Identifying Subjects With Facets

If facets are aspects of subjects, then for every group of facets, I am identifying the subject that has those facets.

If I have the facets, height, weight, sex, age, street address, city, state, country, email address, then at the outset, my subject is the subject that has all those characteristics, with whatever value.

We could call that subject: people.

Not the way I usually think about it but follow the thought out a bit further.

For each facet where I specify a value, the subject identified by the resulting value set is both different from the starting subject and, more importantly, has a smaller set of members in the data set.

Members that make up the collective that is the subject we have identified.

Assume we have narrowed the set of people down to a group subject that has ten members.

Then, we select merge from our application and it merges these ten members.

Sounds damned odd, to merge what we know are different subjects?

What if by merging those different members we can now find these different individuals have a parent association with the same children?

Or have a contact relationship with a phone number associated with an individual or group of interest?

Robust topic map applications will offer users the ability to navigate and explore subject identities.

Subject identities that may not always be the ones you expect.

We don’t live in a canned world. Does your semantic software?

8 Responses to “Identifying Subjects With Facets”

  1. […] Another Word For It Patrick Durusau on Topic Maps and Semantic Diversity « Identifying Subjects With Facets […]

  2. […] failed to mention semantically equivalent facets in either Identifying Subjects With Facets or Facets and “Undoable” […]

  3. If two Topic Maps subjects have the same subject identity they are supposed to be merged, right? Other aspects with the topic are disregarded and only the identfier matter. So how is this supposed to work with the current Topic Maps ISO spec?

  4. Patrick Durusau says:


    Using the TMDM legend, the equivalent would be having a subject that is a group, say the local fraternity and members of that fraternity treated as role playing “members” of that group. Which is one way to model it.

    I was suggesting that facets can be used to identity group subjects, which don’t have explicit treatment subjects even though clearly they are.

    I am not sure where you are reading that the presence of equivalent subjectIdentifier properties mean that topic merge and all other aspects of a topic are ignored?

    First, nothing prevents a topic from having both a subjectIdentifier and a subjectLocator (indeed, multiple ones of both types) and merging on more than one of them. Odd probably given most of the TMDM legend topic maps I have seen but nothing in the 13250-2 says otherwise.

    Second, 13250-2 also allows merging to occur on conditions other than the ones listed.

    Third, I am not familiar with the “if merge on subjectIdentifier and ignore other bases for merging” rule. That may be how some software is written but it isn’t how 13250-2 reads.

    Finally, I was positing the treatment of all the members of the group subject as instances of the group rather than members. Hmmm, if you think of bits of a substance, say iron filings, for some purposes we can treat them as individuals and for others we can treat them all simply as “iron filings.”

    My point in the initial post was that we mistake people as always being individuals. For some purposes, dating services, yes (maybe), for other purposes, predicting shopping behavior, no.

    BTW, having the same subject identity isn’t a basis for merging. That representatives for subjects have specified properties which are defined as being the basis for merging, is the reason merging occurs. Subject identity may or may not have anything to do with the merging process. Supposed to but there are no guarantees.

  5. Ok, so it is context dependent, perhaps it should be called “Pragmatics” rather than Facets? Pragmatics and Topic Maps, perhaps?

  6. Patrick Durusau says:

    Sorry? Not sure what you are saying is “context dependent?”

  7. I may be way off since English is not my native language, but I meant “context” as used in Linquistics as semantics is part of this science.

    -“(linguistics) the text in which a word or passage appears and which helps ascertain its meaning”


    , and “Pragmatics” in a Linquistics meaning:

    -“Pragmatics is a subfield of linguistics which studies the ways in which *context contributes to meaning*”


  8. Patrick Durusau says:

    Ah, I think what I meant was slightly different. Let me try to re-say it and you tell me if it makes any sense:

    1) Facets (as I used it above): Aspects/properties that identify a subject that isn’t explicit in a navigation system. That is as I “choose” facets, some group subject is being identified, even if I don’t think about it that way.

    Hmmm, consider it to be further specification of membership in a set and as I “choose” more facets, the set gets smaller. (Could get larger but generally the purpose is to get a more manageable result set.)

    2) Pragmatics: “context” in the sense of occurrence in a group of words, also known as a phrase, sentence, etc.

    I wasn’t thinking of a navigation system, i.e., the facets, as a “context.” I can see how an argument could be made for a navigation system providing a “context” but that isn’t how I usually understand the term.

    I usually hear it as being closer to your “pragmatics” but now that you point it out, it could be quite broader.

    Think of a group of journals in a discipline as a “context” for example. We interpret all the terms/subjects in that “context” quite differently than if we were another discipline.

    Apologies for the length but I wasn’t being clear in my earlier post.