<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Bayesian on the bin</title><link>https://thebin.net/tags/bayesian/</link><description>Recent content in Bayesian on the bin</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Mon, 14 Sep 2015 00:00:00 +0000</lastBuildDate><atom:link href="https://thebin.net/tags/bayesian/index.xml" rel="self" type="application/rss+xml"/><item><title>Zimbra 8.6: Bayesian Poisoning</title><link>https://thebin.net/2015/09/zimbra-8-6-bayesian-poisoning/</link><pubDate>Mon, 14 Sep 2015 00:00:00 +0000</pubDate><guid>https://thebin.net/2015/09/zimbra-8-6-bayesian-poisoning/</guid><description>&lt;p&gt;Let me start by saying that this problem is not unique to Zimbra and it certainly isn’t unique the version 8.6, however I was using Zimbra 8.6 when I ran into this problem, so this is how I fixed it.&lt;/p&gt;&#10;&lt;h2 id="what-is-bayesian-poisoning"&gt;What is Bayesian Poisoning?&lt;/h2&gt;&#10;&lt;p&gt;One of the core tenants of spam filtering using Bayesian probability to increase or decrease a particular messages score based on the likelihood it is spam.  This is done by compiling a database, often called the Bayes DB, which contains tokens resulting from the Bayesian filtering, these tokens are keywords and combinations that will either push up or down the probability that a given message is spam.  So Bayesian Poisoning is when that DB is intentionally populated with invalid references which result in either more spam being marked as not spam, or more legitimate mails being marked as spam.&lt;/p&gt;</description></item></channel></rss>