updates to classifier primer

This commit is contained in:
Kyle Drake 2016-01-01 01:52:21 -06:00
parent d9f8d1b688
commit c42196e16d

View file

@ -292,13 +292,16 @@ task :update_screenshots => [:environment] do
end end
=end =end
desc 'train_classifier' desc 'prime_classifier'
task :train_classifier => [:environment] do task :prime_classifier => [:environment] do
Site.select(:id, :username).where(is_banned: false, is_deleted: false).all.each do |site| Site.select(:id, :username).where(is_banned: false, is_deleted: false).all.each do |site|
next if site.site_files_dataset.where(classifier: 'spam').count > 0
html_files = site.site_files_dataset.where(path: /\.html$/).all html_files = site.site_files_dataset.where(path: /\.html$/).all
html_files.each do |file| html_files.each do |html_file|
site.train html_files.path print "training #{site.username}/#{html_file.path}..."
site.train html_file.path
print "done.\n"
end end
end end
end end