Database Random Name Generator

[jonathan_phang]

Hello,

quick one here - for those of you more versed in data anonymisation, just wondering if you know of a place where I can generate a large list of random names. I need to rename 8046 people, which are split by gender, so would like to specifically generate a 2 csv lists of x length.

I can then just write a quick C# program to parse each pair and update the names in the DB.

However, most places I find let me do about 30 at a time. Any ideas or better ways to achieve this?

Thanks in advance

JP

[chadders]

Probably missing the point but could you not take the 30 @ time names (first & last) whack them in two tables and cross-join (ie 30 first names x 30 last names = 900 first & last name combinations)?

edit: you'd only need 90 first and last names to get 8,100 different names - does it matter if you get every combination of "Smith"... John Smith, Mary Smith, etc...?

[scaryjim]

Can you not use C# to access the database and SHA1 hash a concatenation of the first and last names? That'll give you psuedonymised data (as long as every person the the database has a different name)! Or do you need something that looks like an actual name?

If you're not sure about generating a SHA1 hash from a String in C# I can post some sample code I've written for work (my work receives ~ 1.2million personal sensitive data records a month, so pseudonimsation of data is something we are quite familiar with ).

[jonathan_phang]

@chadders - yeah, ended up doing that temporarily and it worked fine... Its only a temp fix for now, but better than the character scrambled script that we had before. Might look into a better way for larger sets.

@Jim, unfortunately, the whole point I even got asked about this is because what we had meant that the names were nonsense. Mind you though, if you dont mind, I would be interested in taking a look at the hash method you use, if only to see how others do things!

Thanks for the input chaps

[scaryjim]

No worries. Generating a SHA1 is easy in .NET as the libraries *should* be included with the default SDK.

Use System.Text.Encoding.UTF8.GetBytes(String) to return a Byte array based on the name - you can just concatenate the first and last names, but we use an identifier comprising first and last initial, DOB and sex (M or F).

Use System.Security.Crypotgraphy.SHA1.Create().ComputeHash(Byte[]) - this creates a default SHA1 cipher then uses it to create the hash based on the Byte array you generated in the first step.

Then, step through the array building a String (or using a StringBuilder ) by using the .ToString("X2") method of the Byte struct, which converts the byte into a 2 character Hex string.

Because the hash of the message will always be the same, you can then use the hash as an identifier in the database without storing personally identifiable data, and match it to other datasets which can generate a hash of the same identifier. It allows data matching across sensitive data sources (e.g. matching police data to health data) without disclosing any personally identifiable information, which is important for data protection (the hash cannot be algorithmically reversed - although with sufficient computation power and knowledge of the structure of the data that was hashed you *could* determine the original message by regenerating the hash). By the sound of it it's not relevant to the work you're currently doing, but if you ever get deeper into data anonymisation you may find it useful...