Ideas about easy finding of misspelled names

[Ton Spies]

Re: Ideas about easy finding of misspelled names

Steve,
Making the order of letters insignificant is really an eyeopener.
Thank you for sharing the idea. I will do some tests with it.
I do not the extras as you have like SSN and DOB. But the ZIP and address are also important comparison data to use.

Bill,
Of course other info is important and should be used to, but is not always passed by phone. So in practice people start with the name and are unaware of their deficiency and have to settle it.

G that's an impossible task and wortless as it also contains all synonyms to.

Just to clarify the question: Lastnames are correct in the file, but human interpretation on the spelling - passed by phone - makes it not easy to find them again. To find a way out is the mission.
My first thought about the possible use the soundex function was not feasible so I started this tread for a discussion about new and fresh ideas.


So please add your comments. You are welcome.

Ton

[Bill Griffin]

Re: Ideas about easy finding of misspelled names

Steve,
Making the order of letters insignificant is really an eyeopener.
Thank you for sharing the idea. I will do some tests with it.
I do not the extras as you have like SSN and DOB. But the ZIP and address are also important comparison data to use.

Bill,
Of course other info is important and should be used to, but is not always passed by phone. So in practice people start with the name and are unaware of their deficiency and have to settle it.

G that's an impossible task and wortless as it also contains all synonyms to.

Just to clarify the question: Lastnames are correct in the file, but human interpretation on the spelling - passed by phone - makes it not easy to find them again. To find a way out is the mission.
My first thought about the possible use the soundex function was not feasible so I started this tread for a discussion about new and fresh ideas.


So please add your comments. You are welcome.

Ton

Just so we understand the real problem here, what is the sequence of events when someone calls in? What type of business is this that only the name is supplied to the person answering the phone, or is that all they are trained to ask for? I am a little confused as to WHY the name is the most important piece of information and why more information cannot be supplied over the phone.

[G Gabriel]

Re: Ideas about easy finding of misspelled names

Not really sure I entirely understand the question. I interpret what you said to mean:
You have on your file someone named "Brown". Somebody calls in and gives the name of "Brown" but either the caller or the operator spells "Brown" as "Braun" and the operator starts looking for "Braun" and can't find it.

Is that it?

If it is, I will leave aside how is it possible that the caller does not know how to spell his own name to perhaps entail, as an explanation, a worldwide epidemic of mental retardation as a result of global warming, or perhaps the caller is not Mr. Brown himself but he is providing this information thinking that's how it is spelled.

Assuming further that we are still on the right question, have you looked at things like: like() or any of the functions that use wild cards?

[Ton Spies]

Re: Ideas about easy finding of misspelled names

I see that I am not clear enough about it.

I am thinking at an organization which is often very busy on the phone at specific times. When other people also help answer the telephone and write a messages this sometimes can cause a problem.

Because I think this is a common problem which occurs sometimes in any organisation I started this discussion.
Not to talk about the way organisations work, or should do, but to recognise a case like this. (G Don't make it too heavy).
I am open for any solution and like to discuss about the several ways there are to find a right person in an easy way (= with AlphaFive).

Recognise the case in the perspective of a user, think about a solution as programmer. Isn't that the way we always do?

Ton

[G Gabriel]

Re: Ideas about easy finding of misspelled names

Ton:
On the light side.. the scenario I mentioned is that what you are referring to? Where a caller mentions a name but not sure of the spelling and the operator does not know what exactly to look for?
Is that it?

[Ton Spies]

Re: Ideas about easy finding of misspelled names

Ton:
On the light side.. the scenario I mentioned is that what you are referring to? Where a caller mentions a name but not sure of the spelling and the operator does not know what exactly to look for?
Is that it?

Where a caller mentions his name and some one else writes it down as message and gives it to the person who has to find that caller in the file.
That person tries to find the name of the message.

[G Gabriel]

Re: Ideas about easy finding of misspelled names

Ton:
If the operator wrote down the wrong spelling, who is to say that whatever record you pull out by way of approximation is for sure the correct record?
There has to be some other way to confirm the identity of the record in question.. maybe tel #.. Account # etc. But you can't simply say a caller is asking for Shawanda and in your file Shuanda is the closet thing and therefore it is the correct one because it sounds like it or it is pronounced the same way.

Maybe what you should be looking for is a second and a third layer of identification rather than "best guess". Perhaps the operator should write down an account number, a phone number etc.

And once again, if the accuracy of the record is not so critical, as I mentioned earlier, you could use one of many alpha functions that use wild cards e.g. like() or something like contains() or $ etc.

[Steve Wood]

Re: Ideas about easy finding of misspelled names

My personal opinion is that all the discussion about how to properly guess how the name is spelled, is a fun exercise. The right approach is to ensure you know who you are talking to, and to match that up to a database record that already has their name properly spelled, and phonetically as well, if that is important. And, if a record cannot be found, one is added.

The only time I might disagree with my statement is if I was recording one-off conversations with random individuals, and never intended to create a permenent record of each client.

Fundamental to database design, the key field is NOT the persons name. It is their AccountNumber or similar. And furthurmore, the client's name is never entered ad hoc in to the notes, it just "client said this or that". Or, even if the name is in the notes, the client's official record is still the key when trying to refer to that person.

As an example, every single time I call any company where I have, or could have an account, they go through several steps to identify me, and if necessary to be absolutely sure I am who I say I am. At that point they have me right on their screen, they don't go to step 2 until that is done. That may take an extra 5-45 seconds, but overall it is more effecient than not knowing who I am. AND IT MAKES ME FEEL SAFE.

Thinking-out-of-the-box is all well and good, but it is also our job to tell the client how it has to work.

(I see I agree with Gabriel)

Where a caller mentions his name and some one else writes it down as message and gives it to the person who has to find that caller in the file. That person tries to find the name of the message.

Unless they are a one-person company, they will be out of business soon. At a minimum, the message taker has to be trained to ask, "...and what is your phone number?"

[Bill Griffin]

Re: Ideas about easy finding of misspelled names

A person who took messages that no one could decipher or had incorrect information on them would not last very long in our company :) (unless it was the owner, and if that is the case - the company is in big trouble)

[Ton Spies]

Re: Ideas about easy finding of misspelled names

Let it be completely clear that I share your opinion and views.

Because we do not currently see any more opportunities other than all that are put forward, we just arrived on a different track than discussion about technical solutions.

In summary, there are a number of possibilities.

1) The Soundex function.
This feature is recently discussed in a previous thread.
The use is for many people reasonably well acceptable even though there is no parameter to set.

2) Comparing the first n letters of a name in alphabetical order.
Here you can express the matching characters in a rate and calculate a scoring opportunity ..

3) Create (a part of) the name for a searchstring to use in the functions $ or Like ()

Some people have forwarded a script as a solution of their specific needs.
As they can be adjusted as desired, worth to investigate.


If you have a new idea, feel free to add it.


Thank you very much for your contributions

Ton Spies