Exploring metadata providers reliability and update behavior

Abstract

Metadata harvesting is used very often, to incorporate the resources of small providers to big collections. But how solid is this procedure? Are the metadata providers reliable? How often are the published metadata updated? Are the updates mostly for maintenance (corrections) or for improving the metadata? Such questions can be used to better predict the quality of the harvesting. The huge amount of harvested information and the many sources and metadata specialists involved makes prompt for answers by examining the actual metadata, rather than asking about opinions and practices. We examine such questions by processing appropriately collected information directly from the metadata providers. We harvested records from 2138 sources in 17 rounds over a 3-year period, and study them to explore the behaviour of the providers. We found that some providers are often not available. The number of metadata providers failing to respond is constantly increasing by the time. Additionally, the record length is slightly decreasing, indicating that the records are updated mostly for maintenance/corrections.