Daniel’s Blog

Following my previous post on “The Tribes of the Semantic Web“. There is another Semantic Tribe: The RDFization Tribe.

The RDFization(or RDFisation to be slightly more English English) Tribe. RDFization is quite a simple term to understand if you think about it, because “isation” means the process of turning into something, and RDF is RDF…. therefore RDFisation is turning something into RDF. This is a mechanical process, and therefore automatic.

RDFization will take unstructured or semistructured text and use Statistical and AI techniques (such as Natural Language Processing and Latent Semantic Indexing) in order to extract named entities (and relationships), decypher subjects and disambiguate naming.

The people that take this approach are not necessarily “Semantic Purists” (who also use AI Techniques) because the idea here is to extract data which can be linked with the relevant object (as a foreign key represents an object). With RDFization the clever stuff happens at the higher (semistructured or free text) level, rather than in the heart (data) level which the “Semantic Purists” deal with.

RDFization is theoretically the same as Alex Iskold’s “Top-Down Approach“, except I have two problems with the term “Top-Down Approach”.

1. “Top-Down”/”Bottom-up” is very one dimensional. There are lots of ways to encapsulate semantics into a data source.
2. “Top-Down”/”Bottom-up” is used in so many other aspects of computer science, and all instances of these terms tend to have different meanings or different perceptions.

The Semantic Web / Linked Data is clear, it is to provide meaning in any/every way possible whether that through annotation, fully structured descriptions, “RDFization” through AI/Statistical means or some other way.

More information about RDFization

If you do a search for RDFization on Kingsley Idehens blog, you get some very interesting blog posts about RDFization.