JPH09168028A - Method and device for deciding distribution destination of electronic mail - Google Patents

Abstract

PROBLEM TO BE SOLVED: To distribute an electronic mail over a wide range without distribution to all members by deciding a distribution destination based on a keyword database section and a user classification database section and providing an output of the result to an electronic mail system. SOLUTION: The user informs a user name to an access information extract section 1 to set up connection to a news server. An access recording section 2 calculates frequency of access and records the result based on information obtained for the use of the user. Prior to decision of distribution, a keyword for each user to the keyword database section 5 is registered. A sender of an electronic mail makes a request of decision of distribution destination to a host electronic mail system 7 by which the sender desires to decide recipients. Then a user extract section 6 obtains users for distribution based on similarity of users obtained by the keyword user classification database section 4 registered by the keyword database section 5 based on the extracted condition and provides the result as a distribution of the original message to the electronic mail system 7.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an electronic mail system operated by using an information processing apparatus and its network, and more particularly to a method and apparatus for determining a distribution destination of broadcast mail and the like.

[0002]

2. Description of the Related Art E-mail is widely used for electronically exchanging messages between Unix systems, the Internet, LAN, and more simply, users within the same host computer. The advantages of these e-mails include the short communication time and ease of broadcast transmission. Therefore, depending on the user of the electronic mail system, a message may be flooded because it is included in the address of the broadcast mail.

As a result, many "trash mails" (junk mail: junk mail) are delivered to the recipients of the mail.
l) will arrive. The term "garbage mail" as used herein refers to an email whose content is of little interest to the recipient.

[0004] Japanese Patent Laid-Open No. Sho 63-1 discloses a technique for filtering out these mails and helping the users of the mail system organize information.
Means such as 3545 (hereinafter referred to as Document 1), JP-A-3-229545 (Reference 2), and JP-A-4-130839 (Reference 3) have been considered to limit distribution destinations when a broadcast mail is transmitted. In addition, various means have been devised for a method called filtering, such as removing unnecessary mail on the receiving side or lowering the display priority.

In order to reduce the flow rate of "garbage mail" by limiting distribution destinations in the present invention, Documents 1 and 2 will be described as prior arts.

According to the invention "keyword search method for electronic mail" of Document 1, a user registers a keyword in advance, searches the sent electronic mail for a keyword, and receives a matching one.

The invention "document distribution method" of Document 2 is a distribution obtained by creating a keyword table in which distribution destinations corresponding to each keyword are set in advance and drawing the keyword table with the keywords attached to the document to be transmitted. The method is to distribute the document first. As this document states that it can be distributed as an email message, the word "document" will be used synonymously with email hereinafter unless otherwise noted.

To further summarize, although there is a difference in whether the document is searched for a keyword or the keyword is attached to the document in advance in the inventions of Documents 1 and 2, the keyword is used in advance and associated with the keyword in advance. It can be reduced to distribution to the specified distribution destination.

[0009]

However, since the inventions of Documents 1 and 2 are methods of registering keywords in advance, in order to receive an e-mail that matches interests, interests, tasks in charge, etc., It is necessary to constantly update the keywords.

In the keyword search, a synonym (thesaurus) may be used to expand the search range. However, the synonym dictionary does not always include the keyword relating to the current field of interest.

An object of the present invention is to solve the above problems.

[0012]

A method of deciding a distribution destination of an electronic mail system implemented on an information processing apparatus, in which a user of the information processing apparatus extracts information for accessing a network An information extraction unit, an access recording unit that records the access information extracted by the access information extraction unit, and an access record for each user held by the access recording unit are used to obtain the similarity of the access states between users. A user classification unit, a user classification database unit that records the user classification obtained by the user classification unit, a keyword database unit that registers a keyword for each user, and a keyword database that reads a designated keyword of a distribution destination from an electronic mail system. Section and the user classification database section determine the distribution destination and distribute to the e-mail system And wherein the use of user extracting unit for outputting as.

In the present invention, the situation in which the user of the information processing apparatus accesses a document or other information resource on the apparatus or another apparatus connected via a network is recorded together with the time, and the While measuring the time transition, the degree of similarity between users is calculated from the situation, and the users with similar interests are searched for as the destination of the electronic mail.

[0014]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention. Hereinafter, it will be described as being mounted on the UNIX system.

The access information extraction section 1 is a section for extracting the network news of the user of the information processing apparatus and the access status of network resources such as FTP (File Transfer Protocol) and the World Wide Web. Specifically, it is resident in the system. It is realized as a daemon program.

The extraction of access information will be described by taking network news as an example. Here, the user name, the news group read, and the number of news articles read are obtained.

The transfer of the article of network news is performed by the procedure defined in the transfer protocol (hereinafter referred to as NNTP), which is described in detail in the document RFC977. Also, there are types of nntpd, innd, etc. in the news transfer daemon using NNTP, but here, a method of mediating the totalizing program between the news reader client and the news server will be described.

When reading a news article using the NNTP connection, the tabulation program waits for a socket connection request from the news reader program. If there is a socket connection request, the tallying program issues a connection request to the news server, and after establishing the connection with the news server, the tallying program establishes a connection with the news reader.

Although not specified in RC977, a widely used NNTP command is AUTHINFO. In this command, AUTHINFO is followed by US
Access restriction is realized by sending ERS and user name, PASSINFO and PIN after AUTHINFO.

The tally program relays the communication between the news reader and the news server, and the AUTHI
When sent from the NFO command, the user can be known from the USER and user name following the command.

When reading an actual article by NNTP, a GROUP command followed by a group name
After selecting the newsgroup, the contents of the article are transferred using the ARTICLE command or the BODY command. Therefore, in the total command, when the GROUP command is sent from the news reader, the following news group name can be known. Furthermore, ARTICLE
Alternatively, by counting the number of article transfers by the BODY command, the number of articles actually read can be obtained for each news group.

The above operation is shown in the flow chart of FIG.

In the above, the network news is taken as an example, but by changing the corresponding port number and protocol, access information about other network resources can be obtained. In that case, the objects that can be aggregated may differ. For example, HTTPP used in WWW
In the case of, URL (Uniform Resource
e Locator) only the address pointed to,
And so on.

The access recording unit 2 records the access information extracted by the access information extracting unit 1. The access information is recorded for each user, and at least the access frequency for each news group is recorded. Other information that may be recorded may include the date and time of last access by the user or its equivalent data, and the date and time of last access by the user or its equivalent data for each newsgroup.

As such a recording method, an independent file is created for each user, and the last access date and time is required at the head of the file, followed by the information of the news group name, last access time of the news group, and access frequency. There are ways to describe minutes. An example of this file is shown in FIG.
The first line is the last access time. This number is the total number of seconds since 0:00:00 on January 1, 1970, Greenwich Mean Time.

The second and subsequent lines show the news group name, the last access time and the access frequency of the news group, which are represented in the same manner as the first line for each line.

The method of calculating the access frequency is performed as follows, for example. For a news group, the number of articles read by the access information extraction unit is N, the current time is T, the access frequency up to that time is M, the last access time is S, and a new access frequency F is as follows. To define

[0029]

[Equation 1]

Here,

[0031]

[Outside 1]

Is the contribution ratio of the previous access frequency, τ
Is the time constant until it is halved. Also, α is a positive number that is the contribution rate of the number of newly read articles. By increasing τ and further decreasing α, it is possible to obtain the short-term access number, that is, the short-term access frequency, and by decreasing τ and increasing α, it is possible to obtain the long-term access frequency.

The user classifying unit 3 includes the access recording unit 2
Calculate the similarity between one of the users recorded in and other users. At this time, the user to pay attention to may be designated from the outside, or may pay attention to all the users recorded in the access recording unit 2 one by one. The index of the similarity is calculated by the following formula, for example.

[0034]

(Equation 2)

However, v, u, A, and N are V = {v _i | News group accessed by user v and access frequency} U = {u _i | News group accessed by user u and access frequency} A = V∩U However, the product of sets is applicable only to newsgroups. The number of elements of N ... U (3). Equation (3) is an operation that results only in newsgroups that are common to both parties. Instead of calculating the product of the set by the formula (3), the sum of the set is calculated, and for a news group which exists only on one side, the access frequency is calculated as 0, and the like. Further, the following equation may be used instead of the equation (2).

[0036]

(Equation 3)

The user classification database unit 4 records and holds the similarity between users calculated by the user classification unit 3. These similarities are stored in a format in which it is possible to reproduce which of the users is what the similarities of the users are. FIG. 4 is an example of such a format. Here, the information about the similarity from which user is recorded as a file name,
Inside the file, the user name and the degree of similarity with the user are written line by line.

The keyword database unit 5 records the keywords representing the area of interest for each user. Also, a plurality of keywords may be described in the form of a logical expression. FIG. 5 is an example of the description in such a database. In each line, the user name is written first, and ":" is used as a delimiter to describe the logical expressions after that. Here, “*” represents a logical product, “+” represents a logical sum, and “!” Represents negation.

The electronic mail system 7 is an electronic mail system implemented on the UNIX (registered trademark) system, and a message handled by this system is an indirect message in addition to at least a header and a body in which a recipient can be designated. Has fields that can be specified for dynamic distribution. FIG. 6 shows an example of such a message. In the "To:" field, specify the delivery address of the mail and click "X-Distribute-c
The conditions for specifying the distribution destination are described in the “onions:” column.

The user extraction unit 6 receives the mail with the distribution destination designation condition from the electronic mail system 7, expands the mail to a specific distribution destination, and retransmits it.

The user extraction unit 6 and the electronic mail system 7
Connection with is done as follows. For example, from the electronic mail system 7 to the user extraction unit 6, the mail alias function of the electronic mail system 7 can be used to connect to a processing program that realizes the user extraction unit 6. Figure 7 shows BSD U
It is an example of setting a mail alias in a NX host.
These alias configuration files are usually / etc / aliases or / usr / li on UNIX hosts.
It is often implemented as b / aliases.

In this example, "user-" on that host
By sending to the address "extract",
The executable file "/ usr / local / bin / extract" is started, and the message body of the mail is the program "/ usr / loca" from the standard input.
1 / bin / extract ”.

Further, since the connection from the user extraction unit 6 to the electronic mail system 7 is a mail transmission, various connection methods are possible. For example, there are a method of starting a mail sending program from the inside of the program that realizes the user extraction unit 6, and a method of making an SMTP connection from the inside of the program by a socket. Details of this SMTP protocol are detailed in the document RFC821.

The user extraction unit 6 is an electronic mail system 7
Thus, the message indirectly specifying the distribution destination is received, and the specific distribution destination is determined using the keyword database unit 4 and the user classification database unit 5. The procedure of this determination is as shown in the flowchart of FIG.

Here, an example will be given below in which a set operation is used to specify the distribution destination. keyword
Let (a) be a set of user names satisfying the keyword a in the keyword expression registered in the keyword database unit 5. Also, closeuser (i, r) is a union of users whose similarity is within r from each set i of user names.

The designation of the distribution destination is keyword () and cl
It is performed by a polynomial in which the riser () is combined. For example, as an operator, "*" is the product of sets, "+" is the sum of sets,
Let “!” Be the complement. keyword () and close
The specific value of the user () is obtained by searching the keyword database unit 5 and the user classification database unit 4, respectively.

The state of the apparatus using each of these parts at the time of execution will be described below.

In this apparatus, the operation of extracting the user's access information centering on the access information extracting section 1 and the operation of determining the distribution destination centering on the user classification database section 4 operate almost in parallel. In the following, first, access information extraction will be described.

First, before the operation, some initial settings are made. In this initial setting, the system administrator or operator performs the selection of the protocol for collecting the access information of the user in the access information extraction unit 1 and the setting of the time constant in the access recording unit 2. Hereinafter, in this embodiment, the protocol is NN
TP, and the time constant advances as it is.

Users of this distribution system access network news using their respective network news readers. The case where each user accesses will be explained as an individual case. In the actual implementation, each time a connection request is made from a single daemon program, a duplicate can be created by using the UNIX system's FORK system call, and each request can be processed independently. Does not lose generality.

The user informs the access information extraction unit 1 of his / her user name through the AUTHINFO command through the news reader and establishes a connection with the news server. After that, each time the user reads a news article using the NNTP command, it holds the read news group, the number of articles, the last read time, and the like. When the connection is completed, control is transferred to the access recording unit 2.

The access recording unit 2 calculates and records the access frequency based on the above-mentioned information obtained for each use of the user, that is, the number of read news groups and articles.

The operation of determining the distribution destination will be described.

Prior to the operation of determining the distribution destination, the keyword for each user is registered in the keyword database unit 5. This registration work may be performed at the start of use, and it is not necessary to perform the registration each time the distribution destination is determined. Also, since the registration keyword can be changed asynchronously with the distribution destination determination, it is not mentioned here.

The sender of the electronic mail issues a request for determining the distribution destination to the electronic mail system 7 of the host who wants to determine the recipient through the address set in FIG.

The user extraction unit 6 extracts the distribution destination specifying condition in the message received from the electronic mail system 7. Based on the extracted conditions, the users to be distributed are obtained based on the keyword registered in the keyword database unit 5 and the similarity between the users obtained from the user classification database unit 4, and as the distribution address of the original message,
Output to the electronic mail system 7 and send.

[0057]

As described above, according to the present invention, e-mail can be distributed in a sufficiently wide range without being distributed to all the users, and similar interest can be achieved without updating the keyword indicating the interest. It can be distributed to users.

[Brief description of the drawings]

FIG. 1 is an embodiment of an apparatus for realizing an electronic mail distribution destination determining apparatus of the present invention.

FIG. 2 is a procedure for acquiring a user name and a news group name from an NNTP connection.

FIG. 3 is an example of a database that records an access state for each user.

FIG. 4 is an example of a file that holds the degree of similarity between users.

FIG. 5 is an example of an interest area keyword file for each user.

FIG. 6 is an example of an e-mail header used when sending an e-mail using this system.

FIG. 7 is an example of description when describing this service on the host side.

FIG. 8 is a procedure for determining a specific distribution destination from a distribution destination designation expression.

[Explanation of symbols]

 1 Access Information Extraction Section 2 Access Recording Section 3 User Classification Section 4 User Classification Database Section 5 Keyword Database Section 6 User Extraction Section 7 E-mail System

Claims (2)

[Claims] 1. An electronic mail for determining a similarity of an access state between users based on a situation where a user accesses a network, searching for a user having similar interest from the similarity, and determining a distribution destination of the electronic mail. How to determine the distribution destination. 2. An access information extraction unit for extracting information for a user to access a network, an access recording unit for recording the access information extracted by the access information extraction unit, and each user held by the access recording unit. A user classification unit that obtains the similarity of the access states between users using the access record, a user classification database unit that records the user classification obtained by the user classification unit, and a keyword database that registers keywords for each user. And a user extraction unit that reads a designated keyword of a distribution destination from the electronic mail system, determines a distribution destination from the keyword database unit and the user classification database unit, and outputs the distribution destination to the electronic mail system. , E-mail distribution destination determining device.