JPH11145957A - Multi-address communication system supporting method and device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a multi-address communication system supporting device capable of automatically performing grouping without obtaining the information of individual delivery destinations and easily reconstructing a group accompanying changes. SOLUTION: A message delivered in the past and the delivery destination are related and registered in a message/delivery destination address DB 10. In a delivery destination feature amount preparation part 12, history data stored in the DB 10 are called, the feature amounts of the respective messages are extracted, the extracted feature amounts are clustered and the messages whose feature amounts are similar are grouped. Further, the sum set of delivery destination data to which the messages belonging to the same group are delivered is obtained and it is turned to the delivery destination group of the group. Also, the average value of the feature amounts of the respective messages of the same group is obtained, it is turned to the representative feature amount of the group and the delivery destination group and the representative feature amount are related and stored in a delivery destination feature amount DB 14.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a broadcast communication system supporting method and apparatus.

[0002]

2. Description of the Related Art As is well known, a group broadcast transmits the same information at a time only to devices connected to a network which have been previously grouped for a specific purpose. Conventionally, group broadcasting is applied in a LAN environment. In addition, due to the recent development of the Internet, etc., instead of a limited relatively small network such as a LAN,
The above-mentioned group broadcasting is adopted even in a large-scale network. The grouped distribution destinations may be individual persons (addresses) instead of terminal devices.

When a certain message (mail, news) is delivered, the message must be delivered to a person who needs it. Therefore, whether or not delivery destinations can be grouped according to the situation depends on whether or not an effective broadcast communication system can be constructed. In general, when performing such grouping, it is necessary to obtain personal information of a delivery destination such as occupation, job content, job title, hobby, etc., and to group persons having common items into one group based on the information. Will be

[0004]

However, in the above-mentioned method, since the personal information of the delivery destination is changed with the passage of time, the latest personal information must be constantly obtained and the group must be reconstructed. In addition, not only is it complicated, but even under a relatively small network environment such as a LAN, it is practically impossible to reconstruct a group if the usage form is large.

In delivering a new message,
In many cases, it is unclear to which group the message should be delivered, and the message may not be delivered to the person who really needs it. In addition, the number of groups also increases as the use form increases, so that such a problem becomes conspicuous.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned background, and has as its object to solve the above-described problems and to automatically perform grouping without obtaining information on individual delivery destinations. It is another object of the present invention to provide a broadcast communication system supporting method and apparatus which can easily reconstruct a group due to a change with time and can easily specify a delivery destination of a new message.

[0007]

In order to achieve the above-mentioned object, a broadcast communication system supporting method according to the present invention is directed to a broadcast communication system for delivering the same message to delivery destinations constituting the group. A method for supporting a broadcast communication system for determining information on a delivery message, wherein a delivery message is associated with a delivery destination and stored in a storage unit as history data. Then, the history data stored in the storage means is called out, and a feature amount is extracted for each message from the history data. Messages having similar feature amounts are grouped by clustering the extracted feature amounts, and messages having similar feature amounts are grouped. Determining a delivery destination belonging to the group based on delivery destination data that delivered the message belonging to the group, and determining a representative feature amount of the group based on the feature amount of each grouped message; The delivery destination belonging to the determined group and the representative feature are associated with each other and stored in the delivery destination feature storage means.

[0008] By doing so, it is possible to automatically perform grouping based on the past messages and the history data of the delivery destination. That is, if the messages have similar feature values, the delivery destination is often common, and it can be estimated that the messages belong to the same group. The message to be delivered to the group of delivery destinations (groups) thus collected may be a message having a feature amount similar to the feature amount of a message delivered in the past. Therefore, a representative feature is determined based on each feature of messages belonging to the same group as a guideline for that.

The delivery destinations belonging to the same group are:
In the embodiment, the union of the delivery destinations to which the message belonging to the delivery destination is set is used. However, the present invention is not limited to this. May be comprehensively determined in consideration of other requirements such as the number of appearances, such as not adding to the delivery destination group of the group.

[0010] As an apparatus suitable for carrying out such a method, for example, a broadcast communication system supporting apparatus for determining information on a group in a broadcast communication system for delivering the same message to destinations forming the group. A storage unit for storing and holding the delivered message and the delivery destination in association with each other as history data, calling up the history data stored in the storage unit, extracting a feature amount for each message from the history data, and ,
The extracted feature values are clustered to group messages having similar feature values, a delivery destination belonging to the group is determined based on delivery destination data for delivering a message belonging to the same group, and the grouping is performed. A delivery destination feature amount creating means for determining a representative feature amount of the group based on the feature amount of each message, and a delivery destination belonging to the group determined by the delivery destination feature amount creating means and a representative feature amount stored in association with each other. (Claim 6).

In the case of delivering a new message, for example, a feature amount extraction process is performed on the new message to be delivered to obtain the feature amount of the new message. A quantity is matched with representative feature quantity data of each group stored in the delivery destination feature quantity storage means obtained by executing the method of claim 1, and a delivery destination belonging to a group having a high similarity is determined. The delivery destination of the new message may be determined (claim 2).

Also, instead of such automatic determination, when a new message is delivered, the same feature extraction processing as in claim 1 is performed to determine the feature of the message. Matching between the obtained feature quantity and representative feature quantity data of each group stored in the delivery destination feature quantity storage means obtained by executing the method of claim 1, and a delivery belonging to a group having a high degree of similarity. The destination is a delivery destination candidate for the new message. Then, the delivery destination candidate may be modified according to a predetermined condition to finally determine the delivery destination (claim 3).

Here, "high similarity" in claims 2 and 3 means that one group having the largest similarity among the obtained similarities may be extracted, or in such a case,
Taking a predetermined threshold value into consideration, the group having the highest similarity exceeding the threshold value may be set. When there are a plurality of groups exceeding a certain threshold, all (or a predetermined number) of the plurality of groups may be extracted.

The correction according to claim 3 may have at least one function of correction, addition, and deletion of delivery destination candidates. Then, the correction may be performed manually, or a person who does not deliver is registered in advance as a close list, and a person registered in the close list is automatically removed from the delivery destination. Various methods can be adopted.

[0015] The above-mentioned feature amount can be an important word present in the message and a weight of the important word. In such a case, the feature amount extraction processing described in each claim can use various keyword extraction system algorithms that automatically extract keywords from phrases in a message.

An apparatus suitable for carrying out such a method is, for example, a broadcast communication system support for determining a delivery destination of a new message in a broadcast communication system for delivering the same message to delivery destinations forming a group. 2. An apparatus, comprising: a feature amount extraction unit for extracting a feature amount of the new message; a feature amount extracted by the feature amount extraction unit; and a feature amount stored in the delivery destination feature amount storage unit according to claim 1. It comprises matching means for matching with representative feature amount data of a group and obtaining a similarity thereof, and delivery destination determining means for determining a delivery destination of the new message based on the similarity obtained by the matching means. (Claim 7). Then, on the premise of the invention, the delivery destination determining means may be further provided with a function of correcting the delivery destination determined based on the similarity (claim 8).

Further, the delivery destination of the new message is determined by executing the method of claim 2 or 3, and the new message and the delivery destination are additionally stored as history data in the storage means in association with each other at predetermined timing. Based on the newly added history data and the history data stored before that, at least one of the delivery destination belonging to the group and the representative feature stored in the delivery destination feature storage is updated. (Claim 5). By doing so, the group can be reconstructed, and it is possible to cope with a change over time. Then, this update processing is performed, for example, on the history data up to that time (the history data when the current delivery destination feature amount is generated and the history data delivered thereafter).
May be combined, and grouping may be performed again according to the method created first, or the destination feature may be updated by distinguishing the history data before and after the generation of the destination feature.

As an apparatus suitable for carrying out the method, for example, the new message and the determined delivery destination are stored in the storage means in association with each other, and the newly added history data is added at a predetermined timing. And updating means for updating at least one of a delivery destination belonging to the group and a representative feature stored in the delivery destination feature storage based on history data stored before. (Claim 9).

* Definition of Terms The term "broadcast communication system" as used in the present invention refers to a system for realizing group broadcast which simultaneously delivers the same message realized in a LAN environment to terminals belonging to a German group. However, the delivery destination may be a person, and the network is not limited to a relatively small network such as a LAN, but may be of a large or small network or an open or closed network. In short, any system can be used as long as it can deliver a message having the same content to a destination belonging to a predetermined group.

[0020]

FIG. 1 shows a preferred embodiment of the present invention. As shown in the figure, messages (news, mail, etc.) delivered in the past are stored in the message / delivery destination address database 10 in association with the delivery destinations of the messages. This database 1
0 stores various information in a data structure as shown in FIG. That is, when a certain message is delivered to a plurality of persons, as shown in FIG. 2A, the message ID (the specific message content (text data) is obtained by the ID data structure shown in FIG. Is specified), a leading pointer to a delivery destination list for specifying a recipient to deliver the message is stored in association with the message. Then, as shown in FIG. 2 (b), the delivery destination list specifies a delivery destination ID for specifying a person who needs the message (this delivery ID indicates a specific delivery by the ID data structure shown in FIG. 2 (c)). When the destination address is specified), the pointers of the next delivery destination belonging to the same group requiring the message are stored in pairs. And if there is no next destination (the last person in the same group)
A code indicating the last is stored.

Of course, various data structures are used to store the data. In short, any data structure can be used as long as the specific contents of the delivered message and all the delivery destinations to which the message has been delivered can be known. Good.

Then, the delivery destination characteristic amount creation unit 12 newly creates or corrects a delivery destination characteristic amount based on the data (message and all delivery destinations) stored in the message / delivery destination address database 10. -Update processing such as deletion is performed. As a specific function, a flowchart shown in FIG. 3 is realized.
FIG. 4 shows such a function conceptually.

That is, a message and a delivery destination address (delivery destination set) are acquired from the message / delivery destination address database 10 (ST1). Then, a message vector is generated for each message (ST2). That is, in the present embodiment, the message feature amount is defined by a set (message vector) of important words (key words) selected from the message.

An example of the message vector generation process is as shown in a flowchart in FIG. First, an important word list of a message to be processed is extracted (ST2a). The extraction of the important word list is performed, for example, by extracting all nouns present in the message, weighting each noun based on the number of appearances of the noun, and arranging a predetermined number from the top of the important words in the important word list. Automatic extraction of various keywords used when creating a database, such as extracting unnecessary words or creating unnecessary word dictionaries, and if unnecessary words related to a message exist, removing them and performing weighting. An algorithm can be used (ST2b). Then, the data in which the extracted important words are associated with the weights is set as a message vector for the message (ST2).
c).

As a more specific example, for a word Tj cut out from a message document by unnecessary word removal processing, the importance Vj is calculated by the following equation, for example.

Vj = fj / max (fk) fj: the number of appearances of the word Tj in the message document K
The word set is extracted and the set of words is used as a feature quantity representing the content of the message.

Next, when the above processing is executed to generate message vectors for all messages, the process proceeds to step 3, where clustering is performed, and messages having similar message vectors are put into the same group. As this clustering method, a method generally performed conventionally can be used. At this time, the number of classifications can be specified, and the classification can be performed according to the specified number of groups. As a result, as shown in FIG. 4A, a delivery destination set including one or a plurality of delivery destinations is individually associated with one message in the beginning, whereas FIG. Thus, a plurality of messages come to form one group. Also,
The delivery destination of this group is the union of the delivery destination sets of the messages constituting the group.

Next, a representative vector of the group (delivery destination group) is generated (ST4). This representative vector performs an averaging process on the message vectors of the messages belonging to the group, and sets it as the representative vector of the destination group. This averaging process is performed, for example, in step 2
Since the message vectors of all the messages belonging to the same group obtained in the above are composed of important words and their weights, all the important words extracted from each message are picked up, and the average of the weights assigned to them is averaged. Take a value. As a simple method of calculating the average value, the weight of each important word in the same group is obtained by adding all the weights of the same important word and dividing the added value by the number of messages belonging to the same group.

Then, all the important words extracted from at least one message may be reflected in the feature vector as important words in the group,
An important word having a weight equal to or more than a certain value obtained by the averaging process may be finally left, or a predetermined number of words having a large weight may be finally used as an important word (keyword).
Alternatively, various methods may be used.

Further, when the same important word is extracted in each message, it can be said that the important word is more important as characterizing the message belonging to the same group, and is extracted only in one or a small number of messages. The important words given are not so important as characterizing the messages belonging to the group. Therefore, the weight of the extracted important words may be increased in the messages belonging to the same group together with the weight of each message.

The representative vector of each group (delivery destination group) thus obtained has a data structure as shown in FIG. 6A, for example. Here, the ID is a number for specifying each group, and the delivery destination representative vector is the vector obtained in step 4 and has a data structure as shown in FIG. Here, KWj is an important word specifically extracted, and Wj is a weight for KWj. Further, as shown in FIG. 6A, the number of history messages (the number of messages belonging to the same group)
And a leading pointer to a delivery destination list for specifying a delivery destination of a message belonging to the group. Then, according to this head pointer, FIG.
All the delivery destinations belonging to the same group are specified by the delivery destination list data in which the delivery destination ID and the pointer to the next delivery destination ID as shown in FIG. Then, in the format of each data structure shown in FIG. 6, various information created by the delivery destination feature amount creating unit 12 is registered in the delivery destination feature amount database 14.

[0032] By doing so, it is possible to obtain the information of the individual destination and the message content without knowing the contents of the message.
A message can be automatically divided into a group (delivery destination group) to which the same message is sent, and the type (characteristic) of a message required for the group is specified as a delivery destination representative vector, so that a new message is delivered. In this case, it is easy to determine which group should be delivered.

That is, when a new message is to be delivered, the message is sent to the feature extracting unit 16 and the delivery destination determining unit 20. The feature amount extraction unit 16 extracts a feature vector from the received delivery message. This extraction processing can be realized by executing the same processing as in step 2 in which the message vector of each message is obtained in the delivery destination characteristic amount creation unit 12. Then, the extracted feature vector is sent to the pattern matching unit 18.

The pattern matching section 18 compares the given feature vector with a representative vector of an existing delivery destination group (delivery destination group) stored in the delivery destination feature amount database 14 to obtain similarity. A close group (delivery destination group) is determined. At this time, in order to make the number of keywords constituting both feature vectors to be compared the same, k words are extracted from the important words having a large weight to obtain topic information about each delivery destination group, and based on the important words and the weights, The matching is performed, and the delivery destination group of the most matching representative vector is determined. Then, the determined delivery destination group and similarity are sent to the delivery destination determination unit 20.

The delivery destination determining section 20 has a function for realizing the flowchart shown in FIG. That is, if the matching result is received (ST10), first, the similarity is compared with a predetermined threshold, and if the similarity is larger than a predetermined threshold, a list of delivery destinations corresponding to the representative feature vector is prepared. Used as a candidate list (ST11, 1
2). Although not shown, the candidate list is output through an output device such as a display or a printer, so that the operator can confirm the list. If there is a person who misses the candidate list, a destination can be further added to the candidate list (ST13). If it is determined in step 11 that the value is smaller than the prescribed threshold value, there is no candidate from the existing delivery destination for the delivery message having this content characteristic.
Again, manually specify a new delivery destination (ST1
3). Further, when there is a close list in which a non-delivery destination is registered, the close list may be collated with the close list and the destination registered in the close list may be removed. By executing such various processes, a delivery destination is determined. In this embodiment, the delivery destination candidates are obtained and corrected. However, such processing may not be performed, and a group having a high degree of similarity may be determined as a delivery destination and delivered as it is.

Then, the received message is delivered to the delivery destination determined in this way (ST1).
4). Furthermore, the delivery result is message /
It is stored in the delivery destination address DB, and is also sent to the delivery destination feature amount creating unit 12 (at least that the delivery has been executed),
An update process of the delivery destination feature is performed. This update process
It can be performed by the same processing as the above-described new creation. The update process may be performed each time a message is delivered, or may be performed every predetermined number of times. Also, in principle, every predetermined number of times,
When a new address is generated (input addition in step 13 is executed), various methods can be adopted, such as immediately executing an update process.

As another example of the updating process, when the weight of the updated important word j (j = 1, 2, 3,...) Is expressed as Wj, Wj is defined by the following equation. The representative vector may be obtained as described above. By doing so, the number of calculation processing targets can be reduced, so that the update processing can be easily performed.

[0038]

(Equation 1)

[0039]

As described above, in the broadcast communication system supporting method and apparatus according to the present invention, the destinations are grouped based on the characteristic amount of the message delivered in the past. Grouping can be done without knowing the information. Since the groups can be divided based on the history data of the messages sent in the past, the groups can be easily reconstructed with the lapse of time. further,
The delivery destination of the new message can also easily specify a group to which a similar message has been delivered in the past by obtaining the feature amount of the new message.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a preferred embodiment of the present invention.

FIG. 2 is a diagram showing a data structure of a message / delivery destination address database.

FIG. 3 is a flowchart illustrating functions of a delivery destination feature amount creating unit;

FIG. 4 is a diagram illustrating an operation of a delivery destination feature amount creating unit.

FIG. 5 is a flowchart illustrating a part of a function of a delivery destination feature amount creating unit;

FIG. 6 is a diagram showing a data structure of a delivery destination feature amount database.

FIG. 7 is a flowchart illustrating a function of a delivery destination determining unit.

[Explanation of symbols]

 Reference Signs List 10 Message / delivery destination address database 12 Delivery destination feature amount creation unit 14 Delivery destination feature amount database 16 Feature amount extraction unit 18 Pattern matching unit 20 Delivery destination determination unit

Claims (9)

[Claims] 1. A broadcast communication system supporting method for deciding information on a group in a broadcast communication system for delivering the same message to delivery destinations constituting a group, wherein a history of the delivered message is associated with the delivery destination. The history data stored in the storage means is called out, and the feature amount is extracted for each message from the history data, and the extracted feature amounts are clustered so that the feature amounts are similar. Messages that belong to the same group, determine a delivery destination belonging to the group based on delivery destination data that has delivered the messages belonging to the same group, and represent a representative of the group based on the feature amount of each grouped message. Determine the feature quantity and associate the delivery destination belonging to the determined group with the representative feature quantity In addition, a broadcast communication system supporting method that is stored in a delivery destination characteristic amount storage unit. 2. When delivering a new message, the same feature extraction processing as in claim 1 is performed to obtain a feature of the new message, and then the obtained feature and the method of claim 1 are executed. Then, matching is performed with the representative feature amount data of each group stored in the delivery destination feature amount storage means obtained as described above, and a delivery destination belonging to a group having a high similarity is determined as a delivery destination of the new message. Broadcast system support method. 3. When delivering a new message, the same feature extraction processing as in claim 1 is performed to determine the feature of the message, and then the determined feature and the method of claim 1 are executed. Matching with the representative feature amount data of each group stored in the delivery destination feature amount storage means obtained as described above, and a delivery destination belonging to a group having a high degree of similarity is set as a delivery destination candidate for the new message; A method for supporting a broadcast communication system, wherein the distribution destination candidate is corrected according to the above-mentioned method to finally determine the distribution destination. 4. The feature amount is an important word present in a message and a weight of the important word.
4. The broadcast communication system support method according to any one of 3. 5. The method according to claim 2, wherein the destination of the new message is determined, and the new message is associated with the destination and additionally stored as history data in the storage means. Based on the newly added history data and the history data stored before that, at least one of the delivery destination belonging to the group and the representative feature stored in the delivery destination feature storage is updated. 2. The broadcast communication system supporting method according to claim 1, wherein: 6. A broadcast communication system supporting apparatus for determining information on a group in a broadcast communication system for delivering the same message to delivery destinations constituting the group, wherein the history is obtained by associating the delivered message with the delivery destination. A storage unit for storing and holding as data, and calling history data stored in the storage unit, extracting a feature amount for each message from the history data, and clustering the extracted feature amounts so that the feature amounts are similar. Messages that belong to the same group, determine a delivery destination belonging to the group based on delivery destination data that has delivered the messages belonging to the same group, and represent a representative of the group based on the feature amount of each grouped message. A destination feature amount creating unit that determines a feature amount; A broadcast communication system support apparatus, comprising: a delivery destination feature storage unit that stores a delivery destination belonging to the determined group and a representative feature in association with each other. 7. A broadcast communication system supporting device for determining a delivery destination of a new message in a broadcast communication system that delivers the same message to delivery destinations forming a group, wherein a feature amount of the new message is extracted. A feature amount extraction unit, and matching between the feature amount extracted by the feature amount extraction unit and representative feature amount data of each group stored in the delivery destination feature amount storage unit according to claim 1. A broadcast communication system supporting apparatus comprising: a matching means for determining a degree; and a delivery destination determining means for determining a delivery destination of the new message based on the similarity determined by the matching means. 8. The broadcast communication system according to claim 7, wherein said destination determining means is further provided with a function of correcting a destination determined based on said similarity. Support equipment. 9. The new message and the determined delivery destination are additionally stored in the storage unit in association with each other, and based on the history data newly added at a predetermined timing and the history data stored before that. 9. The broadcast communication system support device according to claim 7, further comprising an update unit for updating at least one of a delivery destination belonging to the group and a representative feature stored in the delivery destination feature storage.