Metadata quality issues in learning repositories

Resumen

Metadata lies at the heart of every digital repository project in the sense that it defines and drives the description of digital content stored in the repositories. Metadata allows content to be successfully stored, managed and retrieved but also preserved in the long-term. Despite the enormous importance of metadata in digital repositories, one that is widely recognized, studies indicate that what is defined as metadata quality, is relatively low in most cases of digital repositories. Metadata quality is loosely defined as "fitness for purpose" meaning that low quality of metadata means that metadata cannot fulfill its purpose which is to allow for the successful storage, management and retrieval of resources. In practice, low metadata quality leads to ineffective searches for content, ones that recall the wrong resources or even worse, no resources which makes them invisible to the intended user, that is the "client" of each digital repository. The present dissertation approaches this problem by proposing a comprehensive metadata quality assurance method, namely the Metadata Quality Assurance Certification Process (MQACP). The basic idea of this dissertation is to propose a set of methods that can be deployed throughout the lifecycle of a repository to ensure that metadata generated from content providers are of high quality. These methods have to be straightforward, simple to apply with measurable results. They also have to be adaptable with minimum effort so that they can be used in different contexts easily. This set of methods was described analytically, taking into account the actors needed to apply them, describing the tools needed and defining the anticipated outcomes. In order to test our proposal, we applied it on a Learning Federation of repositories, from day 1 of its existence until it reached its maturity and regular operation. We supported the metadata creation process throughout the different phases of the repositories involved by setting up specific experiments using the methods and tools of the MQACP. Throughout each phase, we measured the resulting metadata quality to certify that the anticipated improvement in metadata quality actually took place. Lastly, through these different phases, the cost of the MQACP application was measured to provide a comparison basis for future applications. Based on the success of this first application, we decided to validate the MQACP approach by applying it on another two cases of a Cultural and a Research Federation of repositories. This would allow us to prove the transferability of the approach to other cases the present some similarities with the initial one but mainly significant differences. The results showed that the MQACP was successfully adapted to the new contexts, with minimum adaptations needed, with similar results produced and also with comparable costs. In addition, looking closer at the common experiments carried out in each phase of each use case, we were able to identify interesting patterns in the behavior of content providers that can be further researched. The dissertation is completed with a set of future research directions that came out of the cases examined. These research directions can be explored in order to support the next version of the MQACP in terms of the methods deployed, the tools used to assess metadata quality as well as the cost analysis of the MQACP methods.