[ptsefton.com] | [CV & Bio]

RDF appears on my short-range radar


I wrote today about Zotero, which uses RDF metadata, in a way that I have not yet grasped.

RDF has been coming up a lot in our work on the RUBRIC project as well lately. It's been around for quite a while, but I have never had to deal with it, or felt that I needed it. Now I'm beginning to see how it can be used for descriptive metadata, like in Zotero and how it might be useful for constructing interfaces and authorization systems for institutional repositories. The bit I had never thought of before is to use RDF to describe ontologies such as subject classification or organization structures, and be able to build applications that understand the relationships between items based on these ontologies.

I'm using the word ontology in the computer-science sense sounds a bit odd to me in this context but that's how the RDF community talk. Here's part of what Wikipedia said about Ontology (computer science):

In both computer science and information science, an ontology is a data model that represents a set of concepts within a domain and the relationships between those concepts. It is used to reason about the objects within that domain.


Here's how it might work, in three steps:

  1. Repository manager defines an ontology for, say, the structure of a university, with all the faculties, schools, departments and so on. RDF does not have to be a tree-structure, so you can express complex relationships such as you might get when an institute has multiple parents.

  2. Repository manager defines metadata mappings between descriptive metadata 'this item belongs to the Maths Department' and RDF expressions that make the link between the item and the organization structure.

  3. The repository software finds and infers all the relationships inherent in the metadata and the university structure; it will know that a paper written by an author in Maths, is automatically part of the Faculty of Science, for example. This is the 'reason about' mentioned by Wikipedia; and it's the ability to do this reasoning, within reason, that is of value when using RDF.

    This means being able to broaden and narrow searches by department, control access across the university heirarchy and build browse-based indexes all automatically.

I think there will be a lot of RDF in the rest of my year 2007, trying now to get the various projects I'm involved in to come up with some demos of the three-step RDF-ization of an Institutional Repository.