Abstract

This paper will present the possibilities of Linked Data within the SpokenWeb collections of audiovisual (AV) recordings documenting literary events and activities held in Canada since the 1950s. Over the past six years the SpokenWeb research network has digitized and described with a bespoke SpokenWeb metadata schema approximately 7000 AV recordings of literary readings, talks, radio programs, interviews, spoken word performances, and other forms of literary speech. This metadata has been entered into the SpokenWeb Collections project in the custom-developed Swallow Metadata Management System that stores metadata information in no-SQL format, and that allows manual and batch ingest and export of data to and from different systems. These configuration files are defined as JSON objects. With an extensive range of fields and subfields in the SpokenWeb schema, including technical and material data, information about production context, genre, date, location, related works, and the contents of the recordings, there is great potential for the enrichment of this data through interlinking the metadata we have produced with other data sources.

Simply defined, linked data is structured data that is interlinked with other data. Among other benefits, linked data expands interoperability and reusability, and improves search and discovery within and beyond a data set. In the SpokenWeb metadata schema, this interlinking occurs through different fields such as location, creator, contributors, and most notably within the content’s field, which is the most robust and complex field of the entire schema. In this presentation, we wish to share the approaches we have taken thus far to enhance our metadata through a tool developed for the supervised integration of wikidata links to data that appears in our contents fields.

To this end we will first describe the nature of SpokenWeb data, its purpose and potential uses, and will show what it looks like in the Swallow database. We then will demonstrate how Swallow handles linked data as a step towards providing a demonstration of a tool we have developed to integrate the task of interlinking select terms of our metadata task into the cataloguing process. We will explain how the Wikidata import function works within the Swallow system and reflect on some of the lessons we have learned from our experiments with this tool about achieving greater accuracy and speed in the links produced. Finally, we will share some ideas (through examples) of potential uses of the linked data we have integrated into SpokenWeb metadata to generate temporal and location-oriented visualizations of select data sets.

All of this will be offered as an invitation to participants in the conference to share their knowledge, ideas and observations about our approach, and about the myriad possibilities of linked data technology that might be of interest to a community of scholars interested in doing literary research with audiovisual collections, moving forward.