Innovation = Breaking Conventions June 22, 2010Posted by Andre Vellino in Information retrieval, Knowledge Representation, Logic.
A few months ago I was listening to Beethoven’s Piano Sonatas and observing to myself that the genius of the great western composers (my favourites: Monteverdi, Bach, Mozart, Beethoven, Stravinsky, Shostakovitch, Messiaen) is that they all “broke the rules”. I have heard that said about visual artists as well.
In challenging the conventions of their time, they created something new, exciting, different and unexpected that nevertheless expressed a coherent musical thought, evoked specific moods and created a musical narrative.
I was thinking (analogically) that the same might be true in computer science. Consider for instance relational data modeling. The core assumption here is that there are “entities” (things) and “relations” (associations) between them. What if we broke the rules for representing data either in the Object model or in the conventional Entity-Relationship model?
Following suggestions by Peter Turney in an early blog post on relations and attributes consider what a “database” might look like if the fundamental constituent was the relation instead of the entity.
In the conventional model, E-R diagrams allow database developers to constrain the fields in each entity to have some kind of association (1-1, 1-many, etc.) with other fields in other entities. But suppose instead that the fundamental assertion was that a relationship be true between, say, “events”.
The idea I am entertaining is similar to what Donald Davidson sketches in his paper Mental Events. In this view, the ontological primitive is an “occurrence” in some domain or field – e.g. a bank transactions / address changes / christenings – in which the constituents of the event (people, banks etc.) are not the primitive “objects” that need to be modeled by the database. Rather, it is the events themselves that are at the hub.
This is an awfully sketchy idea, but that’s one of the nice things about blogging – you have the freedom to just put out these half-baked ideas out there and get some early feedback.