Patterns are more important than words (barring "#FF0000", you can be pretty much 100% confident that anything sent as an HTML mail will "#FF0000" in it is not worth reading). Take a look at http://en.wikipedia.org/wiki/Bayesian_spam_filtering and the references it has for one approach (IIRC, one of the first experiments with technique found that after training the filter, "#FF0000" was the expression most likely to indicate Spam, see I told you so).
发布评论
评论（3）
模式比文字更重要（除了“#FF0000”，您几乎可以 100% 确信以 HTML 邮件形式发送的任何内容中包含“#FF0000”的内容都不值得阅读）。查看 http://en.wikipedia.org/wiki/Bayesian_spam_filtering 及其引用有一种方法（IIRC，第一个技术实验发现，在训练过滤器后，“#FF0000”是最有可能指示垃圾邮件的表达方式，请参阅我告诉过你的）。
Patterns are more important than words (barring "#FF0000", you can be pretty much 100% confident that anything sent as an HTML mail will "#FF0000" in it is not worth reading). Take a look at http://en.wikipedia.org/wiki/Bayesian_spam_filtering and the references it has for one approach (IIRC, one of the first experiments with technique found that after training the filter, "#FF0000" was the expression most likely to indicate Spam, see I told you so).
您可能想查看 Paul Graham 垃圾邮件过滤文章 您还可以查看 使用朴素贝叶斯分类器的垃圾邮件过滤器的 C# 实现
You may want to check out Paul Graham spam filtering article You can also have a look at C# implementation of spam filter using Naive Bayes Classifier
这是基于单词黑名单的简单手工垃圾邮件过滤器： 基于黑名单的垃圾邮件过滤器的 LINQ 查询
当您没有动力添加库或复杂的自定义解决方案时，此解决方案适用。
Here is the trivial hand-made spam filter based on word blacklist: LINQ Query for Blacklist-Based Spam Filter
This solution is applicable when you don't find motivation to add libraries or complicated custom solution.