-
Chris Webb authored
examples/test trains dictionaries on read and junk mail between 100 and 800 days old, then reports classification statistics for read and junk mail less than 100 days old. For my mail at the time of writing, this yields a training set of around 10k read messages and 50k junk messages, with a test set of around 1k read messages and 10k junk messages. examples/whitelist/generate creates a whitelist using the From: headers of read mail less than 800 days old, filtering out my own addresses as spam is often sent with a From: address to match the recipient. examples/whitelist/deliver demonstrates how this automatically generated whitelist can be integrated into a delivery filter.
Chris Webb authoredexamples/test trains dictionaries on read and junk mail between 100 and 800 days old, then reports classification statistics for read and junk mail less than 100 days old. For my mail at the time of writing, this yields a training set of around 10k read messages and 50k junk messages, with a test set of around 1k read messages and 10k junk messages. examples/whitelist/generate creates a whitelist using the From: headers of read mail less than 800 days old, filtering out my own addresses as spam is often sent with a From: address to match the recipient. examples/whitelist/deliver demonstrates how this automatically generated whitelist can be integrated into a delivery filter.
Loading