【Movable Type】スパム対策プラグインBayesSpamFilterの今日の成果

Posted by kumacchi on 2011年1月26日 , No comment

昨日公開したベイズ理論を利用したMovable Type用のスパム対策プラグインBayesSpamFilterですが、毎日悩んでいた英文スパムをほぼブロックしてくれるようになったのでとてもいい感じです。ちゃんと普通の文章はハム判定してくれます。

【Movale Type】ベイズ理論を利用したスパムフィルタプラグインを作ってみた。【人柱用】 – KUMA TYPE

現在BayesSpamFilterはプラグインのディレクトリの中にdebug_log.cgiというデバッグログを吐くようになっています。下がそれですが判定結果や投稿された文章などの情報が出力されるのでこのプラグインを利用される方は参考にしてください。一応内容の解説もしておきます。下のログが昨日の成果です。

2011-01-25-03-13-44の部分は日付と時刻を表しています。

prefabrik ev fiyatlari
Michavd@yahoo.com
http://www.prefabrikev.biz
Your RSS feed doesn’t work in my browser (google chrome) how can I fix it?

の部分はコメントの投稿者・メールアドレス・本文になります。
prefabrik ev fiyatlari Michavd@yahoo.com http: www.prefabrikev.biz Your RSS feed doesn’t work in my browser (google chrome ) how can I fix it?
の部分は分かち書きされた結果です。英語の場合は元々分かち書きされているので改行が無くなる程度ですが下の方にある日本語のテスト書き込みを見るとそれなりに分かち書きされているのが分かるかと思います。
2011-01-25-03-13-44 01 spam = 0.999999434725567
2011-01-25-03-13-44 02 ham = 0.0010632725646637

の部分はスパムである確率とハムである確率を数値で表しています。数値が大きい方が優先されます。
2011-01-25-03-13-44 スパム判定
は、判定結果です。英文のスパム投稿を全てスパム判定してくれています。

2011-01-25-03-13-44
prefabrik ev fiyatlari
Michavd@yahoo.com
http://www.prefabrikev.biz
Your RSS feed doesn’t work in my browser (google chrome) how can I fix it?
2011-01-25-03-13-44 getHash:plugins/BayesSpamFilter/ham/_ham_log.cgi
2011-01-25-03-13-44 getHash:plugins/BayesSpamFilter/spam/_spam_log.cgi
2011-01-25-03-13-44
prefabrik ev fiyatlari Michavd@yahoo.com http: www.prefabrikev.biz Your RSS feed doesn’t work in my browser (google chrome ) how can I fix it?
2011-01-25-03-13-44 01 spam = 0.999999434725567
2011-01-25-03-13-44 02 ham = 0.0010632725646637
2011-01-25-03-13-44 スパム判定
2011-01-25-05-25-25
sesli erotik
Seward91@gmail.com
http://www.seslierotik.com
WONDERFUL Post.thanks for share..more wait .. …
2011-01-25-05-25-25 getHash:plugins/BayesSpamFilter/ham/_ham_log.cgi
2011-01-25-05-25-25 getHash:plugins/BayesSpamFilter/spam/_spam_log.cgi
2011-01-25-05-25-25
sesli erotik Seward91@gmail.com http: www.seslierotik.com WONDERFUL Post.thanks for share..more wait .. …
2011-01-25-05-25-25 01 spam = 0.999936800947123
2011-01-25-05-25-25 02 ham = 0.0112425135816203
2011-01-25-05-25-25 スパム判定
2011-01-25-12-58-28
Bosworth
Thomas.L.Woodson@myspace.com
http://www.framefrom.com/story.php?title=drugs-are-tested
daaaaam dawg I would like admirers like a mofo
2011-01-25-12-58-28 getHash:plugins/BayesSpamFilter/ham/_ham_log.cgi
2011-01-25-12-58-28 getHash:plugins/BayesSpamFilter/spam/_spam_log.cgi
2011-01-25-12-58-28
Bosworth Thomas.L.Woodson@myspace.com http: www.framefrom.com story.php?title =drugs-are-tested daaaaam dawg I would like admirers like a mofo
2011-01-25-12-58-28 01 spam = 0.999559214833113
2011-01-25-12-58-28 02 ham = 0.0296879780754909
2011-01-25-12-58-28 スパム判定
2011-01-25-16-21-24
convertible sofas
Ack7516@yahoo.com
http://www.futoncreations.com/sofa-beds.html
Frankly speaking,I love your blog’s style. I believe your opinions are affordable.But I really do not concur with you to some extent.
2011-01-25-16-21-24 getHash:plugins/BayesSpamFilter/ham/_ham_log.cgi
2011-01-25-16-21-24 getHash:plugins/BayesSpamFilter/spam/_spam_log.cgi
2011-01-25-16-21-24
convertible sofas Ack7516@yahoo.com http: www.futoncreations.com sofa-beds.html Frankly speaking ,I love your blog ’s style. I believe your opinions are affordable.But I really do not concur with you to some extent.
2011-01-25-16-21-24 01 spam = 0.999999847121269
2011-01-25-16-21-24 02 ham = 0.000552953379076641
2011-01-25-16-21-24 スパム判定
2011-01-25-20-42-30
mercedes benz of isle of man
Pangilinan1@ymail.com
http://used.mercedes-benz.co.uk/mercedes/Mercedes-Benz-of-Isle-of-Man/c-class/used-cars/2-mbid-432
It’s really a nice and helpful piece of information. I’m glad that you shared this helpful info with us. Please keep us informed like this. Thanks for sharing.
2011-01-25-20-42-30 getHash:plugins/BayesSpamFilter/ham/_ham_log.cgi
2011-01-25-20-42-30 getHash:plugins/BayesSpamFilter/spam/_spam_log.cgi
2011-01-25-20-42-30
mercedes benz of isle of man Pangilinan1@ymail.com http: used.mercedes-benz.co.uk mercedes Mercedes-Benz-of-Isle-of-Man c-class used-cars 2-mbid-432 It ’s really a nice and helpful piece of information. I ’m glad that you shared this helpful info with us. Please keep us informed like this. Thanks for sharing.
2011-01-25-20-42-30 01 spam = 0.999999593482426
2011-01-25-20-42-30 02 ham = 0.000901684525104682
2011-01-25-20-42-30 スパム判定
2011-01-25-23-05-37
Rolf Femat
Dodsworth146@hotmail.com

I feel strongly about it and enjoyed learning more on this topic. If possible, as you become an expert, would you mind updating your blog with more info? It is highly helpful for me. Big thumb up for this blog post!
2011-01-25-23-05-37 getHash:plugins/BayesSpamFilter/ham/_ham_log.cgi
2011-01-25-23-05-37 getHash:plugins/BayesSpamFilter/spam/_spam_log.cgi
2011-01-25-23-05-37
Rolf Femat Dodsworth146@hotmail.com I feel strongly about it and enjoyed learning more on this topic. If possible , as you become an expert , would you mind updating your blog with more info? It is highly helpful for me. Big thumb up for this blog post!
2011-01-25-23-05-37 01 spam = 0.999999984402018
2011-01-25-23-05-37 02 ham = 0.000176623791665328
2011-01-25-23-05-37 スパム判定
2011-01-25-23-24-50
Bud Dukart
Calico92@hotmail.com

I have read few of the posts on your website these few days, and I really like your style of blogging. I added it to my bookmark internet site list and will be checking back soon.
2011-01-25-23-24-50 getHash:plugins/BayesSpamFilter/ham/_ham_log.cgi
2011-01-25-23-24-50 getHash:plugins/BayesSpamFilter/spam/_spam_log.cgi
2011-01-25-23-24-50
Bud Dukart Calico92@hotmail.com I have read few of the posts on your website these few days , and I really like your style of blogging. I added it to my bookmark internet site list and will be checking back soon.
2011-01-25-23-24-50 01 spam = 0.999999843255156
2011-01-25-23-24-50 02 ham = 0.0005599014769393
2011-01-25-23-24-50 スパム判定
2011-01-25-23-38-43
prefabrik yapilar
Checa@yahoo.com
http://www.prefabrikyapifiyatlari.com
Ohh very much thanks admin
2011-01-25-23-38-43 getHash:plugins/BayesSpamFilter/ham/_ham_log.cgi
2011-01-25-23-38-43 getHash:plugins/BayesSpamFilter/spam/_spam_log.cgi
2011-01-25-23-38-43
prefabrik yapilar Checa@yahoo.com http: www.prefabrikyapifiyatlari.com Ohh very much thanks admin
2011-01-25-23-38-43 01 spam = 0.999953280422332
2011-01-25-23-38-43 02 ham = 0.00966628018515203
2011-01-25-23-38-43 スパム判定
2011-01-26-00-03-23
くまっち

コレはテストの書き込みですちゃんと日本語の普通の文章が書き込めることを確認するためのテストです。
2011-01-26-00-03-23 getHash:plugins/BayesSpamFilter/ham/_ham_log.cgi
2011-01-26-00-03-23 getHash:plugins/BayesSpamFilter/spam/_spam_log.cgi
2011-01-26-00-03-23
くまっちコレはテストの書き込みですちゃんと日本語の普通の文章が書き込めることを確認するためのテストです
2011-01-26-00-03-23 01 ham = 0.999998774293423
2011-01-26-00-03-23 02 spam = 0.00156569845464989
2011-01-26-00-03-23 ハム判定

英文スパムは全てスパム判定されていますが、実際には二つのコメントスパムがブロックされずに投稿されました。原因はMovable Typeに標準搭載されているSpamLookupがリンクを含まない書き込みを好評価するようになっているためでBayse Spam Filterが-1.0しているのに対してSpamLookupが+1.0しているので相殺されてスコアが0になったためにブロックされずに投稿されてしまいました。

20110899