Semantic Metadata and Evidence-Based Policy Assessment

An increasingly wide range of research resources is being placed online, and the traditional journal article is being supplemented with access to data sets, interview transcripts, and survey results. While these resources are of use to other researchers, the large quantity of resources available makes the management of resources an increasingly important subject. One solution to the problem of potential information overload is the adoption of metadata and ontologies to not only describe the resources but the concepts within the resources and the relationships between them. In their 2009 article, e-Social Science and Evidence-Based Policy Assessment, Edwards et al. detail the process of creating metadata and ontologies to support evidence-based policy assessment. They demonstrate not only the complexities of attempting to conceptualise the necessary components of social science resources, but the importance of the user-interface, and how web 2.0 and semantic technologies can be combined to create community driven metadata.

Evidence-based policy assessment is increasingly important as policy-makers move away from opinion-based decision making, and instead require that policy is grounded in research that provides sufficient evidence to support any policy recommendations and an audit trail that allows policy-makers to understand how conclusions were reached. Researchers need to be able to not only discover available resources, but the provenance of those resources, i.e. how the data was created and its context.

There are various difficulties in creating a framework for the management of evidence, including the range of resources available and the wide variety of both quantitative and qualitative research methods (as well as mixed methods) that may have been adopted in the research process. The creation of a provenance model is not easily done by the analysis of data sets alone, but by interviewing social science researchers, ontologies can be created that capture the necessary metadata to provide the provenance record.

The ability of an ontology to capture the necessary metadata is not enough on its own, an appropriate user interface for the creation of the metadata is equally important if researchers are not only going to go through the inconvenience of creating the metadata, but create useful metadata. The ourSpaces virtual research environment has been designed to support researchers conducting evidence-based policy research, encouraging users to describe the resources themselves through a novel natural language interface that also shows a tag cloud of other terms that have been used within particular fields to create community driven metadata.

The creation of ontologies and metadata to describe resources and their provenance is not only important in policy-related research, but in aiding scientists in the management of resources in all disciplines, including those conducting interdisciplinary research which is likely to introduce new complexities. Importantly, the provenance concept has the potential to encourage greater reflection at all stages of the research process, and greater methodological rigor.

Edwards, P. et al. (2009) e-Social Science and Evidence-Based Policy Assessment: challenges and solutions. Social Science Computer Review, 27(4), 553-568.