Computerised authorship attribution
- From: Matt B <mattb333@xxxxxxxxxxxxx>
- Date: Tue, 24 Jan 2006 17:55:51 GMT
Hi all,
As part of my computing course, I'm currently working on a project in
authorship attribution.
The aim is to create a software tool which, when given two candidate
authors and a large set of training text for each author, can analyse
a document and attribute it to one or other of the authors, based on
the supplied training text of their existing works.
I've reviewed the literature and found a few techniques that have been
trialled in previous authorship attribution studies: K-Nearest
Neighbour, Naive Bayes, CUSUM etc...
Now, the crux of the matter is that my background is in computing, so
whilst I can program capably, I have very little understanding of
statistics or statistical techniques.
The techniques I've looked at so far, all seem to be beyond my grasp
(I don't have a mathematical brain sadly). Actually - I think I
understand CUSUM, but from what I've read, it's not much cop.
I was wondering if anyone knows of a technique that I could possibly
use? It must be:
A: Able to be programmed into a computer
B. Able to be understood by a simple person
C. Potentially capable of attributing authorship of a document when
given two possible authors
Appreciate any advice! Cheers -
Matt
.
- Follow-Ups:
- Re: Computerised authorship attribution
- From: Ross Clement (Email address invalid - do not use)
- Re: Computerised authorship attribution
- Prev by Date: estimating the inverse covariance matrix
- Next by Date: Re: estimating the inverse covariance matrix
- Previous by thread: estimating the inverse covariance matrix
- Next by thread: Re: Computerised authorship attribution
- Index(es):
Relevant Pages
|
