JPH09330335A - Device and method for learning and classifying message - Google Patents

Abstract

PROBLEM TO BE SOLVED: To classify and store received messages and thereby, lighten the burden of a classifying process on a user, by learning a key word which is common to the messages as a new classification condition, and classifying and storing subsequent received messages in a storage means under the classification condition. SOLUTION: A storage control means 2 stores a received message in a corresponding classification of a 1st storage means 3 when the message meets classification conditions. When the received message meets none of the classification conditions, it is stored in a 2nd storage means 4. The message receiver when receiving the message stored in the 2nd storage means 4 opens the message to confirm its contents, and classifies the message as predetermined and stores it in the 1st storage means 3. A classification condition learning means 6 reads respective stored messages on the basis of respective classifications in the 1st storage means 3 and learns the classification conditions corresponding to the respective classifications on the basis of extracted key words.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to classification and storage of received messages in a messaging system represented by electronic mail and facsimile.

[0002]

2. Description of the Related Art For example, Japanese Patent Application Laid-Open No. 7-212392 discloses a conventional method for classifying received messages. This is an identifier I by classifying received e-mails according to preset importance such as subject and comment.
D is added and classified and accumulated. As another example, Japanese Patent Laid-Open No. 3-175763 has the same meaning. As described above, the conventional method adopts a method of assigning an identifier in advance on the receiving side or the transmitting side prior to receiving and classifying based on this.

Other than the messaging system, for example, in reclassification in a database disclosed in Japanese Patent Laid-Open No. 7-78186, there is a system in which a specific item is registered as an attribute and the classification process is performed by paying attention to this attribute.

[0004]

The conventional messaging device is configured as described above, and when automatically classifying received messages, a classification condition is given in advance with an identifier or the like, and this identifier or attribute is used. Since there is an increase in the number of classification items or detailed classification, the labor in advance becomes large, or classification becomes unpredictable if there is an unexpected classification.

The present invention has been made in order to solve the above problems. A classifying apparatus learns and classifies and stores subsequent received messages to obtain a classifying apparatus that reduces the burden of classification processing on the user. With the goal.

[0006]

A message learning classifying apparatus according to the present invention comprises a first accumulating means for classifying and accumulating a received message based on a classification condition which will be described later, and a received message which is not classified by this classification classification. When the second accumulation means for accumulating in the unlearned section and the message accumulated in the second accumulation means which is this unlearned section are classified into the section corresponding to the first accumulation section, The learned classification is provided with keyword extraction means for extracting a keyword in the received message and classification condition learning means for learning as a new classification condition a keyword common to a plurality of messages classified into a category corresponding to the first category. The conditions are held, and subsequent received messages are divided and stored in the first storage means based on this classification condition.

Furthermore, when a keyword conversion table that regards other terms as the same keyword is added to a specific keyword and the subsequent received messages are classified into the first storage means, refer to the keyword conversion table. Then, the range of common keywords is expanded and classified and accumulated.

[0008] Furthermore, when the held keywords as the learned classification conditions and arbitrary keywords are displayed, and the displayed arbitrary keywords are designated, the specified keywords are the learned classification condition keywords. In addition to the above, a classification condition changing means for expanding the classification condition is added, and when the received message thereafter is classified into the first accumulating means, the classification message is classified and accumulated according to the classification condition obtained by the classification condition changing means. I chose

Further, when a predetermined area extracting means for designating a predetermined area of the message is added and the subsequent received message is divided into the first accumulating means, the keyword of the predetermined area of the received message is classified. According to the above, it was divided and accumulated.

Further, a reception message list display means for extracting and displaying information on predetermined items of the reception message is added, and when requested, the above item information is extracted from the reception message and displayed as a reception message list. I decided to do it.

The message learning classification method according to the present invention includes a first accumulation step of dividing a received message according to a previously learned classification condition and a second accumulation step of dividing a received message into an unlearned classification when the received message cannot be classified according to the learned classification condition. When the accumulation step and the message accumulated in the unlearned section are classified into a predetermined section, the keyword extraction step of extracting the keyword in the received message of this predetermined section and the message classified into this predetermined section Also has a classification condition learning step of learning a keyword that appears in common as a new classification condition.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1. The message learning classification device according to the first embodiment of the present invention will be described. 1 is a configuration diagram of a messaging device according to a first embodiment of the present invention. In the figure, 1 is a receiving means for receiving a message from the other party, 2 is a storage control means for classifying the received message according to a classification condition corresponding to each classification category, and accumulating in the accumulating means, 3 is a first for accumulating the received message. 1 storage means, 4 is second storage means for storing received messages according to classification categories, 7 is keyword extraction means for extracting keywords from each received message, and 5 is each classification category for classifying received messages The classification condition holding means for holding the classification condition corresponding to is a classification condition learning means for determining or updating the classification condition for classifying the received message based on the keyword extracted from each message. FIG. 2 is a diagram showing an example of the keyword holding status of the classification condition holding means and the storage status of the message storage means. 3 to 5 are diagrams showing specific examples of received messages.

Next, the operation of the apparatus having the above configuration will be described. When the receiving means 1 receives the message, the storage control means 2 determines the matching between the received message and the classification condition corresponding to each classification category held in the classification condition holding means 5, and the received message is If the classification conditions are met, the data is stored in the corresponding classification section of the first storage means 3. If the received message does not match any of the classification conditions, it is stored in the second storage means 4. When the user of the messaging device, that is, the receiver of the message, receives the message stored in the second storage unit 4 that does not belong to any of the classification categories, the message is opened, the contents are confirmed, and then this message is displayed. The data is classified into one of the predetermined classification categories and stored in the first storage means 3. The classification condition learning means 6 reads each message accumulated for each classification based on each classification category in the first accumulation means 3, and the classification condition corresponding to each classification category based on the extracted keywords described later. To learn.

The learning of this classification condition will be described more specifically. For the first storage means 3, as shown in FIG.
For example, there are three classification categories, "fishing-related", "karaoke-related", and "golf-related", and the classification condition holding unit 5 already has the keywords "Karaoke" and "Karaoke-related" as the classification conditions. It is assumed that a classification condition of the keyword “golf” is set as the classification condition of the classification category “golf-related”, and a classification condition corresponding to the classification category of “fishing-related” is not set. In addition, here, the keyword extracting means 7 analyzes the character string in each message and outputs the Japanese "noun" as a keyword. The classification condition learning means extracts a keyword commonly used in each message from the keywords extracted from each message, and determines / updates the classification condition based on this. That is, a set of keywords extracted from the message A, the message B, ..., The message X is KA = {KA1, KA2, ...
., KA1}, KB = {KB1, KB2, ..., KB
m}, ..., KX = {KX1, KX2, ..., KX
n}, the product of each set is KA∩KB∩ ... ∩
The classification condition shall be determined based on KX.

When the receiving means 1 receives a message as shown in FIGS. 3 and 4, for example, the keyword extracting means 7 reads the message shown in FIG.
"Hello", "North", "now", "Sunday", to extract the "go" as a keyword, also, read the message shown in FIG. 4, "fishing", "Hello",
Extract “East”, “Meeting”, and “Notification” as keywords. Since the keywords extracted from none of the messages do not match the classification conditions of "karaoke" and "golf", the storage control means 2 stores these messages in the second storage means 4. The user who receives the message opens the two messages shown in FIGS. 3 and 4, determines the contents, and stores both messages in the classification item “fishing relation” of the first storage means 3. Classification condition learning means 6
At the proper timing in the first storage means 3,
"Fishing relation" 3 stored in the classification item of the keywords extracted from the message shown in FIG. 4 { "fishing", "Hello", "North", "turn", "Sunday", "go"} and { "fishing", "Hello",
"East", "Games", and the extracted "Notice" is used in common to both the message from} "Hello," the two keywords of "fishing", and classification conditions the "Hello" * "fishing", which Is held in the classification condition holding means. Here, "*" means logical AND, "Hello" * "fishing" refers to a condition that includes both keywords of "Hello" and "fishing".

Further, in this state, the user who receives the new message as shown in FIG.
In the case of accumulating in the classification item “fishing relation” in the accumulating means 3 of, the keyword extracting means 7 extracts {“fishing”, “invitation”, “south”, “recent”, “beginning”, “next time” from this message. , "Teaching"} as keywords. The classification condition learning means updates the contents of the classification condition holding means 5 with “fishing”, which is a keyword common to all the messages in FIGS. 3, 4, and 5, as a new classification condition. After that, if a message including the keyword “fishing” is received, for example, as shown in FIGS.
The accumulation control means 2 accumulates these messages in the classification section “fishing relation” in the first accumulation means 3.

By automatically determining the classification conditions by the classification condition learning means in this way, the user does not need to perform complicated setting work of the classification conditions, and the load on the user in the user's message operation / processing environment. Has the effect of being reduced. In addition, since the classification condition corresponding to the classification category of the message is determined from the contents of the message classified by the user's own will, the classification condition will be in accordance with the user's wish, and the effect of improving the operation / processing environment of the user's message can be obtained. is there. Further, since the classification condition corresponding to the classification classification of the message is dynamically changed by learning, there is an effect that the classification condition becomes more accurate. In addition, in the present embodiment, an example in which the first accumulating means and the second accumulating means are configured as independent accumulating means is shown, but the same accumulating means is logically divided into the first accumulating means and the first accumulating means. The same effect can be obtained by using it for both purposes of the second storage means.

Embodiment 2. A message learning classification device according to the second embodiment of the present invention will be described with reference to the drawings. FIG. 6 is a configuration diagram of a messaging device according to the second embodiment of the present invention. In the figure, reference numeral 8 is a keyword conversion table for converting a plurality of different keywords into one common keyword representing them. 1 to 7
As for the above, the description is omitted because it is the same element as in FIG.
FIG. 7 is a diagram showing an example of the contents of the keyword conversion table of FIG. FIG. 8 is a diagram showing keyword holding, and FIG. 9 is a diagram showing another received mail.

Next, the operation of the apparatus having the above structure will be described. The case where the classification condition holding means holds the classification conditions as shown in FIG. 2 will be described as an example. In this state, if the user receives a message as shown in FIGS. 3, 4 and 5, and the user sees the contents and classifies them and the classification learning means learns, as described in the first embodiment, The classification condition held in the classification condition holding means 5 is updated, and the classification condition holding means stores the classification conditions corresponding to the classification category "fishing relation" in the keywords "fishing" and "karaoke" as shown in FIG. The classification condition of the classification category of "relationship" holds the keyword "Karaoke", and the classification condition of the classification category of "golf relationship" holds the classification condition of the keyword "golf". Here, the keyword conversion table 8 is, for example, a correspondence table of a plurality of different keywords as shown in FIG. 8 and a representative keyword representing them, and the classification condition holding means 5 uses this table based on this table. It is converted into one representative keyword as a plurality of consented keywords.

In this state, when a message as shown in FIG. 9 is received, the keyword "fishing" is interpreted by the accumulation control means 2 as "fishing", classified as "fishing relation", and stored in the first accumulation means 3. Accumulated in. As described above, by converting a plurality of consent keywords into one representative keyword by using the keyword conversion table, it is possible to prevent the same keywords from being included but not correctly classified because they do not include the same keyword. It becomes possible to classify messages more accurately.

Embodiment 3 A message learning classification device according to the third embodiment of the present invention will be described with reference to FIG. In the messaging device of FIG. 10, 9 is a classification condition changing means for changing the classification condition obtained by learning. The description of 1 to 7 is omitted because it is the same device as in FIG. Further, in this example, the classification condition changing means 9 is composed of a classification condition display means and a classification condition input means. FIG. 11 is a diagram showing an example of keyword holding by the classification condition holding means when keywords are expanded by the device having the configuration of FIG.

Next, the operation of the apparatus having the above configuration will be described. The case where the classification condition holding means holds the classification conditions as shown in FIG. 2 will be described as an example. In this state, if the user receives a message as shown in FIGS. 3, 4 and 5, and the user sees the contents and classifies them and the classification learning means learns, as described in the first embodiment, The classification condition held in the classification condition holding means 5 is updated, and the classification condition holding means stores the classification condition corresponding to the classification category “fishing relation” as shown in FIG.
The classification condition is held. In this state, when the message as shown in FIG. 9 is received, this message is
Since it does not correspond to any of the classification conditions held in the classification condition holding means 5, it is stored in the second storage means 4. However, the message is related to “fishing” and should be stored as “fishing relation” in the first storage means 3 as it should be.

Therefore, the user changes the classification condition of the classification category. For example, when the classification condition as shown in FIG. 8 is set, the classification condition display means visualizes and displays the classification condition corresponding to each classification item. As a result, the user can find that the classification condition corresponding to the classification category "fishing relation" is the keyword "fishing", and the classification condition for the classification category "karaoke relation" is the keyword "karaoke" and the classification condition for the golf category "golf relation". Is the keyword "golf"
Know that In order to change the classification condition corresponding to the classification category "fishing relationship", the user inputs "fishing" + "fishing" as the classification condition corresponding to the classification category "fishing relationship" by the classification condition input means. Here, “+” means a logical sum, and “fishing” + “fish”
ing is the keyword “fishing” or the keyword “fi
The term "shing" is included. The classification condition input means uses the inputted classification condition “fishing” + “fishi”.
ng ”is held in the classification condition holding means 5 as a classification condition corresponding to the classification category“ fishing relationship ”.

Thus, thereafter, even if the message shown in FIG. 9 is received, as shown in FIG. 11, the message is stored in the first storage means 3 classified as "fishing relation". As described above, since the classification condition changing means can change the classification condition, it becomes possible to set the classification condition when learning is insufficient or learning cannot cope, and the message classification more suitable for the user's request is obtained. Is possible.

Embodiment 4 A message learning classification device according to the fourth embodiment of the present invention will be described with reference to FIG. In the messaging device of FIG. 12, 10 is a predetermined area extracting means for extracting a predetermined area in the message. Since 1 to 7 are the same elements as in FIG. 1, the description thereof will be omitted.

Next, the operation of the apparatus having the above configuration will be described. When the receiving means 1 receives the message, the storage control means 2 determines the matching between the received message and the classification condition corresponding to each classification category held in the classification condition holding means 5, and the received message is If the classification conditions are met, the data is stored in the corresponding classification section of the first storage means 3. If the received message does not match any of the classification conditions, it is stored in the second storage means 4. However, the matching determination range in this case is limited to only the keywords included in the predetermined area of the received message.

The specific area extraction will be described more specifically. As a matter of course, the predetermined region also specifies the range region where the classification condition learning means 6 learns. As a predetermined area, "the word that appears at the beginning of the message" subjec "
The case where t: "from the character after the line to a line feed" is set will be described. When the received message is as shown in FIGS. 3 and 4, the outputs of the predetermined area extracting means 10 corresponding to the messages shown in FIGS. 3 and 4 are “let's go fishing” and “fishing”, respectively. Becomes Classification condition learning means 5
Does not read the entire contents of the received message, but refers only to this predetermined area, and holds the keyword “fishing” commonly included in both messages as a classification condition in the classification condition holding means 5. Other operations are the same as the operations of the apparatus described in the first embodiment, and therefore the description thereof is omitted. As described above, by providing the predetermined area extraction unit 10 and limiting the range of the message to be referred to for learning the classification condition, the number of calculation processes for learning is reduced, and the learning speed is increased. . Further, in the case of e-mail or the like, since the term “subject:” often contains many keywords that represent the content of the message, it is possible to set highly accurate classification conditions.

Embodiment 5 A message learning classification device according to the fifth embodiment of the present invention will be described with reference to FIG. In the messaging device of FIG. 13, 11 is a facsimile receiving means for receiving a facsimile document, and 12 is a sender name information adding means for adding sender name information sent from a facsimile sender according to a facsimile communication protocol. . Other 2
The description of 10 is omitted because it is the same as the element of FIG. FIG. 14 is a diagram showing an example in which sender information is added to a received message.

Next, the operation of the apparatus having the above structure will be described. The facsimile receiving unit 11 receives the facsimile document according to the facsimile communication protocol and at the same time receives the sender name information according to the facsimile communication protocol. As shown in FIG. 13, the sender name information adding means 12 adds the sender name information, which is the size and character string of the sender name information, to the facsimile document received by the facsimile receiving means 11 to generate a received message. To do. The storage control unit 2 determines the matching between the classification condition corresponding to each classification classification held in the classification condition holding unit 5 and the sender name information included in the received message, and the received message is classified as the classification condition. If it matches, the data is stored in the corresponding classification section of the first storage means 3.
If the received message does not match any of the classification conditions, it is stored in the second storage means 4.

When the message receiver does not belong to any of the classification categories and receives the message accumulated in the second accumulating means 4, it opens the message, confirms the content, and then predetermines this message. Classify into one of the classification categories,
It is stored in the first storage means 3. Predetermined area extraction means 10
Extracts a predetermined condition, that is, the source name information area in this case. The classification condition learning means 6 uses the source name information in each message accumulated based on each classification division in the first storage means 3 extracted by the predetermined area extraction means 10 to classify corresponding to each classification division. Determine the conditions.
As described above, since the classification conditions are learned based on the sender name information that is sent according to the facsimile communication protocol, it becomes possible to classify a facsimile document and further improve the message operation / processing environment. be able to.

Embodiment 6 FIG. A message learning classification device according to the sixth embodiment of the present invention will be described with reference to FIG. In the messaging device of FIG. 15, 13 is the first
Of the messages stored in the storage means and the second storage means, the reception time, the flag indicating whether or not the user has read the message, the classification category of the message, and the classification conditions are visualized and output on the screen. The received message list display means 14 is for displaying the contents of the message and the received message list. Note that the other numbered elements are the same as those in the other figures, and thus the description thereof is omitted. FIG. 16 is a diagram showing a display example of the received message and the additional information which are visualized and displayed by the received message list display means.

Next, the operation of the apparatus having the above structure will be described. As described in the first embodiment, the message received by the receiving means 1 is classified according to the classification condition held by the classification condition holding means 5 and accumulated in the first or second accumulating means. Further, the facsimile document received by the facsimile document receiving means 11 is classified according to the classification condition held by the classification condition holding means 5 and accumulated in the first or second accumulating means. As shown in FIG. 16, the received message list display means displays the type of received message (facsimile document or electronic mail in this example) stored in the first or second storage means, the reception time, and the message displayed by the user. The flag indicating whether or not it has been read out, the classification category of the message, and the classification condition are visualized according to the user's request, and displayed on the display means 14. Further, when there is a new received message that the user does not recognize, it is visualized so as to notify the user, and is displayed on the display unit 14. In this way, the message management environment of the user is displayed by visualizing and displaying the list of the received message, the type of the received message, the reception time, the flag indicating whether or not the user has read the message, the message classification category and the classification conditions. It becomes possible to improve.

Embodiment 7 The above embodiments have been described as devices. The above purpose may be configured by software using a general-purpose computer. In the present embodiment, a method of performing automatic classification as a stack of processing steps will be described. A message learning classification method according to the seventh embodiment of the present invention will be described with reference to FIG. In the messaging method of FIG. 17, S1 is a reception step of receiving a message from the other party, S2 is a storage control step of classifying the received message according to classification conditions corresponding to each classification item and controlling storage, and S3 is a reception message. A first accumulation step of accumulating, S2 a second accumulation step of accumulating received messages by classification category, S5 a classification condition retaining step of retaining classification conditions for classifying the received message into each category, and S6 This is a classification condition learning step of determining classification conditions for classifying received messages by learning.

Next, the operation will be described. When the message is received in the receiving step S1, in the storage control step S2, the matching between the received condition and the classification condition corresponding to each classification classification held in the classification condition holding step is determined, and the received message is classified. If the condition is met, the corresponding classification item is stored in the first storage step S3. If the received message does not match any of the classification conditions, it is stored in the corresponding unlearned section in the second storage step S4.

The recipient of the message opens the message in the unlearned category that does not belong to any of the category items accumulated in the second accumulation step, confirms the content, and then determines this message as the predetermined category item. It is classified into any of the above, and is accumulated in the first accumulation step. In the classification condition learning step S6, each message accumulated based on each classification category in the first accumulation step is read out and the classification condition corresponding to each classification category is determined. Specifically, the keywords are extracted, and further, all common keywords are obtained as classification conditions in each received message. Note that the specific operation of the classification condition learning is the same as that described in the first embodiment, so description thereof will be omitted.

By automatically determining the classification conditions in the classification condition learning step in this way, the user does not need to perform complicated classification condition setting work, and the load on the user in the message operation / processing environment of the user is reduced. Has the effect of being reduced. Further, since the contents of the message are used for the determination, the classification conditions meet the user's wishes, and the operation / processing environment of the user's message is improved. Further, since the classification condition corresponding to the classification classification of the message is dynamically changed by learning, there is an effect that the classification condition becomes more accurate. 18 to 21 are
FIG. 16 is a diagram showing an operation flow of a message learning classification method corresponding to each of the third to sixth embodiments. Each operation is the same as the above, and since it becomes redundant, detailed description will be omitted.

[0037]

As described above, the message learning classifying apparatus according to the present invention includes the first accumulating means for accumulating the received message in the classified classification condition, the second accumulating means for accumulating in the unlearned classification, and the second accumulating means. A keyword extracting means for extracting a keyword in a message corresponding to a category when the message in the second accumulating means is classified, and a classification condition learning means for learning a keyword common to the messages corresponding to the first classification as a new classification condition. Therefore, subsequent received messages can be learned and classified, which has the effect of omitting complicated classification work.

Furthermore, since a keyword conversion table for expanding a specific keyword is added, there is an effect that the classification work is further reduced.

Further, since the classification condition changing means for displaying the specific keyword and the arbitrary keyword and displaying the classification condition and expanding the keyword of the classification condition by additionally designating the keyword is added, there is an effect that the classification work is further reduced.

Furthermore, since the predetermined area extracting means is added, there is an effect that determination processing of matching with message classification conditions and keyword extraction processing can be simplified and shortened.

Furthermore, since the received message list display means for displaying the specific item information of the received message is added, there is an effect that the classification process by the keyword is further simplified.

In the message learning classification method according to the present invention, a first accumulation step of classifying by the learning classification condition, a second accumulation step of accumulating as an unlearned section when the classification cannot be performed, and a message of the unlearned section are predetermined. When the message is classified into the categories, the keyword extraction step of extracting the keywords in the received message of this category and the classification condition learning step of learning the keywords commonly appearing in any message are provided. This makes it possible to classify them in a simple manner, which has the effect of eliminating complicated classification work.

[Brief description of drawings]

FIG. 1 is a configuration block diagram of a message learning classification device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing an example of a keyword holding status of a classification condition holding means and a storage status of a message storage means.

FIG. 3 is a diagram showing a specific received message.

FIG. 4 is a diagram showing a specific received message.

FIG. 5 is a diagram showing a specific received message.

FIG. 6 is a configuration block diagram of a message learning classification device according to a second embodiment of the present invention.

7 is a diagram showing an example of the contents of the keyword conversion table of FIG.

FIG. 8 is a diagram showing an example of a keyword holding state of a classification condition holding unit.

FIG. 9 is a diagram showing a specific received message.

FIG. 10 is a configuration block diagram of a message learning classification device according to a third embodiment of the present invention.

FIG. 11 is a diagram showing an example of a keyword holding state of a classification condition holding unit.

FIG. 12 is a configuration block diagram of a message learning classification device according to a fourth embodiment of the present invention.

FIG. 13 is a configuration block diagram of a message learning classification device according to a fifth embodiment of the present invention.

FIG. 14 is a diagram showing an example in which sender information is added to a received message.

FIG. 15 is a configuration block diagram of a message learning classification device according to a sixth embodiment of the present invention.

FIG. 16 is a diagram showing a display example of a received message list display means.

FIG. 17 is an operational flowchart showing a message learning classification method according to the seventh embodiment of the present invention.

FIG. 18 is an operation flowchart showing another message learning classification method according to the seventh embodiment of the present invention.

FIG. 19 is an operational flowchart showing another message learning classification method according to the seventh embodiment of the present invention.

FIG. 20 is an operational flowchart showing another message learning classification method according to the seventh embodiment of the present invention.

FIG. 21 is an operational flowchart showing another message learning classification method according to the seventh embodiment of the present invention.

[Explanation of symbols]

2 accumulation control means, 3 first accumulation means, 4 second accumulation means, 5 classification condition holding means, 6 classification condition learning means, 7 keyword extraction means, 8 keyword conversion table, 9 classification condition changing means, 10 predetermined area Extraction means, 11 Facsimile reception means, 12 Source information addition means, 13 Received message list display means.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G06F 15/401 310A

Claims (6)

[Claims] 1. A first accumulating unit for classifying and accumulating a received message based on a classification condition described later, and a second accumulating unit for accumulating a received message that is not classified by the classification class in an unlearned class. When the messages accumulated in the second accumulating means which is the unlearned category are classified into the categories corresponding to the first accumulating means,
A keyword extracting means for extracting a keyword in the received message of the category and a classification condition learning means for learning a keyword common to a plurality of messages classified into the category corresponding to the first category as a new classification condition. A message learning classifying device which holds the learned classification conditions and divides and stores subsequent received messages in a first accumulating unit based on the classification conditions. 2. When a keyword conversion table that regards other terms as the same keyword is added to a specific keyword and subsequent received messages are classified into the first storage means, refer to the keyword conversion table. The message learning classifying apparatus according to claim 1, wherein the range of common keywords is expanded and classified and accumulated. 3. A keyword which is held as a learned classification condition and an arbitrary keyword are displayed, and when the displayed arbitrary keyword is designated, the designated keyword becomes the keyword of the learned classification condition. In addition, when a classification condition changing unit for expanding the classification condition is added and subsequent received messages are classified into the first accumulating unit, it is classified and accumulated according to the classification condition obtained by the classification condition changing unit. The message learning classification device according to claim 1, wherein: 4. When a predetermined area extracting means for designating a predetermined area of a message is added and subsequent received messages are classified into the first storage means, the keyword of the predetermined area of the received message is classified according to the classification condition. The message learning classification device according to any one of claims 1 to 3, wherein the message learning classification device is classified and accumulated. 5. A received message list display means for extracting and displaying information on predetermined items of a received message is added, and when requested, the above item information is extracted from the received message and displayed as a received message list. The message learning classification device according to any one of claims 1 to 3, characterized in that. 6. A first accumulating step of classifying a received message according to pre-learned classification conditions, a second accumulating step of accumulating in a non-learned category if the learned message cannot be classified by the learned classification condition, When a message accumulated in the category is classified into a predetermined category, a keyword extraction step of extracting a keyword in a received message of the predetermined category and a message commonly appear in any of the messages classified into the predetermined category A message learning classification method comprising a classification condition learning step of learning a keyword as a new classification condition.