Classifier¶
In the console: Settings › Spam Filter › Classifier
The filter that learns from the mail people mark as spam and not spam.
Spam Classifier¶
Model¶
The spam classifier model to use.
Settings of their own: see Spam classifier model. Name in the API: model.
Training¶
Min. Ham samples¶
Minimum number of ham samples required for training.
A number. Default: 100. Name in the API: minHamSamples.
Min. Spam samples¶
Minimum number of spam samples required for training.
A number. Default: 100. Name in the API: minSpamSamples.
Training frequency¶
Frequency to train the spam classifier.
A length of time. Default: 12h. Name in the API: trainFrequency.
Hold samples for¶
Duration to hold training samples for.
A length of time. Default: 180d. Name in the API: holdSamplesFor.
Reservoir Sampling¶
Capacity¶
The capacity of the training sample reservoir.
A number. Default: 1024. Name in the API: reservoirCapacity.
Auto-learn¶
RBL hits¶
Number of DNSBL servers that list the sender to auto-learn as spam.
A number. Default: 2. Name in the API: learnSpamFromRblHits.
Learn spam from spam traps¶
Train as spam messages sent to spam trap addresses.
On or off. Default: On. Name in the API: learnSpamFromTraps.
Learn ham from address books¶
Whether to automatically learn ham messages from senders in the user's address book.
On or off. Default: On. Name in the API: learnHamFromCard.
Learn ham from trusted replies¶
Whether to automatically learn ham messages that are replies to messages sent by the recipient.
On or off. Default: On. Name in the API: learnHamFromReply.