Datatypes R2 Issue 101

Clarification of Name Semantics

Skype

[11:12:09 AM] Grahame Grieve says: From the abstract Spec: Two name values are considered equal if both conatain the same name parts [11:12:41 AM] Grahame Grieve says: From the ISO Spec that was balloted: Note for balloters: is it right that this is independent of ordering? and do the qualifiers count? [11:12:49 AM] Grahame Grieve says: From Lloyd's response: [11:13:22 AM] Grahame Grieve says: Ordering matters in comparison, as do asserted qualifiers. However, missing qualifiers should match. E.g. If one name declares the qualifier 'Initial' and the other specifies no qualifiers, that should match. [11:13:34 AM] Grahame Grieve says: From NHS: think that it is correct that this is free of ordering. The ordering will be determined by the system. For example one system might want to display Ian Townend whereas another may wish to use Townend, Ian [11:14:00 AM] Grahame Grieve says: we have agree that order matters on first names, so I think that equality has to matter in this regard [11:14:26 AM] Grahame Grieve says: I think Lloyd is wrong about qualifiers - either they matter or they don't [11:16:13 AM] Grahame Grieve says: Getting ahead of an argument Gunther and I are about to have, equality has to be commutative, and Lloyd's logic would make it not so [12:13:40 PM] Gunther Schadow says: it was a holiday, Memorial day. But Lloyd is on vacation and Dale and Lee are not always on anyway. [12:20:02 PM] Gunther Schadow says: Since a name is a LIST of ENXP one would think the order would matter. What are the real use cases for equality for names? I think it's mostly about set membership and how you know if something was changed. I don't think that "matching" should be the issue, because name matching is going to work very differently. I don't feel strongly about this, as long as equal is symmetric, reflexive and transitive and as long as it is not anal about things that make no semantic difference. But what is semantics with names anyway? [12:21:32 PM] Grahame Grieve says: actually, perhaps the most compelling argument for nailing this down well comes from updateMode. [12:21:53 PM] Grahame Grieve says: which is a variation on the set membership theme, I guess [11:16:26 PM] Gunther Schadow says: agree [11:17:03 PM] Grahame Grieve says: so has to be symmetric. so either qualifiers matter or they do not [11:17:30 PM] Grahame Grieve says: I think they matter. And I think that order has to matter, since John Patrick Morgan is not the same as Patrick John Morgan [11:17:35 PM] Gunther Schadow says: right. It could even be as detailed as some matter some don't. [11:17:37 PM] Grahame Grieve says: I don [11:17:46 PM] Gunther Schadow says: right. [11:18:01 PM] Grahame Grieve says: I don't think that order matters for AD though [11:18:05 PM] Grahame Grieve says: even though it's a list [11:18:11 PM] Gunther Schadow says: With EN it's probably very straight forward. [11:18:59 PM] Gunther Schadow says: AD streetAddressLine is intended to repeat, so, order does matter there. But who cares if the city comes first or last, that is a matter of style (i.e., Japan first, US, Europe last.) [11:19:09 PM] Gunther Schadow says: But that is similar with names. [11:19:25 PM] Gunther Schadow says: Only given is at issue (Patrick Morgan vs. Morgan Patrick) [11:20:01 PM] Grahame Grieve says: the real problem with AD is that we have never documented what is meant to repeat [11:20:07 PM] Gunther Schadow says: So, we could say, order only matters inside groups of elements with same type. [11:20:14 PM] Grahame Grieve says: and even with AD, so what if address line repeats [11:20:29 PM] Grahame Grieve says: Immunology Department 65 Victoria Parade [11:20:35 PM] Grahame Grieve says: 65 Victoria Parade Immunology Department [11:20:39 PM] Grahame Grieve says: so who cares. [11:21:00 PM] Grahame Grieve says: so I think that AD rules are correct [11:21:17 PM] Gunther Schadow says: noone. And noone should care about Morgan, John Patrick vs. John Patrick Morgan. [11:21:29 PM] Gunther Schadow says: Only about John Pat vs. Pat John. [11:21:35 PM] Grahame Grieve says: but EN, ok, the rule you propose is probably correct - only order within a given type [11:21:54 PM] Gunther Schadow says: There is more. What about delimiters? [11:22:31 PM] Gunther Schadow says: Is Patrick-John and Patrick John the same? [11:22:38 PM] Grahame Grieve says: ho [11:22:40 PM] Gunther Schadow says: not really [11:22:42 PM] Gunther Schadow says: ho no [11:22:44 PM] Gunther Schadow says: :) [11:23:00 PM] Grahame Grieve says: that's nasty [11:23:30 PM] Gunther Schadow says: But, "Morgan, John Patrick" vs. "Morgan John Patrick" should be the same... [11:23:32 PM] Gunther Schadow says: ... or not. [11:23:38 PM] Gunther Schadow says: Who cares? [11:23:43 PM] Grahame Grieve says: in particular because some would go PatrickJohn and others would do "Patrick-John: [11:23:45 PM] Grahame Grieve says: in particular because some would go PatrickJohn and others would do "Patrick-John" [11:23:57 PM] Grahame Grieve says: and this would actually vary by clerical staff in some systems [11:24:11 PM] Gunther Schadow says: yes. [11:24:25 PM] Gunther Schadow says: One should make it a rule that - is always a delimiter? [11:24:37 PM] Gunther Schadow says: But there are rules what to do with -, I think. [11:24:44 PM] Gunther Schadow says: That's when it comes to initials. [11:25:06 PM] Gunther Schadow says: Eva-Maria Schadow (my mom) would be "E. Schadow", not "E. M. Schadow" [11:25:20 PM] Gunther Schadow says: Eva Maria Schadow would be "E. M. Schadow" [11:25:35 PM] Gunther Schadow says: the - joins two first names into one. [11:25:42 PM] Gunther Schadow says: - is different from space. [11:26:57 PM] Gunther Schadow says: This is a complicated subject, we may want to defer until we have time? [11:27:15 PM] Grahame Grieve says: huh [11:27:31 PM] Gunther Schadow says: uhu [11:27:36 PM] Grahame Grieve says: yes. defer. But agree with you so far. the trouble is that though it is complicated, it's rather important [11:27:48 PM] Gunther Schadow says: Ah! I agree. [11:28:15 PM] Gunther Schadow says: As with most things. If we don't figure them out, every little Programmer out there needs to figure it out. [11:28:19 PM] Gunther Schadow says: by themselves. [11:28:35 PM] Grahame Grieve says: yes [11:29:04 PM] Gunther Schadow says: OTOH, we might find a way by just setting the scope right. Again, what is the purpose in this case for doing anything else than exact pairwise equality of all pieces? [11:29:31 PM] Gunther Schadow says: This is not for MPI queries. [11:29:37 PM] Grahame Grieve says: but it is [11:29:43 PM] Gunther Schadow says: is it? [11:29:58 PM] Grahame Grieve says: yes. There's no other way to do names. EN covers the whole gamut [11:30:27 PM] Grahame Grieve says: oh no - you mean for the the query parameters? [11:30:31 PM] Grahame Grieve says: not so sure. [11:30:45 PM] Grahame Grieve says: but it is for the responses [11:31:09 PM] Gunther Schadow says: Never mind query parameters, I have nothing to do with them. But still I think that this would be a different scope, beyond what EN.equal needs to do. On the first line, I only need to know if a new name comes in whether I already have this one (exactly). [11:31:55 PM] Gunther Schadow says: If anyone screws around with even an iota in qualifiers or delimiters, it is different. [11:32:09 PM] Gunther Schadow says: if they change the order, it's different. [11:32:12 PM] Grahame Grieve says: I think there is a wider scope here. users want to know that they can connect 2 systems by V3, and have the systems understand names. That seems an essential baseline for V3. Strictly, this is not about equals, but about a full understanding of the semantics, questions like the delimiters, which are questions that underly equals [11:33:17 PM] Gunther Schadow says: That's why the initial definition of equals was so parsimonious. [11:33:24 PM] Gunther Schadow says: But it was also too naive. [11:33:45 PM] Gunther Schadow says: If you really try to unravel it, it gets complicated. Need to go back to some of our sources. [11:34:28 PM] Gunther Schadow says: The BibTeX name rules (or any other bibliographic data base rules), and some European SSS (Strategic Short Study) on names. [11:34:52 PM] Gunther Schadow says: It may be too much work for now. [5/29/2007 11:51:02 PM] Charlie McCay says: for order we did discuss having an attribute to declare whether order was significant -- in particular for addresses -- where the data may have been collected as address parts and so with no order determined by the sender -- and in other cases where the address has been formated. [5/29/2007 11:54:00 PM] Charlie McCay says: this makes sense because the order will sometimes imply semantics (and so be significant), and sometimes will not -- only the sender, (or the implementation profile that the sender signs up to) can determine which is the case [8:51:49 AM] Grahame Grieve says: another comment in the ISO ballot: Are all combinations of EntityNamePartType and EntityName PartQualifier valid? This should be stated explicitly in the document [8:53:12 AM] Gunther Schadow says: Right, should be. Could be. Not sure which ones are invalid. [8:53:24 AM] Grahame Grieve says: just knocking up a table for discussion [8:55:30 AM] Grahame Grieve says: There's some goofy stuff in there. "Indicates that a prefix like "Dr." or a suffix like "M.D." or "Ph.D." is an academic title." is subsumed by "For organizations a suffix indicating the legal status, e.g., "Inc.", "Co.", "AG", "GmbH", "B.V." "S.A.", "Ltd." etc." [8:55:38 AM] Grahame Grieve says: we are not good at this vocab stuff. [8:56:32 AM] Grahame Grieve says: The name assumed from the partner in a marital relationship (hence the "M") - except that code is not M [8:56:54 AM] Grahame Grieve says: do we need a harmonisation proposal to fix that? or can someone just fix it as a technical correction [8:57:17 AM] Grahame Grieve says: And then this: A name that a person had shortly after being born. Usually for family names but may be used to mark given names at birth that may have changed later. [8:57:47 AM] Lee Coller says: My understanding is that it is a harmonization proposal that is a technical correction. [8:57:48 AM] Grahame Grieve says: stupid - the general meaning is the same, but the specific meaning is inverted in the two cases - this is current, and this is not current [8:58:14 AM] Grahame Grieve says: I can't be bothered making it as a harmonisation proposal. [8:59:03 AM] Grahame Grieve says: as far as the table goes, there's "can" and "expected". As in, we don't expect you to mark a given name with SP, but I cannot see why we would disallow it [9:02:00 AM] Grahame Grieve says: FAM GIV PFX SFX DEL LS x X AC X x NB X x PR x X VV X x AD X X X X BR X X X X SP X x x x CL x X x x IN X TITLE X X x = allowed, X = expected and allowed [9:02:07 AM] Grahame Grieve says: hope that table works for everyone [9:02:41 AM] Lee Coller says: x versus X? [9:03:03 AM] Gunther Schadow says: cool, except that I don't remeber all qualifiers :) [9:04:11 AM] Lee Coller says: Aren't we making assumptions about how LS is represented around the world? [9:04:56 AM] Lee Coller says: Why is AC allowed as a prefix but not as a suffix. Ph.D. is a suffix [9:05:49 AM] Grahame Grieve says: FAM GIV PFX SFX DEL Legal Status . x X Academix . X x Nobility . X x Professional . x X voorvegsel . X x Adopted . X X X X Birth . X X X X Spouse . X x x x Callme . x X x x Initial . X Title . X X x = allowed, X = expected and allowed [9:06:08 AM] Grahame Grieve says: AC is allowed as a SFX. I'll upgrade it to expected [9:06:42 AM] Gunther Schadow says: When you say "expected" it means that if you do that type then one of the expected ones should also be used, right? [9:07:52 AM] Grahame Grieve says: no, expected is probably the wrong term. It means that there's combinations that we allow in case and there's combinations that we expect to see sometimes. Not that you are expected to use the qualifier [9:08:05 AM] Grahame Grieve says: happy to use a different word, and would explain it in narrative anyway [9:08:46 AM] Gunther Schadow says: that's fine [9:08:49 AM] Gunther Schadow says: not sure how voorveugsel can be suffix? [9:08:52 AM] Lee Coller says: Sorry, my eyes put a 'dis' in from of the allowed on x [9:10:24 AM] Lee Coller says: I think this is OK too. [9:10:31 AM] Gunther Schadow says: I go back to http://aurora.rg.iupui.edu/v3dt/minutes19990224.html [9:10:48 AM] Gunther Schadow says: to figure out the sophistication of Voorvoegsels. [9:10:51 AM] Gunther Schadow says: (and more) [9:14:16 AM] Grahame Grieve says: I was guessing about vv. I will change it to prefix only [9:15:09 AM] Grahame Grieve says: and yes, I am guessing about the use of LS - I was guided by the definition, but as I pointed out, the definitions are a shocking mess [9:15:36 AM] Grahame Grieve says: but we are going back out to ballot on ISO, to ISO, so this will probably be a reasonable way to try out our assumptions [9:15:43 AM] Grahame Grieve says: then we can build them into R2 [9:16:33 AM] Grahame Grieve says: that's a good set of notes, btw. [9:17:02 AM] Grahame Grieve says: I've bookmarked them for future use. I've seen them before but lost track of them [9:17:39 AM] Gunther Schadow says: There are other things here, like mutual exclusion. Vorvoegsel is an opposite of Nobility (but is a minor distinction). [9:17:59 AM] Grahame Grieve says: should we note this? [9:19:37 AM] Gunther Schadow says: There is semantics in strong and weak prefixes. It was in there initially, but was sacrificed to show readyness to compromise. Of course, if you go back to the analysis http://aurora.rg.iupui.edu/v3dt/minutes19990224.html you will see that inversion and weak association in prefixes actually makes sense to consider. [9:20:46 AM] Grahame Grieve says: we've effectively got weak with title - the definitions are nearly the same [9:21:47 AM] Grahame Grieve says: I do not see what an inverted prefix is, other than a suffix? [9:24:28 AM] Gunther Schadow says: an inverted prefix is not a suffix. It is a prefix which just happened to be pushed backwards. [9:25:04 AM] Gunther Schadow says: Example: Eduard de l'Aigle --> Aigle, de la, Eduard.