摘要： 针对垃圾邮件短小、一定时间内在网络上重复、大量地散发的特点，提出了基于签名的近似垃圾邮件检测算法(ASD)。该算法以句为基本单位，求取邮件所含的全部句子的摘要，垃圾邮件的近似检测转变为两个摘要集近似度的比较。通过与近似文本查询算法DSC、DSC-SS、I-Match 的比较，ASD 算法在近似垃圾邮件查询中，表现出样本集的存储空间大小适中、运算时间短、鲁棒性高、高准确率、高召回率的特征。

关键词: 近似垃圾邮件检测；垃圾邮件过滤；签名；文本近似度；查询

Abstract: In the term of characteristics of spammers sending spam in bulk over a relatively short period of time, this paper presents a signature-based approximate spam detection(ASD) algorithm. ASD detects similarity of E-mails by comparing sets of digests of all sentences in E-mails. The paper compares ASD algorithm with DSC, DSC-SS, I-Match algorithms. To approximate spam detection, ASD has a good performance in samples storage, computing time, robustness, precision and recall

Key words: Approximate spam detection; Spam filtering; Signature; Similarity of documents; Query