Software Sorts Electronic Evidence
securitas writes: "The New York Times has a very interesting article about the legal industry using new search software to sort through electronic evidence such as e-mail, documents and recovered files, and the process that they go through to make the evidence usable. It has spawned an industry."
Lawyers discover grep?
Wasn't sniffing through e-mail done by something called Carnivore, which everyone here bitched about as an invasion of privacy? Bleh, be consistent about this stuff, if you don't want something looking through your e-mail, stick with that opinion.
Here's the correct link for those who'd like to register before they read the article.
Yours Sincerely, Michael.
when lawyers attourneys and judges have been replaced by computer programs.
Lawyer Eliza: (Walks up to the witness) So how are you today?
Witness: I am fine thank you.
L Eliza: How long have you been fine thank i?
Witness: I don't understand the question...
L Eliza: Don't you really understand the question?
Witness: That's right.
L Eliza: Is it really that right?
Prosecutor Eliza: Objection Your Honor! She is harassing the witness!
Judge Eliza: Why are you concerned about my honour she is harassing the witness?
and so on...
Will work for bandwidth
I know the tendency for unix enthusiasts to believe that you invented every useful technology in the 1970's, but it's simply not the case. Grep simply isn't suitable for interactive searches of gigabytes of data broken up into millions of files.
To efficiently work with several years worth email, more advanced techniques are required. Specifically, you need a text indexing program tied to a relational database. While this doesn't give you any more power than recursive searches using the grep and find combo, it's much much faster as your keyword and message attributes can use b-tree index lookups and a cost based optimizer to reduce disk reads.
That being said, it's still not that impressive of software. I'm certain that I could build the search component in a couple of weeks using Microsoft SQL Server (with the neat full text indexing feature) and a moderately adept gui developer could hammer out a decent interface in the same amount of time.
Still, there's a difference between "trivial to implement with any decent rdmbs" and "I can do it with a 2 line bash script". You would do well to remember it.
Your friend,
--Shoeboy