KR101919598B1 - E-mail transmission rule reasoning system using virtual session analysis - Google Patents

Abstract

In order to cope with the case where the mass mail sent normally is classified as spam by the spam mail blocking rule and the mail sending fails, the virtual session analysis that increases the sending success rate by deducing the blocking rule for each receiving domain through the virtual session analysis The present invention relates to an e-mail sending control rule inference system through an e-mail sending log, which can define the start and end of mail sending for each domain as a virtual session, analyze the mail sending rules, So that it is possible to adaptively predict an optimal dispatch rule that can be performed.

Description

TECHNICAL FIELD [0001] The present invention relates to an e-mail transmission control rule inference system using virtual session analysis,

The present invention relates to an e-mail sending control rule inferring system through virtual session analysis, and more particularly, to a system for automatically sending e-mail, And more particularly, to an e-mail sending control rule inferencing system through virtual session analysis that increases the sending success rate by deducing blocking rules through virtual session analysis.

As marketing of e-mail has been rapidly increasing due to the various communication means, there has been a great deal of research and policy limitations for filtering such spam e-mails as the damage caused by spam e-mails, which are transmitted unilaterally, .

Various e-mail service providers introduce such spam filtering technologies and policies to block e-mails that are judged to be spam e-mail, thereby improving the satisfaction of subscribers and reducing unnecessary burden. However, It is difficult to distinguish between spam and normal recipients.

Therefore, it may happen that the mail sending is blocked by the spam mail blocking policy even for the sending of the normal mail receiving the receiving consent of the receiver. For example, a large Internet shopping mall may block regular e-mails or coupon e-mails sent to members who agree to receive e-mails, and the recipient may not receive the e-mails. Also, It may happen that you do not receive the service.

However, large-scale portals and professional hosting companies that provide e-mail services are subject to a variety of spam-blocking rules and policies. In addition to not disclosing spam-blocking rules and policies to spammers, Since rules and policies vary, there are more and more cases in which sending fails in normal bulk e-mail delivery. Failure to send such normal emails is increasingly problematic because it restricts legitimate rights of recipients.

Korean Unexamined Patent Application Publication No. 10-2002-0029526, entitled " Method and System for High-Speed Mass E-mail Transmission " distinguishes between broadcast mail and individual mail for bulk mail delivery, In parallel to perform an electronic mail sending operation, thereby sending a large amount of electronic mail at high speed.

Korean Patent Laid-Open No. 10-2003-0017702, 'Mail Processing System' also discloses a distributed processing method for sending a large amount of e-mails, and increases the load of the user authentication process and the processing speed reduction A user authentication and a search operation are simplified by applying a hash algorithm, thereby reducing an authentication load and reducing a mail sending process time.

Until the spam mail policy was strengthened, more attention was paid to the sending speed than the success rate, so the distributed sending process to increase the number of sending times per hour was mainly studied. However, this method is not practical because the mail sending success rate is low due to the strengthening of the spam mail policy This is not the way.

Research has been started to increase the success rate of mail delivery rather than the recent high-speed mail forwarding and rapid mail forwarding, but a proper solution has not yet been proposed. For example, there is a method of sending a mass e-mail by setting a fixed sending rule considering spam e-mail blocking rules, and collecting only unsuccessfully mailed e-mail again. However, if the e-mail sending fails due to the spam e-mail blocking policy and the success rate is lowered Even if the mail is sent again, the spam mail blocking policy will fail to send mail again, so the effect of improving the success rate is insufficient.

As a result, in order to increase the sending success rate, it is necessary to apply an extremely low level of sending rule so as not to block the spam mail as much as possible, or to send a large amount of e-mail in an inefficient environment There is a problem that the delivery speed is low and the management cost is high.

Accordingly, there is a need for a new mass mailing system and method capable of automatically sending a spam mail blocking policy to a recipient while avoiding an optimal level of spam mail blocking policy in normal mass mail.

Korean Patent Publication No. 10-2002-0029526 Korean Patent Publication No. 10-2003-0017702

The object of embodiments of the present invention to solve the above problems is to define start and end of mail sending for each domain as a virtual session on the basis of an e-mail sending log, analyze the mail session, And to provide an e-mail sending control rule inference system through virtual session analysis that can adaptively predict an optimal sending rule that can be successfully transmitted.

Another object of the embodiments of the present invention is to analyze the life cycle of each virtual session by dividing the dispatch log into virtual sessions and to compare the cause of the dispatch failure with the cumulative information by dividing the cause of the session end based on the predetermined element We propose an e-mail sending control rule inferencing system by analyzing the virtual session by analyzing the exact factors which are selected as the actual problems and optimizing them by repeatedly executing them in parallel to optimize the optimal sending rule that can be successfully sent .

Another object of the embodiments of the present invention is to verify the cause of the mail failure based on the individual elements by cumulative storage for each receiving domain through the virtual session analysis, to optimize using the stored analysis information, The inference information is provided to the domain sending control unit for controlling the sending of the domain by domain of the mail sending system to manage the sending rules so that the mail sending success rate of the mail sending system can be increased, .

In order to achieve the above object, an e-mail sending control rule inferring system based on virtual session analysis according to an exemplary embodiment of the present invention uses a sending control rule inferring result for predetermined mail blocking elements, Mail sending control rule inferring system for providing an inbound sending control rule inference result to a mail sending system which determines a mail sending value by a domain and determining a sending time point, To reconstruct the start and end of mail delivery to one domain as a virtual session, to determine information about the pre-established mail blocking elements during the session's life cycle and the cause of failure including the spam classification And cumulatively managing session information; Information about the mail blocking elements for the normal session and the spam mail session for the predetermined domain based on the session information of the log analyzing unit and if the value of the mail blocking element of the normal session is larger than the spam mail session, The blocking element adjusts the spam session value by excluding it from the spam classification item and determines the sending delay time according to the spam mail classification through the subsequent success session confirmation to send the information about the mail blocking elements and the sending delay time to the mail sending system And an inference unit for providing the inference unit.

The log analyzing unit aggregates the log for each domain to determine whether the mail sending action is stopped from the first mail sending start time and the mail sending is not performed again for a predetermined time, or when the mail sending failure classification is the spam mail classification, So that a virtual session can be constructed in which the period is the life cycle. The inference unit compares the spam mail session with the normal session to exclude an item whose mail blocking factor value of the normal session is larger than the value of the mail blocking factor of the spam mail session, And provides it to the mail sending system as a permitted value of the corresponding exclusion item.

delete

delete

delete

delete

delete

delete

delete

delete

In the e-mail sending control rule inference system according to the embodiment of the present invention, the start and end of mail sending for each domain is defined as a virtual session based on the e-mail sending log, So that it is possible to adaptively predict an optimal dispatch rule that can be successfully dispatched.

The e-mail sending control rule inferring system according to the embodiment of the present invention analyzes the life cycle of each virtual session by dividing the sending log into virtual sessions, classifies the cause of the session end according to a predetermined element, , It is possible to select the elements that are not the cause of dispatch failure and to identify the exact factors that are the actual problems and then optimize them by repeatedly executing them in parallel to adaptively predict the optimal dispatch rule that can be successfully dispatched, Can be increased.

The e-mail sending control rule inference system through the virtual session analysis according to the embodiment of the present invention confirms the cause of the mail failure based on the individual elements based on the virtual session analysis, accumulates the messages according to the receiving domain, and utilizes the stored analysis information And provides the optimized forwarding rule inference information to the domain forwarding control unit for controlling the forwarding of each domain of the mail forwarding system to manage the forwarding rule, thereby improving the mail forwarding success rate of the mail forwarding system .

1 is a configuration example of a conventional e-mail sending system.
2 is a configuration diagram of a mass-mail sending system according to an embodiment of the present invention;
3 is an example of an email sending result log according to an embodiment of the present invention.
4 is a configuration diagram of an inference engine unit according to an embodiment of the present invention;
5 is a flowchart illustrating an example of a method of operating a log analyzer according to an embodiment of the present invention.
6 is a flowchart for explaining an example of a reasoning unit operating method according to an embodiment of the present invention;
FIG. 7 is an example of an MX domain for explaining a delay estimation of a transmission delay time according to an embodiment of the present invention; FIG.
8 is a configuration diagram of a mail sending system according to an embodiment of the present invention;
9 is a flowchart for explaining an example of a method of applying a sending control rule of a domain sending control unit according to an embodiment of the present invention.
FIG. 10 is a flowchart illustrating an example of a configuration method of a sending environment of a domain controller according to an embodiment of the present invention. FIG.
Figure 11 is a graphical representation of the performance of a bulk email delivery system and a conventional email delivery system in accordance with an embodiment of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, features and advantages of the invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.

1 is a configuration example of a conventional e-mail sending system. As shown in the figure, when a user creates and sends an e-mail, a message to be delivered through a MUA (Mail User Agent) 11, which is a client program installed in the client terminal 10, The sending MTA 12 sends an e-mail message to the receiving MTA 22 and sends the e-mail message to the receiving MTA 22, And stores it in the mail database 23.

The POP (Post Office Protocol) / IMAP (Internet Message Access Protocol) server 24 connected to the MUA 21 of the server receiver client terminal 20 receives the e-mail stored in the mail database 23, And the recipient confirms the received mail through this.

These e-mails provide information or data using computer-mediated trust-based SMTP, so e-mail recipients are guaranteed to receive messages sent by the sender unconditionally. Therefore, in the case of the existing mass mail sending system, in order to increase the sending speed, the MTA is distributed in a large amount and the mail is sent. However, according to the spam mail blocking policy for filtering the spam mail, I can not.

Currently, various spam blocking techniques are applied to different portal or email service providers. These spam blocking technologies are classified into content filtering technology for checking spam mail by analyzing mail contents, And a technique for controlling the process of delivering to the destination.

In the case of normal mail transmission, spam filtering is often performed by a delivery process control technique rather than content filtering. As a policy and technique for controlling the delivery process by filtering the spam mail, a real time blackhole list (RBL), an SPF There are technologies such as Sender Policy Framework policy, Open Relay blocking, Sending restriction, Graylisting, Tarpiting, CR (Challenge Response), Blocking of port 25 of dynamic IP users .

Among them, the RBL or SPF policy can be avoided when a normal sender transmits an e-mail received by the receiver. However, various spam filtering techniques, such as Blocking Open Relay, Limiting Sending, Graylisting, Tarpiting, Challenge Response (CR), Blocking Port 25 of Dynamic IP Users, Transmission can still be a problem in the case of mass e-mail transmission. However, due to the delay in transmission or the inconvenience of recipients, it is difficult for some of them to be adopted by a large portal or a large e-mail service provider.

Therefore, SPF and RBL policies have been applied in most cases. As a result, spam mail blocking policies and techniques have been analyzed based on representative portals. As a result, spam and RBL policies have been applied. Major spam And it is used as a mail blocking factor.

Of course, the policies applied to each portal are somewhat different, and the criteria are also different, and there is a possibility of the dynamic change of the reference, so it is difficult to uniformly apply the fixed value to the value of the corresponding domain by measuring the fixed value.

Therefore, in the embodiment of the present invention, it is possible to infer the mail sending control rule by applying an inference engine capable of inferring the optimal value of the mail blocking elements according to different criteria for each recipient domain, We propose a bulk e-mail sending system that improves the success rate of each domain by controlling sending by domain, and enables the user to infer the optimal value by actively responding even if the mail blocking factors are variable, thereby increasing the sending success rate even in mass e-mail sending .

FIG. 2 is a block diagram of a bulk email sending system according to an embodiment of the present invention. Referring to FIG.

As shown in the figure, the bulk email sending system includes an inference engine unit 200 for analyzing a sending result log and providing a sending control rule inferring result including an optimal value for a predetermined mail blocking element and a sending delay time, The mail sending system receiving the sending control rule inferring result from the inferring engine unit 200 applies the sending control rule for each domain on the basis of this and sends the email to the inference engine unit 200 do.

The inference engine unit 200 acquires the transmission result log of the domain-specific e-mail transmission list from the log management unit 140, reconstructs the start and end of the mail transmission for one domain as a virtual session, A log analyzer 220 for cumulatively managing session information in the analysis database 230, and a log analyzer 220 for analyzing a value of a mail blocking element of a normal session for a predetermined domain based on the session information of the log analyzer 220, The value of the allowed mail blocking element is obtained by excluding the corresponding mail blocking element from the spam classification item, and the sending delay time according to the spam mail classification is determined through the subsequent success session confirmation, And an inference unit 210 for providing the mail sending system 100 with a result of sending control rule inference including time.

The mail sending system 100 includes a mailing list generating unit 110 for receiving sender information and contents and dividing them by domain to generate a mailing list 111 and a mailing list generating unit 110 for providing a mailing list And a forwarding control unit 120 for controlling forwarding for each domain according to a forwarding rule inferring result received from the inference engine unit 200 and for providing a result of forwarding to a log.

The transmission control unit 120 includes a list distribution unit 121 for dividing the transmission list 111 provided by the transmission list generation unit 110 according to a domain and providing the distribution list 121 for each domain provided by the list distribution unit 121 A domain controller 122 which manages the shipment list information as a queue and confirms the forwarding suitability based on the per-domain routing control rule received from the inference engine through the mail transmission result, and a domain-specific domain controller 122 List, inquires of the domain controller 122 about whether to send the message, and if it is possible to send the message, it sends the related information to the MTA 130, and provides the result to the domain controller 122, The transmission result log is provided to the log management unit 140 for use by the engine unit 200. If the transmission is completed, Delete (122) and, if sent to the spam classification idle it includes the sending domain control manager (123) to adjust the order in the list.

Through the above configuration, it is possible to increase the success rate by inferring the forwarding rule by analyzing the forwarding result log after sending the mail for each domain, controlling the forwarding rule for each domain based on the inference result, and sending the mail within the allowable range, Even if it fails, this information can be used for dispatch rule inferences, and as the information is cumulatively collected, more optimized mail dispatch rules can be set. Therefore, unlike the general method in which the success rate is low even in the case of the subsequent sending when the mail sending fails, the success rate in the subsequent sending can be high and the succeeding sending time can be optimized even if the mail sending fails.

2 according to an embodiment of the present invention. The inference engine unit 200 will be described in more detail with reference to FIG. The mail forwarding system 100 for controlling the forwarding by domain using the inference result will be described in detail.

The illustrated inference engine unit 200 acquires a mail log from the log database 141 of the log management unit 140 and analyzes it by the log analysis unit 220.

Although the log management unit 140 is shown separately from the mail sending system 100 and the reasoning engine unit 200 for convenience sake, the log management unit 140 may be included in one of the mail sending system 100 and the reasoning engine unit 200, More specifically, it may be included as a part of the domain forwarding control management unit 123.

In the embodiment of the present invention, the mail blocking factor obtained as a result of the analysis of the spam mail blocking policy of the major large e-mail service provider (portal, hosting company, etc.) as described above is determined as the main forwarding rule inferencing factor, The inference engine 200 and the mail forwarding system 100 optimized for mail forwarding control through inference values for such mail blocking elements are provided.

The feature of this spam mail blocking policy is that if a large amount of mail is sent for a short period of time, excessive connection is made to the mail receiving server, and mail is sent to an address (account) that does not exist, This blocking may be released after a certain time depending on the service provider after the reception of the mail is blocked. According to the characteristics of the spam mail policy, the inference engine 200 analyzes the dispatch log to determine the number of mail sent, the number of mail server connections, and the number of email errors for each domain, And provides these mail blocking elements and the sending delay time to the mail sending system 100.

Accordingly, the sending log includes individual unit information for easily checking the information on the mail receiving blocking element.

FIG. 3 shows an example of an e-mail sending result log according to an embodiment of the present invention. As shown in FIG. 3, a log is generated to include the information of Table 1.

Component Explanation MEM_SEQ Recipient sequence number EMAIL Recipient email address MX RECORD MX domain SENDER_IP Email sending server IP ERR1 Whether an error occurred before the domain connection ERR2 At the stage where an error occurred during the email sending phase ERR3 Error code received from the receiving mail server in case of failure SEND_DATE Email sending date ERRMSG Error message received from the receiving mail server upon failure

Here, MX (Mail eXchange) domain means a plurality of lower mail server domains set for an actual single receiving domain, and they are managed as a single domain.

As shown in Table 1, the shipment log basically shows information when one message is sent and various error information generated every time.

The inference engine unit 200 analyzes the shipment logs for the domain-specific shipment list to obtain the number of mail shots, the number of mail server connections, and the number of e-mail errors for each domain, and obtains the time for canceling the blocking after the mail reception is blocked .

FIG. 4 is a block diagram of an inference engine according to an embodiment of the present invention. As shown in FIG. 4, the log analyzer 220 compiles the log log information for each domain and converts it into one virtual session information for analysis And a virtual session switching analysis unit 221. [

The virtual session switching analysis unit 221 counts the transmission logs for each domain, and when the mail sending action is stopped from the first mail sending start time and the mail sending is not performed again for a preset time (for example, one minute) If the cause of the mail failure is a spam mail classification, it is determined that the session is terminated, and a virtual session is established in which the period is the life cycle.

The virtual session may include session data as shown in Table 2 below.

Component Explanation Domain Mx domain Start Time Session start time End Time Session end time Total Connect count Number of connections to the mail server during session maintenance Total Send Count Total Shipments Sent During Session Retention Session Time Session duration Connection Count Number of connections per minute Send count Shipments per minute Total Sucess count Number of successful messages Email Error Count The number of times the mail failure was caused by an email error. Session End Cause of the session termination (spam classification, stop sending mail)

Here, the number of logs is regarded as the number of logs during the life cycle of the session, and the number of connections to the e-mail server is obtained by dividing the number of mails sent from the number of mails connected to the mail server to obtain the number of connections per minute and the number of mails per minute. The number of e-mail errors is calculated by adding up the number of e-mail address failures (nonexistent addresses). The criterion for judging the e-mail address (account) error is that the response code of 5xy (permanently negative completion response) is received in the step of RCPT TO (command defined in the SMTP) during the e-mail sending step and the spam mail classification keyword is not present in the response message . The criterion for determining the spam mail classification, which is one of the causes of session termination, is shown in Table 3 below.

ERR2 Shipping phase Spam classification classification processing standard One connect All codes other than the success code (220) 2 HELO All codes other than success code 250 3 MAIL FROM All codes other than success code 250 4 RCPT TO 4yz code of temporary negative completion response
There is a spam classifier keyword in the response message 5 DATA 4yz code of temporary negative completion response 6 DOT All codes other than success code 250 7 QUIT Success regardless of response

The above-described method of creating a virtual session of the log analyzer 220 can be summarized as shown in the flowchart of FIG.

FIG. 5 is a flowchart illustrating an example of a method of operating a log analyzer according to an embodiment of the present invention. As shown in FIG. 5, the virtual session conversion analyzer 221 of the log analyzer 220 extracts a predetermined domain The number of connections and the number of shots per minute are calculated by analyzing the life cycle of the session.

If the session is terminated due to an error in the e-mail, e-mail error is counted to generate session data including the information, If the spam mail classification is based on the spam mail classification shown in Table 3, generates session data including the spam mail classification information and provides the session data to the analysis database 230 and the reasoning unit 210.

The reasoning unit 210 includes a blocking factor estimating unit 212 for obtaining an actual blocking factor and its optimal value among the mail blocking factors and a transmission delay time estimating unit 211 for obtaining an optimal delay up to the next transmission in the spam mail classification ).

FIG. 6 is a flowchart for explaining an operation procedure of the blocking factor estimator 212 according to an embodiment of the present invention. As shown in FIG. 6, when the session termination cause is a maximum number of shots, Collects the maximum number of e-mail block elements such as the number of connections, and collects the maximum value of e-mail blocking factors such as the maximum number of messages, the maximum number of e-mail errors, and the maximum number of connections in a session with a session termination cause.

If there is only a normal session without a spam session, the normal session value is returned as a session value. If there is only a spam session without a normal session, it also returns a spam session value as a session value. The maximum number of emails, the maximum number of email errors, and the maximum number of connections, and compares the spam session value with the normal session. If the value of the normal session is larger than the value of the spam mail session, ≪ / RTI > That is, if the value of the normal session is larger than the value of the spam mail session, it means that the current value of the corresponding item is allowed, and thus it is not judged as spam. Therefore, It can be judged as an optimum value. Therefore, the corresponding information is included in the transmission control rule information to be provided to the mail sending system as the allowable maximum value information.

Since the determination process can use the dispatch logs obtained through the plurality of IP distributed dispatch information for a plurality of MX domains, a plurality of virtual sessions for one domain can be substantially obtained for each MX domain and IP, The same analysis can be made more easily.

4, the reasoning unit 210 includes a transmission delay time estimating unit 211 for estimating a transmission delay time until the next normal transmission in the spam mail classification, and the transmission delay time estimating unit 211 ) Judges that the spam delay time is from the spam session to the start of the success session for the next same domain based on the spam session for the same domain.

Here, the transmission delay time estimating unit 211 determines the transmission delay time through the subsequent success session check, and uses the other IP information obtained for the same domain through the log or the log analyzer accumulation information, It is possible to determine the delay time of the entire domain or the MX domain of the corresponding domain by referring to all the information about the plurality of MX domains of the corresponding domain.

FIG. 7 shows an example of a case where a plurality of MX domains exist for one domain. In the case of hamail.net, 10 MX domains are configured, so that at least 10 different virtual session information for each MX domain is parallel The inference unit 210 may determine that the mail sending system 100 should attempt to send the mail with a different delay time for each of the plurality of MX domains in the same domain if necessary, And determines the shortest transmission delay time among the transmission delay times obtained in the successful MX domain as the transmission delay time for the corresponding domain.

Of course, in the case of distributed processing through a plurality of IPs, information on more virtual sessions can be obtained in parallel for each IP. Therefore, the blocking factor estimating unit 212 and the transmission delay time estimating unit 211 may calculate a plurality of MX It is possible to obtain mixed information such as normal and spam mail classification, success and failure through the multiple IPs for the domain and the distributed processing, and it is possible to optimize the mail blocking factor or the transmission delay time through such information, As a result, it is possible to estimate the optimal success rule for the domain which is frequently repeatedly sent.

The optimized forwarding control scheme of the mail forwarding system 100, which receives the forwarding control rule inferring result obtained by the inference engine unit 200, will be described with reference to FIG.

FIG. 8 is a block diagram of a bulk email sending system in which the configuration of the mail sending system 100 according to the embodiment of the present invention is shown in more detail.

As shown in the figure, the mail sending system 100 is divided into a sending list generator 110 and a sending controller 120.

The transmission list generation module 110 includes a data monitoring module 111 for checking a reception object file, a domain-specific transmission list generation module 112 for generating a transmission list 113 by dividing the reception object file into domains ).

As described above, the transmission control unit 120 includes a list distribution unit 121, a domain control unit 122, and a domain transmission control management unit 123. The domain control unit 122 includes a domain controller 122, A domain control unit 310 of the domain, and a domain sending control unit 320 which is configured in the domain control unit 310 and manages the shipment list by a queue.

If the domain controller 310 of the specific domain corresponding to the forwarding list of the list distributor 121 does not exist, the domain forwarding control manager 123 generates a corresponding domain controller 310 and manages the domain controller 310 together with other domain controllers (The list of recipients, mail contents, MX domain address information, and the like) to the MTA 130 upon receiving the sendable flag by querying the domain sending controller 320 in the domain controller 310, And provides the result to the corresponding domain control unit 310. [0064] Upon receiving the flag indicating that the sending is completed, the corresponding domain controller 310 is deleted from the list. When receiving the flag of the spam mail classification state when the domain sending control unit 320 in the domain control unit 310 managing the list inquires of the domain sending control unit 320 about the sending state, the corresponding domain control unit 310 adjusts the order in the list, It is usually postponed later, and if it is in the spam classification state, it moves to the last list.

Here, the domain controller 310, which is created to manage the forwarding for a specific domain, receives the forwarding rule inference value of the corresponding domain from the inferring engine 200 and sets and updates the current forwarding state based on the forwarding rule inference value. The received sending control rule reasoning value includes information on the optimal value for the mail blocking elements including the maximum number of shots, the maximum number of e-mail errors, the maximum number of accesses, and the sending delay time in determining the spam mail.

That is, the domain controller 310 receives the forwarding control rule inference value from the inference engine unit 200, manages the current forwarding state, analyzes the actual forwarding control rule according to the request of the domain forwarding control management unit 123, As a result, the recipient list, mail contents, and MX domain information management to be provided and sent are performed by the domain sending control unit 320 configured internally.

The domain forwarding control unit 320 includes a queue for storing MX domain information, recipient information, and contents in the forwarding list. The domain forwarding control unit 320 includes a queue for storing the forwarding state, the forwarding rule configured based on the forwarding control rule information for each domain, And the result is provided. An example of a method of confirming such a dispatch state is shown in the flowchart of FIG. 9 below.

9 is a flowchart for explaining an example of a transmission control rule application method of the domain transmission control unit 320 according to an embodiment of the present invention. As shown in the figure, A session delay flag (Session_delay return), a session maximum flag (Session_MAX return), and the like are determined based on whether the spam mail classification state, remaining recipient, current number of connections, , A flag indicating that the session is temporarily empty (Session_empty return), a completion flag ((Session_End return)), and a sendable flag (Session_send return).

According to the illustrated order, first, it is confirmed whether the current state is the spam mail classification state. If a message indicating that the mail is classified as spam is received from the receiving mail server, even if the mail is sent for a specific time, The system responds with a shipment delay flag during the shipment delay time when it is requested to check whether the shipment can be tried by processing in the spam mail classification mode for the time.

If it is not in the current spam mail classification state, it checks whether the mail recipient list exists, and if there is no remaining recipient in the list, it checks the time when there is no remaining recipient, and if there is no recipient that remains for a preset time (60 seconds in the illustrated case) The domain controller 310 deletes the corresponding domain controller 310 from the list by responding with a completion flag. If the time limit has not reached the predetermined time yet, the additional list may be further received, so that the flag temporarily indicates that the list is empty Reply.

If there are remaining recipients in the recipient list, check whether the current number of connections exceeds the allowed number specified in the sending control rule and whether the current number of shipment exceeds the allowed number of sending per minute specified in the sending control rule. If both of the cases are within the permissible number, the sending flag is transmitted so that the sending process is performed. (Parallel judgment criteria such as the same, big, Can be selected in various ways)

9, when the mail sending is possible and the mail is sent, the domain sending control managing unit 123 (substantially the MTA constituted by the corresponding domain sending control managing unit) transmits the sending result To the corresponding domain controller 310, and the domain controller 310 constructs or reconfigures (updates) the current sending state.

10 is a flowchart for explaining an example of a configuration of a sending environment of the domain controller 310 according to an embodiment of the present invention. As shown in FIG. 10, when the sending result is received, the sending number (number of connections) Determine if the result is a success, and if successful, increase the number of successes.

If the sending fails, the cause is classified according to the spam mail classification, the email error, and the server response timeout. In the spam mail classification, the current status is set to the delay mode so that the sending is delayed during the sending delay time according to the spam mail classification. The failure value is increased while the failure value is readjusted (cumulative increase).

If the cause of the failure is not a spam mail classification but a response is received during a predetermined response waiting time (Soket time out) when the cause of failure is a connection to the receiving mail server and waiting for a response, the response waiting time is further increased by a predetermined level If the response wait time is equal to or longer than the allowable response wait time, this also sets the current state to the delay mode.

Also, if the cause of the failure (not the spam classification) is an email error (the email address does not exist), the email error value is accumulated and the failure value is increased. If the accumulated email error is greater than or equal to the allowed number of email errors Also set the current state to delay mode.

If the cause of the failure is an e-mail error and the cumulative error count is equal to or greater than the allowed e-mail error count (or exceeded), or if the cumulative increase in response wait time over the server response time is greater than or equal to the allowed response time ) By setting the current state to delay mode (which can be set differently from the delay time by the spam mail classification) before being classified as spam mail, the possibility of being classified as spam mail is lowered, so that the delay time can be reduced.

As a result, as described above, the inference engine part deduces the spam mail blocking rule and utilizes the inferred mail sending rule information actively to utilize it in the mail sending system. E-mail can be sent to the state, which can increase the success rate of sending e-mail.

FIG. 11 is a graph illustrating the performance of a mass mail sending system according to an embodiment of the present invention and a conventional mass mail sending system. In the case of an email sending system according to an embodiment of the present invention, Since it is possible to infer sending control rule, the success rate of repeated sending increases, but it can be confirmed that the sending success rate is always the same in the existing system.

The foregoing and other objects, features and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings. However, the present invention is not limited to the above-described embodiments, and various changes and modifications may be made by those skilled in the art without departing from the scope of the present invention. .

100: mail sending system 110: sending list generating unit
120: sending control section 121: list distribution section
122: domain control unit 123: domain sending control management unit
130: MTA 140: Log management unit
200: inference engine 210:
220: log analysis unit 230: analysis database

Claims (1)

A rule for providing a rule-based reasoning result to a mail sending system for determining a value and a sending time point of the corresponding mail blocking elements by using the result of the rule-based inference for the predetermined mail blocking elements, An e-mail sending control rule inference system through session analysis,
Collecting log output results of the domain-specific email sending list to reconstitute the start and end of mail delivery for one domain into a virtual session, and store information about pre-established mail blocking elements during the lifetime of the virtual session And a spam mail classification; and cumulatively managing the cause of the failure as session information;
Information about the mail blocking elements for the normal session and the spam mail session for the predetermined domain based on the session information of the log analyzing unit and if the value of the mail blocking element of the normal session is larger than the spam mail session, The blocking element adjusts the spam session value by excluding it from the spam classification item and determines the sending delay time according to the spam mail classification through the subsequent success session confirmation to send the information about the mail blocking elements and the sending delay time to the mail sending system And an inference unit for providing,
The virtual session is judged to be the end of the session when the mail sending action is stopped from the first mail sending start time and the mail sending is not performed again for the predetermined time, or when the failure cause is the spam mail classification, And the life cycle,
The inference unit compares the spam mail session with the normal session to exclude an item whose mail blocking factor value of the normal session is larger than the value of the mail blocking factor of the spam mail session, And provides the email sending system with a permitted value of the corresponding excluded item to the email sending system.

Images