Automatic email classification (important/unimportant)
Posted: 24 Jul 2014 06:57
This flow takes the k9 email subject and classifies it as important or unimportant. Important mails will be read aloud. Classification is done through text mining by storing the email subjects (standard:the last 50 mails, but you can edit this in the first Action of the flow) and their words as a learn set. Each time a subject is classified, the flow learns from the learnset and the probability of the words being important or unimportant is calculated using the Bayes algorithm (the lowest 2 actions in the flow will reveil the magic...).
Start the flow manually (or send yourself an email without a subject
) to maintain the classified subjects. The flow will learn from your corrections.
!!Important!!
Please be aware that you must maintain the email classification a few times before your emails are categorized correctly. At first all received emails will be classified as "important". You must manually select a few emails and categorize them as "unimportant", so the flow can learn from your input.
Start the flow manually (or send yourself an email without a subject
) to maintain the classified subjects. The flow will learn from your corrections.!!Important!!
Please be aware that you must maintain the email classification a few times before your emails are categorized correctly. At first all received emails will be classified as "important". You must manually select a few emails and categorize them as "unimportant", so the flow can learn from your input.
,