Linked. Open. Data.
As we get spun up for our Linked Ancient World Data Institute (LAWDI) and move toward more interlinkedness between Pleiades, Papyri.info, the Ancient World Image Bank and other resources, I've been refreshing my thinking on Linked Open Data (LOD). Among the fodder for this thinking is Tim Berners-Lee's "Linked Data - Design Issues." Today's Digital Humanities Project Directors meeting at NEH has John Muccigrosso and myself channeling Sebastian Heath in a lightning talk, so I thought I'd better get my thoughts in order. It falls to me to briefly define linked open data so John can talk about the institute.
Berners-Lee's piece presents two lists of criteria/considerations/characteristics of linked open data in a way that doesn't entirely reconcile the two. They're overlapping and complementary, but they clearly represent two different stages in the promotion of the idea. I think those lists can be remixed into a single list with three clear rubrics that are easy to remember.
- Put your content on the web (not under or inside or over it)
- Assign (stable) Uniform Resource Identifiers (URIs) as names to discrete entities and ideas in your content
- Use HTTP URIs to enable lookup of the names and hide implementation detail
- Provide useful information about named entities on lookup
- Link to other URIs to facilitate discovery
- Typed links define relationships between named entities both within and beyond your content
- Typed links facilitate assertion and inference
- Bottom line: my data is your metadata
- Publish under open licenses (e.g., Creative Commons Attribution, GNU Public License)
- Publish in non-proprietary formats
- Publish machine-readable content
- Facilitate discovery via browsing and crawling (i.e., don't require searching/guessing)
- Bottom line: my data is your data
- Publish structured data, not just documents
- Treat both content and links as data
- Express internal and external links in well-defined forms like RDF
- Bottom line: our machines can help