Re: Getting it OUT!

< Next Message | Back to archived message list | Previous Message >

Note: This message is from the outliners.com archive kindly provided by Dave Winer.

Outliners.com Message ID: 2039

Posted by thompson.chris
2004-07-10 21:20:16

Steve, Bayesian learning is just application of Bayes’ theorem on a large scale. The database is broken down into features (usually individual words or phrases), and any individual database item thus has an implicit vector representing the presence or absence of each of these features. Given probability estimates taken from a set of learning examples, we can apply Bayes’ theorem to any new feature vector to determine the probability of that vector being associated with a specific category. If some numerical threshold is satisfied (e.g. 90% probability this email is associated with the Jorgensen account), then the software would either present this as a suggested category or simply automatically assign that category to the item, depending on the user interface design.

—Chris

Back to archived message list