- Taschenbuch: 84 Seiten
- Verlag: Vdm Verlag Dr. Müller (14. April 2008)
- Sprache: Englisch
- ISBN-10: 383645971X
- ISBN-13: 978-3836459716
- Größe und/oder Gewicht: 22,9 x 15,2 x 0,4 cm
|
||||||||||||||||||||
|
||||||||||||||||||||
|
||||||||||||||||||||
Produktinformation |
Produktbeschreibungen
Kurzbeschreibung
Today's attacks against Bayesian spam filters attempt to keep the content of spam emails visible to humans, but obscured to filters, or they attempt to fool the filters with additional good words appended to the spam. Attacks of this kind have characteristics that are detectable by spam filters, but there is another conceivable approach: one could substitute suspicious words in spam emails with innocent words to make the spam emails appear as legitimate emails (i.e., ham emails). In this book, first, Bayesian spam filtering, including its mathematical foundation, is described. Then, other anti-spam approaches are presented to highlight specific strengths and weaknesses of Bayesian spam filters. Existing attacks against Bayesian spam filters are shown, after which substitution attacks are analyzed: the preconditions of this new kind of attack are examined, and the effectiveness of substitution attacks is measured for three different spam filters. "Penetrating Bayesian Spam Filters" is aimed at computer security specialists, developers of Bayesian spam filters, and anyone interested in the limits of currently used spam filtering technology.
Synopsis
Today's attacks against Bayesian spam filters attempt to keep the content of spam emails visible to humans, but obscured to filters, or they attempt to fool the filters with additional good words appended to the spam. Attacks of this kind have characteristics that are de--tectable by spam filters, but there is another conceivable approach: one could substitute suspicious words in spam emails with innocent words to make the spam emails appear as legitimate emails (i.e., ham emails). In this book, first, Bayesian spam filtering, including its mathematical foundation, is described. Then, other anti-spam approaches are presented to highlight specific strengths and weak--nesses of Bayesian spam filters. Existing attacks against Bayesian spam filters are shown, after which substitution attacks are analyzed: the preconditions of this new kind of attack are examined, and the effectiveness of substitution attacks is measured for three different spam filters. "Penetrating Bayesian Spam Filters" is aimed at computer security specialists, developers of Bayesian spam filters, and anyone interested in the limits of currently used spam filtering technology.
Tags(Was ist das?)Bei einem Tag handelt es sich um ein Schlagwort, das zum Produkt passt.
Tags erleichtern allen Kunden die Suche und die Sortierung ihrer Lieblingsprodukte. |
Eine digitale Version dieses Buchs im Kindle-Shop verkaufen
Wenn Sie ein Verleger oder Autor sind und die digitalen Rechte an einem Buch haben, können Sie die digitale Version des Buchs in unserem Kindle-Shop verkaufen.
Weitere Informationen
Kundenrezensionen
Es gibt noch keine Kundenrezensionen auf Amazon.de
5 Sterne
4 Sterne
3 Sterne
2 Sterne
1 Sterne
Die hilfreichsten Kundenrezensionen auf Amazon.com (beta)
Amazon.com:
1 Rezension
1 von 1 Kunden fanden die folgende Rezension hilfreich
try to evade an antispam Bayesian
8. August 2008
Format:Taschenbuch
In this short monograph, Bayler provides a good summary of the spam problem. The main current antispam methods are described. Like blacklisting against fields in the email header, or against domains in links in the email body. Other methods include blacklisting against IP addresses of email servers that open a port 25 connection to your mail server. In this latter case, the problem is the rise of botnets. Where spammers take over many computers, and then use those to inject spam into the Internet, going to recipients at various addresses. The reader should be able to see that it is very difficult to reduce these entry points.
But the main narrative concerns the use of Bayesian methods on the mail servers, typically against incoming mail, though the methods are certainly usable against outgoing messages. The maths behind Bayesians is deliberately kept simple. A few equations about probabilities. A point is that you don't need a strong maths background to understand current usages of Bayesians and what the author experimented with.
Bayler studied how substituting less common synonyms for certain words in an existing spam email was able to increase the probability that it was classified as "ham" (ie. non-spam) by various antispam Bayesians, including that used by the open source Spam Assassin. This method was offered as an alternative to the spammer's typical anti-Bayesian approach of padding a spam with extra words that are typical of non-spam. Hoping to have the Bayesian misclassify it as ham. The problem spammers face is that heuristics might be run to detect the appending of such lines, and remove them, prior to any Bayesian being run. Along with the detection itself being a heuristic that increases the assignation of the message as being spam.
The problem with Bayler's method, at least from the spammer's viewpoint, is that the autogeneration of synonyms can degrade the meaning of the message. Hence lowering its ultimate efficacy. The book's conclusion suggests ways that the method can be improved. It does not take the next step, by looking at countermeasures. Perhaps a topic for another book?
But the main narrative concerns the use of Bayesian methods on the mail servers, typically against incoming mail, though the methods are certainly usable against outgoing messages. The maths behind Bayesians is deliberately kept simple. A few equations about probabilities. A point is that you don't need a strong maths background to understand current usages of Bayesians and what the author experimented with.
Bayler studied how substituting less common synonyms for certain words in an existing spam email was able to increase the probability that it was classified as "ham" (ie. non-spam) by various antispam Bayesians, including that used by the open source Spam Assassin. This method was offered as an alternative to the spammer's typical anti-Bayesian approach of padding a spam with extra words that are typical of non-spam. Hoping to have the Bayesian misclassify it as ham. The problem spammers face is that heuristics might be run to detect the appending of such lines, and remove them, prior to any Bayesian being run. Along with the detection itself being a heuristic that increases the assignation of the message as being spam.
The problem with Bayler's method, at least from the spammer's viewpoint, is that the autogeneration of synonyms can degrade the meaning of the message. Hence lowering its ultimate efficacy. The book's conclusion suggests ways that the method can be improved. It does not take the next step, by looking at countermeasures. Perhaps a topic for another book?
Ist diese Rezension hilfreich?
Wir wollen von Ihnen hören.
Kunden diskutieren
|
Das Forum zu diesem Produkt
Fragen stellen, Meinungen austauschen, Einblicke gewinnen Aktive Diskussionen in ähnlichen Foren
Kundendiskussionen durchsuchen
|
Ähnliche Foren
|
||||||||||||||||||||||||||||||||||
Lieblingslisten
Ähnliche Artikel finden
Anhand des Sachgebietes nach ähnlichen Produkten suchen:
Ihr Kommentar
Möchten Sie die Produktinformationen aktualisieren oder Feedback zu den Produktabbildungen geben?
|
