Project Advice

Project Advice

5. Linked Data

The Semantic web

Tim Berners-Lee – who, as we mentioned right at the beginning of this course, invented HTML, and thus the worldwide web – has proposed a five-tier grading system for data on the web.

Berners-Lee five-star grading system: 1 available on the web; 2 available as machine-readable structured data; 3 as before but non-proprietary; 4 as above but using open standards from W3C; 5 all the above plus linked data linked to other data

Congratulations. You can now produce XML data, which merits three stars on the Berners-Lee rating. If you wanted to take it to the next level you would have to add some linked data to your markup.

The idea here is that we should make data readable by machines, so that they can do all of the hard work for us. A web search would be able to return you all of the data about the 1st Duke of Wellington, without mixing up any information about places called Wellington, or the famous boot. Furthermore, the semantic web engine would pull in data from multiple sources, because they are linked together.

Using technologies such as RDF, you would mark up the First Duke of Wellington as a triple, containing a subject, predicate and object. By pointing this at an authority list, such as the Library of Congress’s list, as a Uniform Resource Indicator (URI). If everyone did this in their data, the theory goes, then we could unambiguously identify occurrences of the First Duke of Wellington.

You've probably heard of a similar thing, the semantic web. To the extent that there is a distinction, semantic web tends to be used of machine reasoning over the kind of linked data that we mentioned above.

The idea of the 'semantic web' has been kicking around for a while, and it’s fair to say that not much has happened yet. When it will take off is a good question. But by producing data that is semantically marked up you’ll be most of the way there already. We hope that this course has given you the information to start doing just that.