Question of id-entity
November 30th, 2002 by jiri
Reading discussions on the definition of words “trust” and “identity” under a DIDW article one cannot miss ambiguity of definitions. Topically, I have come across a text, that illustrates similar problems only with the use of the word “entity”:
“Having surveyed various definitions of “entity” in the literature, Reingruber and Gregory (p. 64) extract the following themes:
An entity is a `thing’ or `concept’ or `object’. . . . Or it may represent the fact that we need to capture information specific to only a subset of the instances of an entity. . . .
An entity is not a single “thing,” but rather a representation of like or similar things that share characteristics or properties. For example, King Lear and Hamlet are both plays, and have properties such as name, author, cast of characters, and lines of verse. The entity describing these might be PLAY, with King Lear and Hamlet as examples or instances or occurrences of PLAY. . . .
Entities involve information. The “things” that entities represent are things about which the enterprise wants or needs to retain. . . .
These sentences shift repeatedly between treating entities as things in the world and treating entities as representations of things in the world. Their choice of example facilitates the confusion, given that the word “play” refers to both the text and the performance, the representation and the thing represented. Reflecting on the above points, they “refine our view of entities” as follows:
Entities are collections of attributes that satisfy a particular set of rules established for relational modeling. (ibid., p. 64)
This definition is perfectly ambiguous. Real things can, with some effort, be viewed figuratively as collections of attributes, but entity-relationship data records are precisely collections of data elements called attributes. These authors may have been misled by the practice, common but usually well defined in the literature, of using the word “entity” to refer both to categories of things (cars, people, companies) and to instances of those categories (my old 240Z, Jacques Martin, IBM). Be this as it may, the conflation of representation and reality is a common and significant feature of the whole computer science literature.”
We won’t get much far if we don’t get the language right. This, however, means starting from description of various scenarios illustrating what trust or identity mean in various real world scenarios, before trying to come up with abstractions. Not the other way around. In this respect, I find work of Roger Clarke more useful.