August 3, 2016 at 11:51 pm
Comments posted to this topic are about the item Proper Storage and Handling of Personal Names
August 4, 2016 at 4:59 am
Thank you for the article Tung. Are you one of the many "one named" people you refer to? 😉
I think the problem you refer to goes deeper than technical implementation. Although all cultures that I am aware of use "names" to identify people, they do it in different ways. And there is the heart of the problem - traditionally computer systems have a "Western European" (and I'll include the USA in that) and "anglo-centric" bias. That's fine if the people you are handling in your system are all from that culture but it can cause problems (which you describe) as you move further away from that.
It is very hard to devise a scheme for holding names which does not have a cultural bias, and if you do, then that may cause problems as well, because you lose meaning which is hidden in the sequence and relationship between the names. As well as the problems you refer to, Western Europe tends to put the "given name" first, and the "family name" later. Some other cultures naturally put the "family name" (if there is one) first. Some cultures (eg Russia) include a structured "father's name". Other cultures use middle names in a less structured way. It makes real, significant differences for grouping and sorting. People change their names with life events. The differences go on and on...
The question is: Does it matter? Sometimes it does and thoughtless imposition of norms from another culture cause inconvenience or even offence. Sometimes it doesn't. And sometimes the limited scope of the system means that it is simpler to accept a standard way of doing things and "force fit" the ones that don't match the norm (and you should find a way of apologising to those affected when you do!).
Whatever you do, it's hard and it deserves a little more thought at the early stages of specifying the system than it receives.
Tom Gillies LinkedIn Profilewww.DuhallowGreyGeek.com[/url]
August 4, 2016 at 5:03 am
1. The full_name column may not a practical one. It is brings more problem than profit, if you think about the possible sequences of the first_name and last_name it is obvious (and not just from that aspect).
2. I'm Hungarian and I have a very interesting problem. If a women is married on Hungary she can chose the 'married name'. This may her original name without any change (a) or her husband surname and her given name (b) or her original surname + her husband surname + her original given name (c) or her husband surname + her original surname + her original given name (c2) or (the most crazy) if her husband has a title (dr. etc) that too (C3) and if she has one it is too (C4) and from other variants and if these are not enough the order of the names is reverse (like the Korean or Japanese) i.e. surname given name the right order. (Example if these was too difficult: An ordinary Hungarian woman name is Nagy Katalin (the Nagy is the surname!), if she has a title (dr. what mean he is a physician) her name (in Hungarian order) dr. Nagy Katalin, if her husband name is dr. Kiss Endre she can chose the next alternatives: dr. Nagy Katalin, dr. Kiss Katalin, dr. Kissné Nagy Katalin, dr. Kissné dr. Nagy Katalin (!), dr. Kiss Endréné dr. Nagy Katalin (!). These are not the all variant but the main versions. :alien:) The problem (if it not obvious for now...): which is which? Which is the surname which is the given name, which is the middle name(s)? This is not only a problem when you in an English database but in a Hungarian too. (And I didn't mention the need of the Unicode...). If you want to ordering these kind of names it is not an easy task but even the storage is a big problem. (And I can write from the old Hungarian names, which is can more complex...:angry: ). 😉
August 4, 2016 at 5:14 am
I would not want to type the column prefix column names you suggested.
For an article thorough about a lot of things, the assumption that you would name columns that way takes away from your argument.
412-977-3526 call/text
August 4, 2016 at 5:19 am
Didn't notice the author's name till afterwards - looks like the article might be motivated by personal experience on sites that don't cater for people with only one name !
August 4, 2016 at 6:26 am
Allow me to propose an alternative. One of the primary reasons Western databases store names in multiple fields (given_name, surname, title, etc.) is to allow for the generation of formal and informal ways to address the person. Perhaps that's what we need: a simple "name" field, plus a formal_name and informal_name fields of sufficient length. This imposes additional burden at entry but should accommodate the different cultural aspects discussed in the article with the exception of the native language issue which would require duplicating all three fields to allow for native language versions. The only thing that is still missing (for Western cultures with multiple names) is a way to sort by surname, which is the norm most of the time. Thoughts?
Edit 1: added colon to make sentence more readable
Edit 2: added sort by surname sentence
August 4, 2016 at 6:48 am
That's definitely an interesting and useful suggestion. Tung has pointed out a real problem. The difficulty with dealing with it in the abstract is that different solutions work in different situations. The "formal" and "informal" name approach would work very well in many situations and across cultures. I'm not sure it would necessarily give you the "grouping" that you get with Western family names but that may not matter.
I've been thinking about this problem, and I would start with a number of questions: 1) What is the geographic/cultural scope of the proposed system/database? - If we are confining ourselves to one culture, then our job is simpler. 2) How different are the various cultures concerned? 3) What are the numbers/proportions involved? 4) How much does it matter to the people/business involved? - Might it cause grave offence or do they not care. 5) Is there a political aspect to this? - "Colonial legacy" or minority ethnic groups 6) Are there features of names that we want to use that we haven't been thinking about - Like grouping with family names. It goes on and on...
...the thing is, this can either be important, or it can be a complete waste of time. We need to decide if it matters, how much it matters and how much effort we want to put into it. We can only really do that for a particular case. I'm not sure if it is possible to come up with a implementable "general" solution.
Tom Gillies LinkedIn Profilewww.DuhallowGreyGeek.com[/url]
August 4, 2016 at 6:48 am
We have a similar structure as your original example. I could see where some slight improvements could help. We have a computed column for the full name.
August 4, 2016 at 6:52 am
szeitzp (8/4/2016)
1. The full_name column may not a practical one. It is brings more problem than profit, if you think about the possible sequences of the first_name and last_name it is obvious (and not just from that aspect).
We would just put whatever name is to be displayed first in the first name column and whatever name is to be displayed last in the last name column. We don't really care whether or not it is the surname or not for either column. We construct the full name in the same manner so it always displays the way someone wants it displayed. We would have a bit of a problem on a long name though. Fortunately for our measly million names, we don't have any that have reached the max length.
August 4, 2016 at 7:20 am
Just an aside to this, under data protection laws (at least here in the UK), personal data should be stored encrypted, so if that is the case, it really doesn't matter at the DB level as it would ideally be salt and hashed.
August 4, 2016 at 7:26 am
Knowing my colleagues they would suggest that the database should not be used like this, but may be suitable to have a column that stores JSON for the names. :rolleyes:
JSON solves everything
August 4, 2016 at 7:28 am
We would just put whatever name is to be displayed first in the first name column and whatever name is to be displayed last in the last name column. We don't really care whether or not it is the surname or not for either column. We construct the full name in the same manner so it always displays the way someone wants it displayed. We would have a bit of a problem on a long name though. Fortunately for our measly million names, we don't have any that have reached the max length.
Yes, sometimes it is the right decision.. but sometimes not. An example: if you write references from an article into a scientific publication you must write the family name first and after this the given name, but if it is the wrong order, you must put a comma between the two names, but if it is the right order not. It seemingly a little problem, but if you manage hundreds or thousands of it the problem a little bigger. And this is not just an abstract example, see the reference and library managing programs.
August 4, 2016 at 7:44 am
BTW, that last INSERT should be:
INSERT INTO demo.Customers_5(
cst_full_name
, cst_full_name_native_language
, cst_short_name
, cst_short_name_native_language
)
VALUES ('Tyrion Lannister',N'Tyrion Lannister','Tyrion',N'Tyrion')
, ('Lee Jun-fan',N'???','Bruce',N'??')
, ('Sinéad O''Connor',N'Sinéad O''Connor','Sinéad',N'Sinéad')
, ('Cher',N'Cher','Cher',N'Cher')
, ('Nebuchadnezzar',N'????????','Nebuchadnezzar',N'????????')
, ('Gordon Matthew Thomas Sumner',N'Gordon Matthew Thomas Sumner','Sting',N'Sting')
, ('Shreya Ghoshal',N'??????? ?????','Shreya',N'???????')
, ('Smith',N'Smith','Sir',N'Sir')
August 4, 2016 at 7:46 am
Rick-153145 (8/4/2016)
Just an aside to this, under data protection laws (at least here in the UK), personal data should be stored encrypted, so if that is the case, it really doesn't matter at the DB level as it would ideally be salt and hashed.
PII data needs to be protected inappropriate use & dissemination which doesn't always mean that the data needs to be encrypted.
A business customer doesn't need their name encrypted, as it is not personal.
The name and delivery address for a customer doesn't need to be encrypted as how will the postman every deliver the item. What the customer buys may be deemed to be confidential and may warrant a plain packaging and the postman will not need to know the contents other than handling instructions.
August 4, 2016 at 8:15 am
Tom Gillies (8/4/2016)
Thank you for the article Tung. Are you one of the many "one named" people you refer to? 😉
Tom, yes indeed, my name is a mononym. I still think of the USA as a "melting pot" of many different cultures, made up almost entirely of immigrants, so we also need to deal with the way people are named in other countries and cultures. Years ago, many immigrants made changes to their name to make them more anglicized, and some mononymous folks have added a name to make their name "fit". These days, I don't think those kinds of modification are quite as common.
I think the problem you refer to goes deeper than technical implementation. Although all cultures that I am aware of use "names" to identify people, they do it in different ways. And there is the heart of the problem - traditionally computer systems have a "Western European" (and I'll include the USA in that) and "anglo-centric" bias. That's fine if the people you are handling in your system are all from that culture but it can cause problems (which you describe) as you move further away from that.
It is very hard to devise a scheme for holding names which does not have a cultural bias, and if you do, then that may cause problems as well, because you lose meaning which is hidden in the sequence and relationship between the names. As well as the problems you refer to, Western Europe tends to put the "given name" first, and the "family name" later. Some other cultures naturally put the "family name" (if there is one) first. Some cultures (eg Russia) include a structured "father's name". Other cultures use middle names in a less structured way. It makes real, significant differences for grouping and sorting. People change their names with life events. The differences go on and on...
The question is: Does it matter? Sometimes it does and thoughtless imposition of norms from another culture cause inconvenience or even offence. Sometimes it doesn't. And sometimes the limited scope of the system means that it is simpler to accept a standard way of doing things and "force fit" the ones that don't match the norm (and you should find a way of apologising to those affected when you do!).
Whatever you do, it's hard and it deserves a little more thought at the early stages of specifying the system than it receives.
Viewing 15 posts - 1 through 15 (of 38 total)
You must be logged in to reply to this topic. Login to reply