JP2885161B2 - Email classifier - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an electronic mail classification apparatus for classifying transmitted and received mail in an electronic mail system in which a plurality of users transmit and receive documents and data via a network and a computer.

[0002]

2. Description of the Related Art In a conventional e-mail system, e-mails transmitted and received by an individual user are collectively stored in a specific reception box or a transmission copy box. for that reason,
The user has to check each mail individually and classify them according to the contents. If the number of sent and received mails is large, it takes a very long time to sort and sort the mails. As a method for solving this, Japanese Patent Application Laid-Open Nos.
As in the inventions described in Japanese Patent No. 170642 and Japanese Patent Application Laid-Open No. 4-351134, it was possible to automatically classify transmitted and received mails according to a classification rule defined by an individual user. On the other hand, there are a method of automatically analyzing the contents of a mail to extract a keyword and a method of automatically learning a classification rule, as in the invention described in Japanese Patent Application Laid-Open No. 63-175965.

[0003]

In the conventional method of classifying incoming and outgoing mail according to the classification rules defined by the user, the user has to input the classification rules from a keyboard. Therefore, the user needs to learn the format of the classification rule in advance. Also, in recent years, as the range of use of electronic mail has expanded, the number of rules that must be entered by the user has increased and has become more complicated. Furthermore, the user's email classification criteria change over time, so the rules need to be modified each time. As described above, in the method of inputting the classification rules from the conventional keyboard, the cost for creating and changing the classification rules is increasing.

On the other hand, the method of automatically learning the classification rules from the contents of the mail requires an enormous vocabulary dictionary and background knowledge, and is therefore not suitable for use on a personal computer with limited disk capacity and processing speed. Absent. Further, since large-scale mail data is required to learn the classification rules, it is difficult to immediately respond to the user's classification criteria that change in a short period of time.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an e-mail classifying apparatus which allows a user to create and change a desired classification rule with simple operations without having to remember a complicated classification rule format. .

[0006]

According to a first aspect of the present invention, there is provided an electronic mail classifying apparatus, wherein a plurality of users transmit and receive documents and data via a network and a computer. A mail sending / receiving means for sending / receiving mail distributed by the user, a classification rule storage means for storing classification rules, a mail classification means for classifying transmitted / received mails using the classification rules stored in the classification rule storage means, A mail storage unit for storing mail data, a classification result display unit for displaying a classification result on the input / output device, and a classification example included in the classification example input by the user from the input / output device.
The information extracted from the incoming mail is
Presented to the user as a candidate, and
Receives the selection of the information to be used as the classification condition from the user.
And a classification rule creating means for creating a classification rule using the received information .

A second invention is characterized in that the classification rule creating means extracts information common to a plurality of mail data operated by a user and creates a set of classification rules satisfying the plurality of mail data. And

A third invention is characterized in that the mail classifying means re-determines the classification category of the mail stored in the mail storage means using the classification rules stored in the classification rule storage means.

In the first invention, a classification rule for classifying transmitted / received mails is a logical expression composed of mail data patterns as a condition part and a classification category name as a conclusion part, which is stored in the rule storage means. Have been.

When the mail sending / receiving means sends or receives an electronic mail, the mail classifying means determines the classification category of each mail using the classification rules stored in the mail storage means. Each mail data is stored in the mail storage means together with the classification category name. The classification result display means displays the mail stored in the mail storage means on the input / output device for each classification category.

When a user wants to classify a mail into a new category, the user inputs a mail to be classified into a new category from the mails displayed on the input / output device. The classification rule creating means extracts information such as “sender” and “title” from the content of the instructed mail, and presents it to the user as a candidate for the condition part of the classification rule,
The user selects information to be used as a classification condition for a newly created category. The classification rule creating means creates a classification rule using the selected information.

As described above, the user does not need to remember the classification rule format, and can create the desired classification category and the classification rule by only a simple operation of directly specifying the mail to be classified.

[0013] In the second invention, when a user specifies to classify a plurality of mails into the same category, the classification rule creating means extracts information common to the plurality of mails, and the user is most likely to select. Information is automatically generated as the classification condition, so that the user's labor can be reduced.

Next, in the third invention, when a new classification rule is added or the classification rule is changed due to the user changing the mail classification standard, the past data already stored in the mail storage means is stored. Mail sent and received can be re-classified according to the new classification criteria.

[0015]

Next, an embodiment of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of one embodiment of the first invention. Referring to FIG. 1, an e-mail classification device according to an embodiment of the first invention is to be classified into a mail transmission / reception unit 1 for transmitting / receiving a mail distributed via a network, and a classification category defined by a user. A classification rule storage unit 2 having a positive rule storage unit 21 storing mail conditions and a negative rule storage unit 22 storing mail conditions not to be classified into the classification category;
E-mail classification means 3 for classifying incoming and outgoing mails using the classification rules stored in the classification rule storage means, mail storage means 4 for storing the classification results and mail data, and classification result display for displaying the classification results on the input / output device and means 5, for classification example input by the user from the input device, the classification example
The information extracted from the included mail is classified into the condition part of the classification rule
To the user as a candidate, and among the presented candidates
From the user, select the information that you want to use as a classification condition from
And a classification rule creating means 6 for creating a classification rule using the received information . Reference numeral 7 denotes an input / output device connected to the classification result display means 5 and the classification rule creation means 6, and is composed of, for example, a CRT 71, a keyboard 72, and a pointing device 73.

FIG. 3 shows an example of mail data distributed to the network by the mail transmitting / receiving means 1. The mail data includes a sender (From), a destination (To), a destination of carbon copy (Cc), and a delivery date and time (Dat
e), fixed information such as a title (Subject) and the body.

FIG. 4 shows an example of the classification rules stored in the classification rule storage means 2. The classification rule is made up of a condition part and a conclusion part. The affirmation rule storage unit 21 stores a mail pattern as a condition part and a classification rule name for a mail that satisfies the mail pattern as a conclusion part. Is stored. For example, the first line of the affirmative rule storage unit in FIG. 4 is “If the destination is“ ski-clu
If the character string “b” is included, the mail is classified into the category “ski part.” The denial rule storage unit 22 stores the mail pattern as a condition part and the mail pattern as a conclusion part. For example, the first line of the denial rule storage unit in Fig. 4 includes "If the sender contains a character string" tanaka "," And the title includes a character string "bug", the mail is not classified into the category "tanaka".

When a mail is transmitted from the mail transmission / reception means 1 or a mail is received by the mail transmission / reception means 1, the mail classification means 3 uses the classification rules stored in the classification rule storage means 2 to classify the mail. To determine. A detailed classification flow will be described with reference to FIG.

For convenience of explanation, the rule stored in the positive rule storage is P _i (i = 1,..., N), and the rule stored in the negative rule storage is N _i (n = 1). ,
..., m). The classification category defined by the user is C _i (i = 1,..., T).

[0021] The mail classification means 3, first read the e-mail data from the mail transmitting and receiving means 1 (P1), to set a classification category set S _c of the mail to the empty set (P
2). Next, it is checked whether the mail satisfies the condition part of each affirmation rule P _i stored in the affirmation rule storage part (P4), and if satisfied, the classification category C written in the conclusion part of the rule Add _i to the classification category set S _c (P5). After all positive rules have been checked,
If _Sc is an empty set, the classification result of the mail is set as an empty set (P7). If S _c is not an empty set, the condition part of each negation rule N _j stored in the negation rule storage part is checked (P10). If the mail satisfies the condition part of the negation rule N _j , S _c to remove the classification category C _j from (P11). Each element of _Sc after the check of all the denial rules is completed becomes a category to which the mail is classified (P13). S _c obtained in this way
Is stored in the mail storage means 4 together with the mail data as the mail classification result.

The classification result display means 5 includes the mail storage means 4
When email data and classification results stored in are updated,
The new storage content is output to the input / output device 7. FIG. 6 shows an example of the classification result display. FIG. 6 shows the arrival date and time, sender, title, and category of each mail. “O” indicates the classification result of the received mail, and “X” indicates the classification result of the transmitted mail. “Unclassified” indicates that the corresponding category does not exist and the classification result is an empty set. For example, the mail on the first line in FIG. 6 indicates that the corresponding category does not exist. The mail in the fourth line indicates that the mail is classified into the category “important” and the category “material management”.

Next, a classification rule creation flow in the classification rule creation means 6 will be described with reference to FIG.

First, when a classification change request for the mail _Mk is input from the input / output device 7 from the user (T1), the classification rule creating means 6 inquires the user whether to add or delete the classification category of the mail. (T
2). In the case of "addition", it is further inquired whether a new category is to be created and classified or to be classified into an existing category (T3). If the answer from the user is "new", the routine extracts the mail _Mk and presents the list to the user (T4). The user selects and inputs information to be set as classification conditions from the list (T5). The classification rule creating means 6 presents the selected information to the user as a category name (T6). The user can confirm the category name and correct the category name C _new if necessary (T
7). Next, the standard information selected by the user is used as a condition part,
An affirmation rule with the category name corrected and confirmed by the user as the conclusion part is created and added to the affirmation rule storage unit 21 (T
8). The mail M _k stored in the mail storage means 4
C _new is added to the classification result of (T9).

On the other hand, if the answer from the user is "existing", an inquiry is made as to the category to be classified (T10).
When the category C _selected is input (T11), C
All positive rules and negative rules having _selected as the conclusion are collected from the classification rule storage means 2 (T12), and it is checked whether _Mk can be classified as C _selected when these existing classification rules are used (T13). If it is determined that can be classified into C _selected adds C _selected classification result mail M _k stored in the mail storage means 4.
If it is determined that the existing rules cannot be classified as C _selected , the routine extracts the standard information from M _k and presents the list to the user (T14). The user selects and inputs information to be set as the classification condition from the list (T15). next,
An affirmation rule is created by using the standard information selected by the user as the condition part and C _selected as the conclusion part, and stores the affirmation rule storage unit 21
(T16). Also, C _selected is added to the classification result of the mail _Mk stored in the mail storage means 4 (T17).

Next, returning to the classification change request from the user,
If the classification change type is "delete", the user is inquired about which category is excluded from the categories into which the mail _Mk is currently classified (T18). Category C from user
_{When omitted} is input (T19), all affirmative rules and negative rules having C _omitted in the conclusion part are collected from the classification rule storage means 2 (T20), and when these existing classification rules are used, _Mk becomes C _{It is} checked whether it can be classified as _omitted (T21). If it is determined that the mail cannot be classified as C _omitted , C _omitted is deleted from the classification result of the mail _Mk stored in the mail storage unit 4 (T25).
If it is determined that the information can be classified as C _omitted by the existing rule, the template information is extracted from _Mk and the list is presented to the user (T22). The user selects and inputs information to be set as the classification condition from the list (T23). Next, a denial rule is created with the standard information selected by the user as the condition part and C _omitted as the conclusion part, and added to the denial rule storage part 22 (T24). Also, C _omitted is deleted from the classification result of the mail _Mk stored in the mail storage unit 4.

The manner in which the classification rules are created will be specifically described with reference to FIG. FIG. 8 shows the mail data in the first and second lines of FIG. Also, assume that the rules stored in the classification rule storage means 2 are as shown in FIG.

When the user requests a change in the classification of the rule in the first line to the existing category "ski part", the classification rule creating means 6 converts information usable as classification conditions such as "sender" and "title" from the mail data. Extract and present the list to the user. The user selects from the list what he or she wants to use as the actual classification conditions. FIG. 9 shows an example of the extracted list. For example, when "destination of carbon copy" is selected in FIG.
"If the destination of the carbon copy is ski-club @
swl. co. jp is classified into the category “skiing section” ”. In addition, if it is desired to classify the e-mails sent by oneself among the e-mails related to “materials” into a new category “material reserve” instead of the category “material management”, the mail in the second line in FIG. , And adding a classification to a new category, for example, "If the title contains the character string" material "and" sender "is midori, it is not classified into the category" material management ". The negation rule, "If the title contains the string" Material "
If the “sender” is midori, the category is classified as “material reserve”.

As described above, the user inputs the mail that he / she actually wants to classify without knowing the format of the classification rule.
A classification rule can be created or corrected only by selecting an applicable classification condition candidate.

Next, the second invention will be described.

FIG. 2 is an explanatory diagram of one embodiment of the second invention. Referring to the figure, the classification rule creating means 6
It comprises a common pattern extraction unit 61 for extracting information common to a plurality of mail data input by the user, and a classification rule set creation unit 62 for creating a classification rule set that satisfies all of the plurality of mail data.

For example, a plurality of mails M _s (s
= 1,..., L). If the change type was an addition to the new category, a common pattern extraction unit 61 extracts the boilerplate information from each mail M _s, see if there is information that is common to all e-mail.
If there is common information, a classification category _Call having the information as the category name is newly created, and an affirmation rule using the common information as a classification condition is created. If there is no information common to all the mails, the information common to the largest number of mails is selected from all the standard information extracted from all the mails, and a positive rule is created. At this time, the value of the common information is used for the category name of the category C _most newly created. Classification rule set creation unit 62 extracts the mail that does not meet the positive rules created from the M _s, presents every single routine information pattern them to the user. When the user selects what he or she wants to use as the classification condition, an affirmation rule is created with category C _most as the conclusion.

The addition to the existing category and the deletion of the category can be performed in the same manner.

As described above, when the user inputs a plurality of mails as classification examples, common information can be extracted and a classification rule can be automatically created. Even if there is no information common to all the input e-mails, the rule can be created by selecting the common information covering the most e-mails, so that the user can save time and effort.

Next, the third invention will be described.

In the third invention, when the user instructs re-sorting, the mail classifying means 3 reads the mail already classified and stored in the mail storage means 4, and stores the mail in the classification rule storage means 2. Classification is performed using the classification rules stored in. The new classification result is stored in the mail storage unit 4.

As described above, when the contents of the classification rules stored in the classification rule storage means 2 are changed, not only mails transmitted and received after the change are classified according to the new classification criteria, but also those transmitted and received in the past. E-mails can be re-categorized according to the new classification criteria.

[0038]

As described above, according to the present invention, the user does not need to memorize the classification rule format, and the classification category desired by the user can be specified only by a simple operation of directly specifying the mail to be classified. Classification rules can be created and modified. For this reason, it is easy to create a new classification category or to change a previous classification standard. If the user specifies to classify multiple emails into the same category, the classification rule creation means extracts information common to multiple emails and automatically creates the information that the user is most likely to select as the classification condition. Therefore, the user's labor can be reduced. Further, when the classification criteria are changed, it is possible to easily reclassify previously transmitted / received emails according to the new criteria.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an embodiment of the first invention.

FIG. 2 is a configuration diagram of an embodiment of the second invention.

FIG. 3 is an example of transmitted / received mail data.

FIG. 4 is an example of a classification rule stored in a classification rule storage unit.

FIG. 5 is a flowchart showing the operation of the mail classification means.

FIG. 6 is an example of a classification result display.

FIG. 7 is a flowchart illustrating the operation of a classification rule creating unit.

FIG. 8 is a specific example of the mail data written in FIG. 6;

FIG. 9 is an example of a list of information extracted by a classification rule creating unit.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Mail transmission / reception means 2 Classification rule storage means 3 Mail classification means 4 Mail storage means 5 Classification result display means 6 Classification rule creation means 61 Common pattern extraction unit 62 Classification rule set creation unit 7 I / O device

Continuation of the front page (56) References JP-A-5-233706 (JP, A) JP-A-5-342272 (JP, A) JP-A-4-351134 (JP, A) JP-A-2-170642 (JP) JP-A-63-175965 (JP, A) JP-A-4-129446 (JP, A) JP-A-4-4640 (JP, A) JP-A-5-110593 (JP, A) 6-214789 (JP, A) JP-A-6-250949 (JP, A) JP-A-5-204975 (JP, A) Fumio Mizoguchi, Masahiro Yanagida, Hayato Owada, "E-mail Agent Using Inductive Learning" Design, ”Proceedings of the 12th Annual Conference of the Japan Society for Software Science and Technology, pp.277-280 (September 12, 1995) (58) Fields investigated (Int. Cl. ⁶ , DB name) 12/58

Claims (3)

(57) [Claims] 1. An apparatus for classifying transmitted and received mail in an electronic mail system in which a plurality of users transmit and receive documents and data via a network and a computer, comprising: (a) a mail transmitting / receiving means for transmitting / receiving a mail distributed via a network; (B) classification rule storage means for storing classification rules; (c) mail classification means for classifying transmitted and received mails using the classification rules stored in the classification rule storage means; and (d) classification results and mail data. Mail storage means for storage, (e) classification result display means for displaying the classification result on the input / output device, and (f) classification examples input by the user from the input / output device ,
Classification rules for information extracted from emails included in classification examples
Presented to the user as a candidate for the condition part of
Select the information you want to use as a classification condition from the candidates.
An e-mail classification device, comprising: a classification rule creating unit that receives from a user and creates a classification rule using the received information . 2. The classification rule creating means extracts information common to a plurality of mail data operated by a user and creates a set of classification rules that satisfies the plurality of mail data. 2. The e-mail classification device according to 1. 3. The mail classification unit according to claim 1, wherein the mail classification unit re-determines the classification category of the mail stored in the mail storage unit using the classification rule stored in the classification rule storage unit. An e-mail classification device according to claim 1.