A Brief Review of the Term “Entity” - from Plato to the Semantic Web

In my previous post I looked at the way in which the term "data model" is used, and the confusion it can create.  Today, I am focusing on a very common term: "Entity".  We occasionally come across references about its usage in the data modeling literature, where "entity" is distinguished from "entity type".  The question then arises as to whether this is just useless nit-picking of terminology or if there is something we need to be aware of - something that perhaps can trip us up.

So what is an entity?  It seems like a silly question to ask a data modeler.  In one sense it is a box that appears in a data model created with a data modeling tool like ERwin that conventionally represents entities as boxes.  However, that is only a way we choose to represent entities when we model them - not an entity itself.

If we look at the way the term "entity" developed over time, we find that there is a dissonance between the community that first used it - philosophers - and the way data modelers use it. 

In philosophy, the term "entity" means an individual thing in the "real world" (whatever that means).  It corresponds to the term "instance" that is used in data modeling.  The problem with data modelers, and everyone in data management for that matter, is that when they think of "instance" they often think of a record that represents the instance rather than the instance itself.  But in philosophy, "entity" means the real individual thing, like the cup I am drinking my coffee from right now, or the laptop that I am using to write this blog post.

For data modelers, "entity" does not refer to an individual thing, but to a type.  We should really use the term "entity type", and indeed, authors such as David Hay and Larry English (and others), have carefully pointed this out in their writing.  However, commonly used terms tend to get shortened, and this included technical terms.  I think that this is why "entity type" has ended up as "entity" in the data modeling community. 

However, it can be confusing when we read books on logic or philosophy to find the term "entity" appearing repeatedly, and not meaning the same thing as "entity" means to data modelers.  And it is quite important.  Plato, for instance, was very interested in entity types, which he called "forms" and which he and his followers thought had a kind of real existence somewhere in the universe.  The writings of the Scholastics are filled with references to "entities" but they really mean "individual things". 

Well, it is all very interesting - but so what?  I think that not appreciating the way "entity" can be used to refer to "type" and "instance" will trip us up when we come into contact with philosophers.  Unlikely you say?  I disagree.  The semantic community, and particularly the ontologists, are heavily influenced by philosophy.  Some (perhaps many), like Barry Smith of Buffalo University are former philosophers.  Their tradition is one of using the term "entity" in the way the philosophers have used it, and not as data modelers do.   The future of data management is going to be heavily influenced by semantics and ontology, so it is important that we understand where these people are coming from, and this includes their usage of "entity".