper attribution for educational and other non-commercial purposes. Other uses may require the permission of the author(s).