Trying out something new for this blog, I’m “live” blogging (with a slight delay to clean up my notes and track down some links) the NISO webinar, Return on Investment (ROI) in Linking the Semantic Web, featuring speakers Ellen Hays, Corey Harper, and Geoffrey Bilder.
Ellen Hays gives an overview of the semantic technologies used by Elsevier in order to offer “smart content” to its users. Semantic data will power more interesting user services and features (data mashups, visualizations, etc.). Hays identifies scale as the biggest challenge for Elsevier; there’s a huge breadth of content and numerous stakeholders to serve. Elsevier Linked Data Repository (LDR) stores RDF/XML graphs as annotation documents/content satellites. The repository allows Elsevier platforms and apps to retrieve and store content enhancements. The LDR is capable of some heavy lifting (reading/writing large amounts of data). Thesauri, etc. are stored in LDR as well.
Corey Harper focuses on how libraries might shift the focus of semantic efforts from publishing linked data to consuming linked data. CKAN lists 51 libraries that have published linked data. Two contentious approaches to data modeling are evident here: modeling data on resources (approach taken by Talis–I’ll post a link if/when I find one!) and modeling data on existing bibliographic standards (approach taken by Metadata Management Associates in expressing MARC21 in RDF). Do these approaches consider how data is being used? The efforts of linked open data in libraries, archives and museums (LOD-LAM) do not focus solely on bibliographic data; the goal of LOD-LAM is to incorporate cultural resources of archives and museums (example: Europeana). Harper says that expressing EAD in an RDF model enables resources such as SNAC (a “Facebook for dead people”). By focusing on identification rather than description of resource, computers can reveal relationships and context. Loosely quoting Harper: “I have this vision of archival data… being embedded into delivery systems.”
Geoffrey Bilder discusses the crucial role of persistent identifiers in an open linked data environment. Bilder outlines the problems that Direct Object Identifiers (DOIs) face in the framework of a semantic web environment. CrossRef’s solution leverages content negotiation to ensure that DOIs are complaint with linked data principles. Bilder briefly discusses how these lessons are informing ORCID linked data design. Bilder references the Den Haag Persistent Object Identifier – Linked Open Data Manifesto, which aligns DOIs with the basic tenants of semantic web. Bilder closes his presentation with a discussion of what linked data might look like without XML and without RDF. Harper cites a link to a blog in which the author points out some failings of RDF.