sisyphus

Author	SHA1	Message	Date
Carlo Strub	54af109661	if word list is too long, only take a random subsample (fixes #11 )	6 years ago
Carlo Strub	e7ac26f264	Add dry run mode that does not move mails after classification (fixes #8 )	6 years ago
Carlo Strub	3a95f5cf1b	Use path/filepath for cleaner and safer path generation	6 years ago
Carlo Strub	cb1d924333	please gometalinter	7 years ago
Carlo Strub	bbee193679	improve memory footprint	7 years ago
Carlo Strub	87b51ebee0	glide up. load mails correctly before learning or classifying them. improve some logging messages	7 years ago
Carlo Strub	3c665ed446	classify also if nothing have ever been learned	7 years ago
Carlo Strub	5413fdb4d4	Do not calculate harmonic mean if there is no probability -- otherwise it crashes	7 years ago
Carlo Strub	1e67467307	use logrus properly (fixes #1 )	7 years ago
Carlo Strub	04cc804722	replace log by logrus	7 years ago
Carlo Strub	276e665172	wrong prior calculation	7 years ago
Carlo Strub	36a3cb6374	make gometalinter happier	7 years ago
Carlo Strub	f76bfbd6eb	add classify test and fix other tests	7 years ago
Carlo Strub	639ffadfe3	fix some static code errors	7 years ago
Carlo Strub	b5b3792efe	implement learning and classifying using Bayes' rule and Hyperloglog data structures -- still way to go though.	7 years ago