Form Data Validation

knashash

Hello...im wondering if anyone has a good way of validating bogus data people submit....specifically names and addresses.

for example:

name: asd asfd

address: 123 fake st.

I know this stuff must be hard to catch, but there has to be a good script out there somewhere.

Thanks All

drawmack

you can buy the national addresses database from the usps and then verify that their address is in that, but that would exclude foreign customers.

TheDefender

I don't know how you could really validate a name per se (aside from names starting with numbers or names that have punctuation, except apostrophes, in them.)

I mean, to you the name Shrek may not seem like a real name. I mean, come on, who would ever name their child Shrek... Who's to say asdf is or isn't a real name? (especially if dealing with a foreign name)

I mean, you could probably write a filter to look for certain Keyboard patterns like Qwerty or Qaz, but you could inadvertently exclude someone based on a typo, or a really cruel parent...

As far as the address is concerned, again, who's to say 123 fake st. isn't a real address. A human may be able to discern it, but you're talking about a computer discerning things, and unless you have an incredibly smart AI, that's taking on too much of a risk to lose clients, customers, etc... If someone was submitting fake info to you, would it really be worth your time to pursue them anyhow?

The bottom line is that I think it is a bit of overkill... If someone wants you to know who they are, or if you NEED to contact them, they'll get the clue when they never hear from you since they provided bogus info anyhow, right?

jam_wil

well said... well said

knashash

Yes you are definatley correct there.....I do however would like to get some sort of percentage of people who are submitting valid data to me.....I am running a business and would like to use that for reporting purposes.

I guess ill just have to come up with some basic rules to help filter it out.....thanks for the replys though

TheDefender

Unfortunately, it all goes back to the point I made earlier, and that is; How do you determine if it really is valid or not? I mean, I don't personally know anyone named Qwerty, but I wouldn't put it past some cybergeek lucky enough to have sex once, and accidentally get some poor girl pregnant, to name their child something like that. So I think that even your filtering methods would have to be too lenient for it to be effective.

Same thing with filtering addresses... Most addresses are some number (123) some street (fake st.) so you could filter words out at the beginning to keep people from putting something like five five five any street... OK, but Dell's Address in Nashville, TN is One Dell Parkway, not 1 Dell Parkway, so someone putting an address like that will get flagged. So again, your filter would have to be too lenient to be effective...

I just don't see there being a very good solution to this. You kind of have to rely on the old human brain to discern it, because remember, computers and scripts are stupid things... You have to tell it what to do, and how to do it, so just cut out the middle man.... er machine, that is.

knashash

Yeah I hear ya man...although with about 15,000 records a day to look at I dont think cutting out the middle man is an option...I appreciate your input though

TheDefender

Good luck with it... If you figure out a good method, I would definitely be interested in taking a look... I've tried several times in the past, and just can't put my brain around a solid method without being too exclusionary.

rachel2004

Originally posted by TheDefender
...but I wouldn't put it past some cybergeek lucky enough to have sex once...

...with a girl or by himself?😉 Depending on how you view it, yeah, I guess a cybergeek could be considered "getting lucky" while the other could be considered "a sure thing".😉

knashash: you may want to check through phpclasses.org. I don't know if they'll have what you're looking for, but they are generally pretty resourceful and who knows... you may be getting luck--okay, I won't go there.😃

drawmack

Defender made two invalid points
1) names can, and do, contain punctuation. For example most places I go ask for my first name, but I don't go by my first name I go by my middle name so in this case my first name is J. Thomas not J Thomas. I have a friend named Ashley-Kay as well. Put rules into place that don't let us sign up with those names and you just lost 2 customers.

2) Addresses can start with letters when you're dealing with the country. My mailing address is P.O. Box and my physical address is Lot 3

TheDefender

Invalid points???? Drawmack, I think your points emphasized my points precisely... and that it's virtually impossible to script a filter for this kind of stuff, without some heavy and smart AI that just doesn't exist.

{sitting in corner, arms crossed, glum look on the face, sobbing, hoping my peers didn't misread my points }:glare:

rachel2004

My best suggestion for filtering for valid data would be (and this is only for United States based addresses):

is the zip code valid (meaning, is it 5 digits with a possible 4 digit suffix)? If yes, continue.
is the zip code in an American state or FPO, APO, etc. that would contain that zip code? If yes, continue.
does the street address contain a numeral? If no, send an e-mail to the administrator to use some "human" filtering on the address. Else, continue and considered the address "valid" since I can't think of any more checks one can perform for validity.

Of course, this is not foolproof. I could still be a visitor your site and enter my zip code as 54321 and say I live in New York state. However, your error checking would know that New York state contains the zip code ranges of xxxxxx-yyyyyy, whereas as we move westward through the continental United States, I believe the numbers get higher.

As far as checking for street addresses that contain (not necessarily begin with a number)... my suggestion is to have a human verify street addresses that do not contain any numbers NOT because I believe all street addresses should contain numbers (as the street address One Microsoft Way illustratres), but simply because most street addresses DO contain numbers (e.g. P.O. Box 1234, 123 Main Street, 1 Infinite Loop. Having a human check those few (relatively speaking) addresses that do not contain numbers in the street address is better than allowing all forms of street addresses through.