On Mon, May 29, 2000 at 03:09:03PM -0500, Ted Pedersen wrote:
>
> Name: Ted Pedersen
> Email: [EMAIL PROTECTED]
> homepage: http://www.d.umn.edu/~tpederse
> Preferred User ID: TPEDERSE
>
> Plannning to Contribute: A Naive Bayesian Classifier. This is a simple yet
> surprisingly effective supervised learning algorithm.
>
> The module will estimate the parameters of the Naive Bayesian
> model using training data that the user must supply - the module will
> perform n-fold cross validation and report accuracy,
> standard deviation, and confusion matrices for the classifications.
>
> This is a cleaned up version of code I have used in previously published
> studies that use Naive Bayesian classifiers for natural language processing
> problems. (Details buried in papers available at my web page). This module
> will be general purpose and useful for any domain where supervised learning
> is viable.
>
> I have not consulted other developers on this project as I searched around
> and did not find other implementations of Naive Bayesian classifiers (or any
> other machine learning algorithms for that matter) so I believe this will be
> a useful module.
>
> I think this would fit under an AI:: category pretty well.
> It could also go under the Decision:: category.
You might also consider it under Text::.
Peace,
* Kurt Starsinic ([EMAIL PROTECTED]) ---------- Senior Network Engineer *
| `There is more ado to interpret interpretations than to interpret |
| the things, and more books upon books than upon all other subjects; |
| we do nothing but comment upon one another.' - Michael de Montaigne |
