Video Term Ontology


The word "ontology" has been acquired by the computer science fraternity to mean a graph which describes some specific thing. That graph may be a tree (a taxonomy) or it may be more and may have properties that allow it to be reasoned over and therefore calculate stuff.

Ontologies are considered the backbone of the semantic web and they can be represented in a language called the Resource Description Framework (RDF). There are many ways to write RDF, but XML is one of the (horrible) ways. Turtle, or N3, is much nicer. A program called Protege allows creation of simple ontologies relatively painlessly.

So, here I present a small ontology for describing videos and shots within videos.

As part of a project I was working on, I needed to analyse a video and classify shots within the video. The idea was to take that forward and classify sets of shots into scenes and types of programme. Anyway, the very first step required a means for describing shots. And that's what this ontology was used for.

It contains information such as the shot type (e.g. long shot, mid-shot, closeup), whether the frame contains people and who they might be, whether the frame is static or moving, and if so how it's moving, etc.

The concepts are all fairly simple concepts, bar the protozoom concept which is the union of two other concepts (zoom and dolly).