WO2019137290A1 - 一种发信人的信誉值生成方法与过滤垃圾邮件的方法 - Google Patents

一种发信人的信誉值生成方法与过滤垃圾邮件的方法 Download PDF

Info

Publication number
WO2019137290A1
WO2019137290A1 PCT/CN2019/070224 CN2019070224W WO2019137290A1 WO 2019137290 A1 WO2019137290 A1 WO 2019137290A1 CN 2019070224 W CN2019070224 W CN 2019070224W WO 2019137290 A1 WO2019137290 A1 WO 2019137290A1
Authority
WO
WIPO (PCT)
Prior art keywords
sender
reputation value
target
reputation
value
Prior art date
Application number
PCT/CN2019/070224
Other languages
English (en)
French (fr)
Inventor
陈磊华
潘庆峰
Original Assignee
论客科技（广州）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 论客科技（广州）有限公司 filed Critical 论客科技（广州）有限公司
Publication of WO2019137290A1 publication Critical patent/WO2019137290A1/zh
Priority to US16/924,263 priority Critical patent/US11343213B2/en

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L51/00—User-to-user messaging in packet-switching networks, transmitted according to store-and-forward or real-time protocols, e.g. e-mail
    • H04L51/21—Monitoring or handling of messages
    • H04L51/212—Monitoring or handling of messages using filtering or selective blocking
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L51/00—User-to-user messaging in packet-switching networks, transmitted according to store-and-forward or real-time protocols, e.g. e-mail
    • H04L51/42—Mailbox-related aspects, e.g. synchronisation of mailboxes

Definitions

  • the present invention relates to the field of email technologies, and in particular, to a method for generating a reputation value of a sender and a method for filtering spam.
  • the filtering technologies for existing spam mainly include mail sending authentication, black and white list, content filtering technology, fingerprint technology, Bayesian filtering technology, etc., but these technologies have more or less shortcomings and cannot be accurately judged 100%.
  • the method for generating a sender reputation based on the mail user behavior of the patent application number 201510401224.8 discloses that the method of machine learning is used to analyze the massive log generated online, and the size of the mail body, the number of successful transmissions, and the number of failed transmissions are selected. Multiple feature dimensions, such as the total number of recipients, the number of recipients, the content of the message, the domain name of the sender, the success of the IP transmission, and the number of failures.
  • the feature model is trained through massive logs, and an overall reputation is generated for these feature values.
  • the score library the real-time mail matches this feature reputation library, generates a sender-specific reputation value for the sender who satisfies the condition, and improves the accuracy of the reputation value.
  • the above sender reputation generation method relies on the sender's transmission history, so that a batch of mailboxes can be specially prepared by the "raising number” method, and a batch of normal mails are sent to the mailboxes controlled by them within a certain period of time. This will improve the credibility of the sender, or send some normal mail to the mailbox controlled by you while sending spam, in order to offset the sender's reputation degradation caused by sending spam.
  • the object of the present invention is to provide a sender's reputation value generation method and a method for filtering spam, which does not depend on the sender's transmission history, and can accurately calculate the sender's reputation value, thereby effectively avoiding the sender by cheating. Reputation value.
  • an embodiment of the present invention provides a method for generating a reputation value of a sender, including:
  • the calculating the initial credential value of the target sender according to the sender identity of the non-spam log, the identity of the recipient, and the number of senders sent by the sender specifically:
  • the sender identity Separating the sender identity, the recipient identity, and the sending amount information of the non-spam log, respectively, and creating a first data table; wherein the first data table includes a sender identity identifier and a recipient identity primary key ;
  • the initial reputation value of the target sender is calculated according to the number of primary keys of the sender and the primary key of the addressee and the preset total reputation value, which specifically includes:
  • N is the preset total reputation value
  • M is the number of primary keys of the sender identity identifier and the recipient identity primary key.
  • the calculating the delivery credit value of the target sender according to the sender identity of the non-spam log and the number of senders sent by the sender, specifically:
  • the current credit value of the plurality of recipients the current credit value of the target sender, the number of transmissions by the target sender to the plurality of recipients, and the total transmission of the target sender to the external sender. Quantity, calculating the delivery reputation value of the target sender.
  • the according to the current credit value of the plurality of recipients, the current credit value of the target sender, the number of transmissions by the target sender to the plurality of recipients, and the target sender The total number of letters sent by the letter, calculating the value of the target sender's delivery reputation, including:
  • the input i is the current reputation value of the i-th recipient
  • n i is the number of transmissions by the target sender to the ith recipient
  • the total is the total transmission sent by the target sender to the external sender.
  • the calculating the current reputation value of the target sender according to the initial reputation value and the delivery reputation value of the target sender specifically:
  • the method for generating a reputation value of the sender further includes:
  • sender reputation table Creating a sender reputation table according to the current reputation value of the target sender; wherein the sender reputation table includes a sender identity primary key.
  • the sender identity is a sender email address and/or a sender domain name.
  • the embodiment of the invention further provides a method for filtering spam, comprising:
  • a method for generating a reputation value of a sender includes acquiring a non-spam log within a set time period; Calculating the sender identity of the non-spam log, the identity of the recipient, and the number of senders sent by the sender, calculating the initial reputation value of the target sender; sending the identity of the sender according to the non-spam log and sending the sender a quantity, a delivery reputation value of the target sender is calculated; and a current reputation value of the target sender is calculated according to the initial credit value and the delivery reputation value of the target sender.
  • the sender's reputation value generation method does not depend on the sender's transmission history, and can accurately calculate the sender's reputation value, thereby effectively avoiding the sender's reputation value by cheating.
  • a method for filtering spam provided by an embodiment of the present invention.
  • FIG. 1 is a flowchart of a method for generating a reputation value of a sender according to an embodiment of the present invention
  • FIG. 2 is a flowchart of a method for filtering spam according to an embodiment of the present invention.
  • FIG. 1 is a flowchart of a method for generating a reputation value of a sender according to an embodiment of the present invention.
  • the method for generating a reputation value of the sender includes:
  • the log that has been determined to be spam is removed, and the non-spam log in the set time period is obtained, that is, when the sender's reputation value is calculated, it is determined as garbage.
  • the log of the mail does not participate in the calculation, and the spam log is pre-excluded, which can avoid unnecessary technical processes, avoid the influence of the mail that has been determined to be spam on the credit value of the sender, and improve the accuracy of calculating the credit value of the sender.
  • S12 Calculate an initial reputation value of the target sender according to the sender identity of the non-spam log, the identity of the addressee, and the number of senders sent by the sender;
  • S14 Calculate a current reputation value of the target sender according to the initial reputation value and the delivery reputation value of the target sender.
  • the sender's initial reputation value and delivery reputation value by selecting the sender identity, the recipient identity, and the sender's number of transmissions from the non-spam log, and calculate the target based on the initial reputation value and the delivery reputation value.
  • the sender's pass-through value does not depend on the sender's transmission history, and can accurately calculate the sender's reputation value, effectively avoiding the credibility of the sender by cheating.
  • S12 calculating an initial reputation value of the target sender according to the sender identity of the non-spam log, the identity of the recipient, and the number of senders sent by the sender, including:
  • the sender identity Separating the sender identity, the recipient identity, and the sending amount information of the non-spam log, respectively, and creating a first data table; wherein the first data table includes a sender identity identifier and a recipient identity primary key ;
  • the initial value of the target sender is calculated according to the number of primary keys of the sender and the primary key of the addressee, and the preset total reputation value, which specifically includes:
  • N is the preset total reputation value
  • M is the number of primary keys of the sender identity identifier and the recipient identity primary key.
  • S13 calculating a delivery reputation value of the target sender according to the sender identity of the non-spam log and the number of senders sent by the sender, specifically:
  • the current credit value of the plurality of recipients the current credit value of the target sender, the number of transmissions by the target sender to the plurality of recipients, and the total transmission of the target sender to the external sender. Quantity, calculating the delivery reputation value of the target sender.
  • the current reputation value according to the multiple recipients, the current credit value of the target sender, the number of signalings sent by the target sender to multiple recipients, and Calculating the total number of transmissions sent by the target sender to the external sender, and calculating the delivery reputation value of the target sender specifically:
  • the inputi is the current reputation value of the i-th recipient
  • ni is the number of transmissions by the target sender to the ith recipient
  • total is the total number of transmissions sent by the target sender to the external sender.
  • the current credit value of the plurality of recipients the current credit value of the target sender, the number of signalings sent by the target sender to the plurality of recipients, and the target sender to the outside
  • the total number of transmissions sent by the sender the transmission credit value of the target sender is calculated, and the credit value of the sender is transmitted to the addressee according to the proportion of the self-issuance, which can improve the accuracy of the transmission of the reputation value.
  • the sender A sends an email to the recipient B. If 1/10 of the email sent by the sender A is sent to the sender B, the sender A's reputation value is transmitted 1/10 to the sender B.
  • S14 calculating a current reputation value of the target sender according to the initial reputation value and the delivery reputation value of the target sender, specifically:
  • the method for generating a reputation value of the sender further includes:
  • sender reputation table Creating a sender reputation table according to the current reputation value of the target sender; wherein the sender reputation table includes a sender identity primary key.
  • the cheater sends a large amount of mail to the mailbox D through a mailbox C that is controlled by itself, but through the credit value generation method of the sender, the reputation value of the technical mailbox C and the mailbox D can be known, since there is no other reputable mailbox.
  • Mail has been sent to the mailbox C, so the credit value of the mailbox C is very close to 0, so the delivery credit value delivered by the mailbox C to the mailbox D is O, even if the mailbox C sends a large amount of mail to the mailbox D, the credit value obtained by the mailbox D is N. /M after a number of iterations will be very close to 0, can not improve the reputation value of mailbox D.
  • the sender identity is a sender email address and/or a sender domain name.
  • the accuracy of the garbage control of the personal mailbox can be improved.
  • the reputation value of the sender's domain name for some newly-created mailboxes, although it has not appeared in the system, since the reputation value of the sender's domain name is high enough, it is possible to relax the garbage control.
  • the credibility value of the sender's email address and/or the reputation value of the sender's domain name can improve the accuracy and flexibility of the email spam control, and greatly reduce the problem of spam misjudgment.
  • the credit value of the sender's email address and the reputation value of the sender's domain name are respectively calculated, which can improve the accuracy of the garbage control of the personal mailbox, and for some newly-created mailboxes, although in the system It has not appeared, but because the reputation value of the sender's domain name is high enough, it can realize the relaxation of garbage control, improve the flexibility of email spam control, and greatly reduce the problem of spam misjudgment.
  • a method for generating a reputation value of a sender includes acquiring a non-spam log within a set time period; Calculating the sender identity of the non-spam log, the identity of the recipient, and the number of senders sent by the sender, calculating the initial reputation value of the target sender; sending the identity of the sender according to the non-spam log and sending the sender a quantity, a delivery reputation value of the target sender is calculated; and a current reputation value of the target sender is calculated according to the initial credit value and the delivery reputation value of the target sender.
  • the sender's reputation value generation method does not depend on the sender's transmission history, and can accurately calculate the sender's reputation value, thereby effectively avoiding the sender's reputation value by cheating.
  • the embodiment of the present invention also provides a method for filtering spam.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

本发明公开了一种发信人的信誉值生成方法与过滤垃圾邮件的方法，所述发信人的信誉值生成方法包括获取设定时间段内的非垃圾邮件日志；根据所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信人的发信数量，计算目标发信人的初始信誉值；根据所述非垃圾邮件日志的发信人身份标识以及发信人的发信数量，计算所述目标发信人的传递信誉值；根据所述目标发信人的初始信誉值和传递信誉值，计算所述目标发信人的当前信誉值。所述发信人的信誉值生成方法不依赖发信人的发信历史，并且能够准确计算发信人的信誉值，有效避免通过作弊的方式提高发信人的信誉值。

Description

一种发信人的信誉值生成方法与过滤垃圾邮件的方法 技术领域
本发明涉及电子邮件技术领域，具体涉及一种发信人的信誉值生成方法与过滤垃圾邮件的方法。
背景技术
随着电子邮件的广泛应用，伴随而来的垃圾邮件问题日益严重。它不仅消耗网络资源、占用网络带宽、浪费用户的宝贵时间和上网费用，而且严重威胁网络安全，已成为网络公害，带来了严重的经济损失。中国互联网协会反垃圾邮件中心发布调查报告显示，垃圾邮件在规模上不断增长，中国网民平均每周收到的垃圾邮件比例为55.65％。迫切需要有效的技术解决垃圾邮件泛滥的问题。
针对现有垃圾邮件的过滤技术主要有邮件发送认证、黑白名单、内容过滤技术，指纹技术，贝叶斯过滤技术等，但是这些技术都或多或少的存在缺点，无法百分之百的准确判断，其中专利申请号为201510401224.8的基于邮件用户行为的发件人信誉生成方法公开了通过机器学习的方法，对线上产生的海量日志进行分析，选取邮件体大小，发信成功数，发信失败数，发信总数，收件人回复数，邮件内容，发件人域名，IP发信成功和失败数等多个特征维度，通过海量日志对特征模型进行训练，对这些特征值生成了一个总体的信誉分值库，实时的邮件匹配这个特征信誉库，对满足条件的发件人生成发件人特定的信誉值，提高信誉值的准确度。但是，上述发件人信誉生成方法依赖对应发件人的发信历史，导致可以通过“养号”方式专门准备一批邮箱，在一段时间之内发送一批正常邮件给自己控制的邮箱，以此来提高发信人信誉，或者在发送垃圾邮件的同时，发送一些正常邮件给自己控制的邮箱，以此来抵消发送垃圾邮件导致的发信人信誉降低问题。
发明内容
本发明的目的是提供一种发信人的信誉值生成方法与过滤垃圾邮件的方法，不依赖发信人的发信历史，并且能够准确计算发信人的信誉值，有效避免通过作弊的方式提高发信人的信誉值。
为解决以上技术问题，本发明实施例提供一种发信人的信誉值生成方法，包括：
获取设定时间段内的非垃圾邮件日志；
根据所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信人的发信数量，计算目标发信人的初始信誉值；
根据所述非垃圾邮件日志的发信人身份标识以及发信人的发信数量，计算所述目标发信人的传递信誉值；
根据所述目标发信人的初始信誉值和传递信誉值，计算所述目标发信人的当前信誉值。
优选地，所述根据所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信人的发信数量，计算目标发信人的初始信誉值，具体包括：
分别统计所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信数量信息，并创建第一数据表；其中，所述第一数据表包含发信人身份标识以及收信人身份标识主键；
统计所述发信人身份标识以及收信人身份标识主键的主键数量；
根据所述发信人以及收信人主键的主键数量以及预设的总信誉值，计算所述目标发信人的初始信誉值；其中，所述预设的总信誉值为所有发信人的初始信誉值的总和。
优选地，所述根据所述发信人以及收信人主键的主键数量以及预设的总信誉值，计算所述目标发信人的初始信誉值，具体包括：
根据公式reputation＝N/M，计算所述目标发信人的初始信誉值；
其中，N为所述预设的总信誉值，M为所述发信人身份标识以及收信人身 份标识主键的主键数量。
优选地，所述根据所述非垃圾邮件日志的发信人身份标识以及发信人的发信数量，计算所述目标发信人的传递信誉值，具体包括：
分别统计所述非垃圾邮件日志的发信人身份标识以及发信数量信息，并创建第二数据表；其中，所述第二数据表包含发信人身份标识主键；
统计所述第一数据表中所述目标发信人分别对多个收信人的发信数量；
统计所述第二数据表中所述目标发信人对外发信的发信总数量；
根据所述多个收信人的当前信誉值、所述目标发信人的当前信誉值、所述目标发信人分别对多个收信人的发信数量以及所述目标发信人对外发信的发信总数量，计算所述目标发信人的传递信誉值。
优选地，所述根据所述多个收信人的当前信誉值、所述目标发信人的当前信誉值、所述目标发信人分别对多个收信人的发信数量以及所述目标发信人对外发信的发信总数量，计算所述目标发信人的传递信誉值，具体包括：
根据公式delta＝reputation 1×(n 1/total)+reputation 2×(n 2/total)……reputation i×(n i/total)……reputation m×(n m/total)，计算所述目标发信人的传递信誉值；
其中，reputation i为第i个收信人的当前信誉值，n i为所述目标发信人对所述第i个收信人的发信数量，total为所述目标发信人对外发信的发信总数量。
优选地，所述根据所述目标发信人的初始信誉值和传递信誉值，计算所述目标发信人的当前信誉值，具体包括：
根据公式reputation’＝reputation×(N/(N+delta)),计算所述目标发信人的当前信誉值。
优选地，所述发信人的信誉值生成方法还包括：
根据所述目标发信人的当前信誉值，创建发信人信誉表；其中，所述发信人信誉表包括发信人身份标识主键。
优选地，所述发信人身份标识为发信人邮箱地址和/或发信人域名。
本发明实施例还提供一种过滤垃圾邮件的方法，包括：
采用上述的发信人的信誉值生成方法生成发信人的当前信誉值；
根据所述发信人的当前信誉值判断所述发信人发送的邮件是否为垃圾邮件；
当所述发信人的当前信誉值低于预设的阈值时，确定所述发信人发送的邮件为垃圾邮件并对所述发信人发送的邮件进行过滤处理。
相对于现有技术，本发明实施例提供的一种发信人的信誉值生成方法的有益效果在于：所述发信人的信誉值生成方法包括获取设定时间段内的非垃圾邮件日志；根据所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信人的发信数量，计算目标发信人的初始信誉值；根据所述非垃圾邮件日志的发信人身份标识以及发信人的发信数量，计算所述目标发信人的传递信誉值；根据所述目标发信人的初始信誉值和传递信誉值，计算所述目标发信人的当前信誉值。所述发信人的信誉值生成方法不依赖发信人的发信历史，并且能够准确计算发信人的信誉值，有效避免通过作弊的方式提高发信人的信誉值。本发明实施例提供的一种过滤垃圾邮件的方法。
附图说明
图1是本发明实施例提供的一种发信人的信誉值生成方法的流程图；
图2是本发明实施例提供的一种过滤垃圾邮件的方法的流程图。
具体实施方式
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
请参阅图1，其是本发明实施例提供的一种发信人的信誉值生成方法的流程图，所述发信人的信誉值生成方法包括：
S11：获取设定时间段内的非垃圾邮件日志；
通过扫描所述设定时间段内的系统日志，剔除掉已经被判定为垃圾邮件的日志，以获取设定时间段内的非垃圾邮件日志，即计算发信人的信誉值时，被 判定为垃圾邮件的日志不参与计算，预先排除垃圾邮件日志，可以避免不必要的技术过程，同时避免已确定为垃圾邮件的邮件对发信人信誉值的影响，提高计算发信人的信誉值的准确性。
S12：根据所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信人的发信数量，计算目标发信人的初始信誉值；
S13：根据所述非垃圾邮件日志的发信人身份标识以及发信人的发信数量，计算所述目标发信人的传递信誉值；
S14：根据所述目标发信人的初始信誉值和传递信誉值，计算所述目标发信人的当前信誉值。
通过从非垃圾邮件日志中选取发信人身份标识、收信人身份标识以及发信人的发信数量计算发信人的初始信誉值和传递信誉值，并根据初始信誉值和传递信誉值，计算所述目标发信人的传递信誉值，不依赖发信人的发信历史，并且能够准确计算发信人的信誉值，有效避免通过作弊的方式提高发信人的信誉值。
在一种可选的实施例中，S12：根据所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信人的发信数量，计算目标发信人的初始信誉值，具体包括：
分别统计所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信数量信息，并创建第一数据表；其中，所述第一数据表包含发信人身份标识以及收信人身份标识主键；
统计所述发信人身份标识以及收信人身份标识主键的主键数量；
根据所述发信人以及收信人主键的主键数量以及预设的总信誉值，计算所述目标发信人的初始信誉值；其中，所述预设的总信誉值为所有发信人的初始信誉值的总和。
在一种可选的实施例中，所述根据所述发信人以及收信人主键的主键数量以及预设的总信誉值，计算所述目标发信人的初始信誉值，具体包括：
根据公式reputation＝N/M，计算所述目标发信人的初始信誉值；
其中，N为所述预设的总信誉值，M为所述发信人身份标识以及收信人身份标识主键的主键数量。
在一种可选的实施例中，S13：根据所述非垃圾邮件日志的发信人身份标识以及发信人的发信数量，计算所述目标发信人的传递信誉值，具体包括：
分别统计所述非垃圾邮件日志的发信人身份标识以及发信数量信息，并创建第二数据表；其中，所述第二数据表包含发信人身份标识主键；
统计所述第一数据表中所述目标发信人分别对多个收信人的发信数量；
统计所述第二数据表中所述目标发信人对外发信的发信总数量；
根据所述多个收信人的当前信誉值、所述目标发信人的当前信誉值、所述目标发信人分别对多个收信人的发信数量以及所述目标发信人对外发信的发信总数量，计算所述目标发信人的传递信誉值。
在一种可选的实施例中，所述根据所述多个收信人的当前信誉值、所述目标发信人的当前信誉值、所述目标发信人分别对多个收信人的发信数量以及所述目标发信人对外发信的发信总数量，计算所述目标发信人的传递信誉值，具体包括：
根据公式delta＝reputation 1×(n 1/total)+reputation 2×(n 2/total)……reputation i×(n i/total)……reputation m×(n m/total)，计算所述目标发信人的传递信誉值；
其中，reputationi为第i个收信人的当前信誉值，ni为所述目标发信人对所述第i个收信人的发信数量，total为所述目标发信人对外发信的发信总数量。
在本实施例中，根据所述多个收信人的当前信誉值、所述目标发信人的当前信誉值、所述目标发信人分别对多个收信人的发信数量以及所述目标发信人对外发信的发信总数量，计算所述目标发信人的传递信誉值，实现发信人的信誉值按自身发信比例传递给收信人，能够提高传递信誉值的准确性。例如发信人A发邮件给收信人B,如果发信人A发出的邮件有1/10都是发给发信人B的，则将发信人A的信誉值传递1/10给发信人B。
在一种可选的实施例中，S14：根据所述目标发信人的初始信誉值和传递信誉值，计算所述目标发信人的当前信誉值，具体包括：
根据公式reputation’＝reputation×(N/(N+delta)),计算所述目标发信人的当前信誉值。
在一种可选的实施例中，所述发信人的信誉值生成方法还包括：
根据所述目标发信人的当前信誉值，创建发信人信誉表；其中，所述发信人信誉表包括发信人身份标识主键。
所述发信人的信誉生成方法通过枚举所有发信人，并将每个发信人的信誉传递给其相关的收信人。迭代完成后，所有收过邮件的发信人，其信誉值都有增加，即所有发信人在所述设定的时间段内对应的总信誉值增加了delta(因为发信人A将信誉值传递给B之后，A本身的信誉值不会减少，所以相当于凭空创造了一些信誉值给B,于是所有发信人在所述设定的时间段内对应的总信誉值增加了delta),使用N记录这次迭代一共增加的信誉值为N’＝N+delta，以使得所有发信人在所述设定的时间段内对应的总信誉值N’＝N+delta>1。为了让所有发信人的当前总信誉值恢复到N，更新每个发信人的信誉值为reputation’＝reputation×(N/(N+delta)),相当于按比例减少每个人的信誉值，直到信誉值总值恢复到N。这个迭代最终会使得每个发信人的信誉慢慢收敛，当两次迭代导致的发信人信誉变化小于某个特定的阀值，说明已经足够收敛，迭代结束。此时发信人信誉表里面的就是我们需要的每个发信人的发信人信誉值了。从上面的计算过程也可以看出，如果一个邮箱，一直只发送邮件，不接受邮件的，则它在初始的时候的信誉值是N/M,然后每次迭代因为没有其他人给它增加信誉值，且每次迭代结束的时候都会按比例减少信誉值，所以迭代收敛之后，发信人的信誉值机会非常非常小，接近于0。
例如作弊者通过一个自己控制的邮箱C，发送大量邮件给邮箱D，但是通过所述发信人的信誉值生成方法技术邮箱C和邮箱D的信誉值可得知，由于从来没有其他有信誉的邮箱发过邮件给邮箱C,所以邮箱C的信誉值是非常接近0，因此由邮箱C传递给邮箱D的传递信誉值为O，就算邮箱C大量发送邮件给邮箱D,邮箱D得到的信誉值N/M经过多次迭代后也会非常接近0，不能提高邮箱D的信誉值。
在一种可选的实施例中，所述发信人身份标识为发信人邮箱地址和/或发信人域名。
通过计算发信人邮箱地址的信誉值，能够提高个人邮箱的发垃圾控制的准确性。通过计算发信人域名的信誉值，对于一些新出现的邮箱，虽然在系统中没有出现过，但是由于发信人域名的信誉值够高，可以实现放宽发垃圾控制。通过发信人邮箱地址信誉值和/或发信人域名的信誉值可以提高邮箱发垃圾控制的准确性以及灵活性，大大减少垃圾邮件误判问题。
请参阅图2，其是本发明实施例还提供一种过滤垃圾邮件的方法的流程图，所述过滤垃圾邮件的方法包括：
S21：采用上述的发信人的信誉值生成方法生成发信人的当前信誉值；
S22：根据所述发信人的当前信誉值判断所述发信人发送的邮件是否为垃圾邮件；
S23：当所述发信人的当前信誉值低于预设的阈值时，确定所述发信人发送的邮件为垃圾邮件并对所述发信人发送的邮件进行过滤处理。
通过上述的发信人的信誉值生成方法分别计算发信人邮箱地址的信誉值和发信人域名的信誉值，能够提高个人邮箱的发垃圾控制的准确性，对于一些新出现的邮箱，虽然在系统中没有出现过，但是由于发信人域名的信誉值够高，可以实现放宽发垃圾控制，提高邮箱发垃圾控制的灵活性，大大减少垃圾邮件误判问题。
相对于现有技术，本发明实施例提供的一种发信人的信誉值生成方法的有益效果在于：所述发信人的信誉值生成方法包括获取设定时间段内的非垃圾邮件日志；根据所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信人的发信数量，计算目标发信人的初始信誉值；根据所述非垃圾邮件日志的发信人身份标识以及发信人的发信数量，计算所述目标发信人的传递信誉值；根据所述目标发信人的初始信誉值和传递信誉值，计算所述目标发信人的当前信誉值。所述发信人的信誉值生成方法不依赖发信人的发信历史，并且能够准确计算发信人的信誉值，有效避免通过作弊的方式提高发信人的信誉值。本发 明实施例还提供的一种过滤垃圾邮件的方法。
以上是本发明的优选实施方式，应当指出，对于本技术领域的普通技术人员来说，在不脱离本发明原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也视为本发明的保护范围。

Claims (9)

  1. 一种发信人的信誉值生成方法，其特征在于，包括：
    获取设定时间段内的非垃圾邮件日志；
    根据所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信人的发信数量，计算目标发信人的初始信誉值；
    根据所述非垃圾邮件日志的发信人身份标识以及发信人的发信数量，计算所述目标发信人的传递信誉值；
    根据所述目标发信人的初始信誉值和传递信誉值，计算所述目标发信人的当前信誉值。
  2. 如权利要求1所述的发信人的信誉值生成方法，其特征在于，所述根据所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信人的发信数量，计算目标发信人的初始信誉值，具体包括：
    分别统计所述非垃圾邮件日志的发信人身份标识、收信人身份标识以及发信数量信息，并创建第一数据表；其中，所述第一数据表包含发信人身份标识以及收信人身份标识主键；
    统计所述发信人身份标识以及收信人身份标识主键的主键数量；
    根据所述发信人以及收信人主键的主键数量以及预设的总信誉值，计算所述目标发信人的初始信誉值；其中，所述预设的总信誉值为所有发信人的初始信誉值的总和。
  3. 如权利要求2所述的发信人的信誉值生成方法，其特征在于，所述根据所述发信人以及收信人主键的主键数量以及预设的总信誉值，计算所述目标发信人的初始信誉值，具体包括：
    根据公式reputation＝N/M，计算所述目标发信人的初始信誉值；
    其中，N为所述预设的总信誉值，M为所述发信人身份标识以及收信人身份标识主键的主键数量。
  4. 如权利要求3所述的发信人的信誉值生成方法，其特征在于，所述根据所述非垃圾邮件日志的发信人身份标识以及发信人的发信数量，计算所述目标发信人的传递信誉值，具体包括：
    分别统计所述非垃圾邮件日志的发信人身份标识以及发信数量信息，并创建第二数据表；其中，所述第二数据表包含发信人身份标识主键；
    统计所述第一数据表中所述目标发信人分别对多个收信人的发信数量；
    统计所述第二数据表中所述目标发信人对外发信的发信总数量；
    根据所述多个收信人的当前信誉值、所述目标发信人的当前信誉值、所述目标发信人分别对多个收信人的发信数量以及所述目标发信人对外发信的发信总数量，计算所述目标发信人的传递信誉值。
  5. 如权利要求4所述的发信人的信誉值生成方法，其特征在于，所述根据所述多个收信人的当前信誉值、所述目标发信人的当前信誉值、所述目标发信人分别对多个收信人的发信数量以及所述目标发信人对外发信的发信总数量，计算所述目标发信人的传递信誉值，具体包括：
    根据公式delta＝reputation 1×(n 1/total)+reputation 2×(n 2/total)……reputation i×(n i/total)……reputation m×(n m/total)，计算所述目标发信人的传递信誉值；
    其中，reputation i为第i个收信人的当前信誉值，n i为所述目标发信人对所述第i个收信人的发信数量，total为所述目标发信人对外发信的发信总数量。
  6. 如权利要求5所述的发信人的信誉值生成方法，其特征在于，所述根据所述目标发信人的初始信誉值和传递信誉值，计算所述目标发信人的当前信誉值，具体包括：
    根据公式reputation’＝reputation×(N/(N+delta)),计算所述目标发信人的当前信誉值。
  7. 如权利要求2所述的发信人的信誉值生成方法，其特征在于，所述发信人的信誉值生成方法还包括：
    根据所述目标发信人的当前信誉值，创建发信人信誉表；其中，所述发信人信誉表包括发信人身份标识主键。
  8. 如权利要求1-7任一项所述的发信人的信誉值生成方法，其特征在于，所述发信人身份标识为发信人邮箱地址和/或发信人域名。
  9. 一种过滤垃圾邮件的方法，其特征在于，包括：
    采用如权利要求1-8任一项所述的发信人的信誉值生成方法生成发信人的当前信誉值；
    根据所述发信人的当前信誉值判断所述发信人发送的邮件是否为垃圾邮件；
    当所述发信人的当前信誉值低于预设的阈值时，确定所述发信人发送的邮件为垃圾邮件并对所述发信人发送的邮件进行过滤处理。
PCT/CN2019/070224 2018-01-09 2019-01-03 一种发信人的信誉值生成方法与过滤垃圾邮件的方法 WO2019137290A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/924,263 US11343213B2 (en) 2018-01-09 2020-07-09 Method for generating reputation value of sender and spam filtering method

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810019589.8 2018-01-09
CN201810019589.8A CN108282394B (zh) 2018-01-09 2018-01-09 一种发信人的信誉值生成方法与过滤垃圾邮件的方法

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/924,263 Continuation-In-Part US11343213B2 (en) 2018-01-09 2020-07-09 Method for generating reputation value of sender and spam filtering method

Publications (1)

Publication Number Publication Date
WO2019137290A1 true WO2019137290A1 (zh) 2019-07-18

Family

ID=62803304

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/070224 WO2019137290A1 (zh) 2018-01-09 2019-01-03 一种发信人的信誉值生成方法与过滤垃圾邮件的方法

Country Status (3)

Country Link
US (1) US11343213B2 (zh)
CN (1) CN108282394B (zh)
WO (1) WO2019137290A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117745433A (zh) * 2024-02-19 2024-03-22 成都理工大学 基于改进pbft共识机制的能源区块链节点共识方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108282394B (zh) * 2018-01-09 2020-09-25 论客科技（广州）有限公司 一种发信人的信誉值生成方法与过滤垃圾邮件的方法
CN111404805B (zh) * 2020-03-12 2022-11-22 深信服科技股份有限公司 一种垃圾邮件检测方法、装置、电子设备及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050198159A1 (en) * 2004-03-08 2005-09-08 Kirsch Steven T. Method and system for categorizing and processing e-mails based upon information in the message header and SMTP session
US20060168024A1 (en) * 2004-12-13 2006-07-27 Microsoft Corporation Sender reputations for spam prevention
CN101540773A (zh) * 2009-04-22 2009-09-23 成都市华为赛门铁克科技有限公司 一种垃圾邮件检测方法及其装置
CN106341303A (zh) * 2015-07-10 2017-01-18 彩讯科技股份有限公司 基于邮件用户行为的发件人信誉生成方法
CN108282394A (zh) * 2018-01-09 2018-07-13 论客科技（广州）有限公司 一种发信人的信誉值生成方法与过滤垃圾邮件的方法

Family Cites Families (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050091320A1 (en) * 2003-10-09 2005-04-28 Kirsch Steven T. Method and system for categorizing and processing e-mails
US8423563B2 (en) * 2003-10-16 2013-04-16 Sybase, Inc. System and methodology for name searches
WO2005116851A2 (en) * 2004-05-25 2005-12-08 Postini, Inc. Electronic message source information reputation system
CN1746916A (zh) * 2005-10-25 2006-03-15 二六三网络通信股份有限公司 网络ip地址信誉度评估方法及其在电子邮件系统中的应用
US8224905B2 (en) * 2006-12-06 2012-07-17 Microsoft Corporation Spam filtration utilizing sender activity data
CN100539555C (zh) * 2007-06-01 2009-09-09 清华大学 基于可扩展消息在线协议和信誉机制的电子邮件传送方法
US7937468B2 (en) * 2007-07-06 2011-05-03 Yahoo! Inc. Detecting spam messages using rapid sender reputation feedback analysis
US7783597B2 (en) * 2007-08-02 2010-08-24 Abaca Technology Corporation Email filtering using recipient reputation
US8209741B2 (en) * 2007-09-17 2012-06-26 Microsoft Corporation Human performance in human interactive proofs using partial credit
US8359632B2 (en) * 2008-05-30 2013-01-22 Microsoft Corporation Centralized account reputation
US7996475B2 (en) * 2008-07-03 2011-08-09 Barracuda Networks Inc Facilitating transmission of email by checking email parameters with a database of well behaved senders
US8924488B2 (en) * 2010-07-27 2014-12-30 At&T Intellectual Property I, L.P. Employing report ratios for intelligent mobile messaging classification and anti-spam defense
US9519682B1 (en) * 2011-05-26 2016-12-13 Yahoo! Inc. User trustworthiness
CN102413076B (zh) * 2011-12-22 2015-08-05 网易（杭州）网络有限公司 基于行为分析的垃圾邮件判定系统
CN104660490B (zh) * 2014-11-26 2018-08-03 论客科技（广州）有限公司 外发邮件控制方法与系统
CN105610833B (zh) * 2015-12-30 2019-01-18 新浪网技术(中国)有限公司 一种基于ip信誉值的邮件反垃圾方法及系统

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050198159A1 (en) * 2004-03-08 2005-09-08 Kirsch Steven T. Method and system for categorizing and processing e-mails based upon information in the message header and SMTP session
US20060168024A1 (en) * 2004-12-13 2006-07-27 Microsoft Corporation Sender reputations for spam prevention
CN101540773A (zh) * 2009-04-22 2009-09-23 成都市华为赛门铁克科技有限公司 一种垃圾邮件检测方法及其装置
CN106341303A (zh) * 2015-07-10 2017-01-18 彩讯科技股份有限公司 基于邮件用户行为的发件人信誉生成方法
CN108282394A (zh) * 2018-01-09 2018-07-13 论客科技（广州）有限公司 一种发信人的信誉值生成方法与过滤垃圾邮件的方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117745433A (zh) * 2024-02-19 2024-03-22 成都理工大学 基于改进pbft共识机制的能源区块链节点共识方法
CN117745433B (zh) * 2024-02-19 2024-05-28 成都理工大学 基于改进pbft共识机制的能源区块链节点共识方法

Also Published As

Publication number Publication date
CN108282394B (zh) 2020-09-25
US11343213B2 (en) 2022-05-24
CN108282394A (zh) 2018-07-13
US20200344191A1 (en) 2020-10-29

Similar Documents

Publication Publication Date Title
US7689652B2 (en) Using IP address and domain for email spam filtering
WO2019137290A1 (zh) 一种发信人的信誉值生成方法与过滤垃圾邮件的方法
US11068795B2 (en) Automatically predicting that a proposed electronic message is flagged based on a predicted hard bounce rate
US7899866B1 (en) Using message features and sender identity for email spam filtering
US8224905B2 (en) Spam filtration utilizing sender activity data
US8863244B2 (en) Communication abuse prevention
US8892673B1 (en) Hybrid challenge-response
US20050198171A1 (en) Managing electronic messages using contact information
US20150288634A1 (en) Removal From a Whitelist Based On an Extracted Email Address
US20050102366A1 (en) E-mail filter employing adaptive ruleset
US8745143B2 (en) Delaying inbound and outbound email messages
US20070180031A1 (en) Email Opt-out Enforcement
US20090327430A1 (en) Determining email filtering type based on sender classification
JP2004220613A5 (zh)
TW201036399A (en) Feedback loop for spam prevention
KR20080058415A (ko) 컴퓨터 시스템에서 평판에 의한 통신 처리 방법 및 평판에의한 통신 처리 시스템
Leiba et al. A Multifaceted Approach to Spam Reduction.
Agrawal et al. Origin (dynamic blacklisting) based spammer detection and spam mail filtering approach
CN106341303A (zh) 基于邮件用户行为的发件人信誉生成方法
Naksomboon et al. Considering behavior of sender in spam mail detection
CN1696949A (zh) 接收/发送服务器反垃圾邮件的方法以及反垃圾邮件系统
Erickson et al. The Effectiveness of Whitelisting: a User-Study.
CN113726806A (zh) 一种bec邮件检测方法、装置、系统及可读存储介质
Yasin Spam Reduction by using E-mail History and Authentication (SREHA)
Sarafijanovic et al. Resolving FP-TP conflict in digest-based collaborative spam detection by use of negative selection algorithm

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19738132

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19738132

Country of ref document: EP

Kind code of ref document: A1