JP2005084858A - NETWORK COMMUNICATION SYSTEM, NETWORK COMMUNICATION PROVIDING SERVER, PROGRAM FOR OPERATING THE SAME, AND RECORDING MEDIUM CONTAINING THE PROGRAM - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a network communication system capable of providing a communication participant with a past communication log made by a topic similar to or similar to a topic of ongoing communication as knowledge material.
A past log division unit 11b divides a past log of communication stored in a past log storage unit 12a into topic units, and a topic feature vector generation unit 11c generates topic unit feature vectors. Further, the topic unit similarity search unit 11d searches for a topic unit similar to the ongoing communication from a plurality of topic units, and outputs and displays it as knowledge data on the client terminal device 30. Therefore, communication becomes more active, and stored message data can be effectively used as knowledge information.
[Selection] Figure 2

Description

  The present invention relates to a network communication system, and more particularly to a network communication system, a network communication providing server, and a network communication providing server capable of reusing a history of communication such as a conversation performed using the network communication system as knowledge information. The present invention relates to a program for operating, and a recording medium on which the program is recorded.

2. Description of the Related Art Today, network communication systems such as chat and electronic bulletin boards are known that can communicate between a plurality of client terminal devices by transmitting and receiving message data via a network. In addition, by using this network communication system, it is not necessary to secure a special place such as a conference hall, and it is possible to easily carry out conference and symposium-style communication between multiple people at remote locations. It has become possible.
On the other hand, communication logs on these network communication systems contain a lot of message data with high utility value as knowledge information. Therefore, it is desired to effectively utilize the log of communication performed by network communication so that it can be reused.
In response to the above problems, the applicant has proposed a network communication providing server capable of easily and effectively reusing stored message data as knowledge information according to No. 2002-301641.
In this network communication providing server, message classification information is added, classification information data adding means for generating message data, and inter-message relation in which message data is added to the message data indicating inter-message relation information. Since the information data input means is provided, the stored message data can be easily searched and communication can be reproduced, and the stored message data can be widely used as knowledge information.

Japanese Patent Application No. 2002-301641

By the way, as a result of repeated earnest research experiments based on the invention of Patent Document 1, the inventors appropriately obtained materials related to the topic of conversation that is currently in progress in network communication. It has been found that exchanging opinions on the topic becomes more active when it is provided to those who participate in communication.
In addition, in network communication, it has been found that usages such as a plurality of communications on the same / similar topic proceeding independently of each other or repeated several times in the past.

  Under the above circumstances, as a method of effective use of communication logs using the network communication system, the same or similar to the main topic of ongoing communication from among the stored communication logs for communication participants If the past communication log made by the topic can be provided as knowledge material, active communication can be promoted and the saved communication log can be used effectively.

In order to solve the above-mentioned problems, a network communication system according to the present invention as claimed in claim 1 is a communication device using a message data transmitted from a client terminal device and received by the message data transmission / reception means. A storage device comprising past log storage means for accumulating the history of the past and storing it as a past log, and a plurality of client terminal devices capable of performing communication using the communication provision server Is a network communication system that is connected via a network,
The control device includes a past log dividing unit that divides a past log stored in the past log storage unit into topic units that are aggregates of message data for each topic, and the storage device includes the past log dividing unit. The communication providing server includes a topic unit similar to a specific topic based on a plurality of topic units stored in the topic unit storage unit. Search means for searching and outputting and displaying on the client terminal device is provided.

  The network communication system according to claim 2 is the network communication system according to claim 1, wherein the search means is a topic feature of each topic unit stored in the topic unit storage means provided in the control device. A topic feature vector generating means for generating a feature vector for each topic unit, and a topic feature vector for storing the topic feature vector generated by the topic feature vector generating means provided in the storage device. From the plurality of topic units stored in the topic unit storage unit, the storage unit and the feature unit of the topic unit obtained by the topic feature vector generation unit provided in the control device are compared. Talk that searches for topic units similar to the topic of the user and outputs and displays them on the client terminal device Characterized in that it consists of a unit similarity retrieval means.

  The network communication system according to claim 3 is the network communication system according to claim 1 or 2, wherein the control device includes a topic unit summary sentence generation unit that extracts the topic unit summary. .

  The network communication system according to claim 4 is the network communication system according to any one of claims 1 to 3, wherein the past log dividing means includes a message classification type, message data transmission / reception time, and message data assigned to message data. The past log division start point and / or the past log division end point are determined by referring to one or two or more factors selected from the mutual relationship and the communication elapsed time limit, and the past log is divided into topic units. To do.

The network communication providing server according to claim 5, wherein the message data is transmitted from a client terminal device, a control device comprising message data transmitting / receiving means for transmitting / receiving message data between a plurality of client terminal devices connected via a network, and the message data A communication providing server comprising: a storage device including a past log storage unit that accumulates a history of communication based on message data received by the transmission / reception unit and stores it as a past log;
The control device includes a past log dividing unit that divides a past log stored in the past log storage unit into topic units that are aggregates of message data for each topic, and the storage device includes the past log dividing unit. The client terminal device includes a topic unit storage unit that stores the topic unit obtained by the above, and searches for a topic unit similar to a specific topic based on a plurality of topic units stored in the topic unit storage unit. And a search means for outputting and displaying.

  The network communication providing server according to claim 6 is the communication providing server according to claim 5, wherein the search unit is a topic for each topic unit stored in the topic unit storage unit provided in the control device. Topic feature vector generating means for generating feature vectors for each topic unit by quantifying the features, and topic features for storing the topic feature vectors generated by the topic feature vector generating means provided in the storage device By comparing the feature unit of the topic unit obtained by the topic feature vector generation unit provided in the control device with the vector storage unit, from a plurality of topic units stored in the topic unit storage unit, Topic units that search for topic units similar to a specific topic and display them on the client terminal device Characterized in that it consists of the search means.

  The network communication providing server according to claim 7 is the network communication providing server according to claim 5 or 6, wherein the control device includes a topic unit summary sentence generation unit that extracts the topic unit summary. And

  The network communication providing server according to claim 8 is the network communication providing server according to claims 5 to 7, wherein the past log dividing means includes a message classification type, message data transmission / reception time, message data assigned to message data. By referring to one or two or more factors selected from the interrelationship between them and the communication elapsed time limit, the past log division start point and / or the past log division end point are determined, and the past log is divided into topic units. Features.

  The invention according to claim 9 is a program for operating the communication providing server according to any one of claims 5 to 8.

  A tenth aspect of the present invention is a computer-readable recording medium on which the program according to the ninth aspect is recorded.

  ADVANTAGE OF THE INVENTION According to this invention, the past conversation log made by the topic which is the same as or similar to the main topic of the conversation now in progress can be provided as knowledge data from the stored conversation log, and active communication is promoted. And the stored conversation log can be used effectively as a database of knowledge information.

  The embodiment of the present invention will be described with reference to FIGS.

  FIG. 1 is a diagram showing a schematic configuration of a network communication system 1 according to an embodiment of the present invention, and what is connected to the network communication providing server 10 via the network communication providing server 10 and the network 20. And a single client terminal device 30. As the network 20, a well-known wired or wireless network such as the Internet or a LAN can be used. As the client terminal device 30, a known terminal device such as a personal computer or a mobile phone can be used.

  In the network communication system 1 configured as described above, the communication providing server 10 receives message data transmitted from a client terminal device connected to the communication providing server 10, for example, the client terminal device 31, and this message data Is immediately transmitted to all the client terminal devices 30 connected via the network 20. In this way, for example, communication in a conference format or a symposium format can be performed between clients operating each client terminal device 30.

FIG. 2 is a schematic configuration diagram of the communication providing server 10 according to the embodiment of the present invention, and includes a control device 11 and a storage device 12.
The control device 11 includes a message data transmission / reception unit 11a, a past log division unit 11b, a topic feature vector generation unit 11c, a topic unit similarity search unit 11d, and a topic unit summary sentence generation unit 11e. The storage device 12 includes past log storage means 12a, topic unit storage means 12b, and topic feature vector storage means 12c. Note that the storage device 12 can be provided in the communication providing server 10 or in another database server connected to the communication providing server 10.
The search means referred to in the present invention includes, for example, topic feature vector generation means 11c, topic unit similarity search means 11d, and topic feature vector storage means 12c.

The message data transmission / reception means 11a receives the message data transmitted from the client terminal device 30 and transmits the received message data to all the client terminal devices 30 connected to the communication providing server 1. The message data includes message information such as a message classification type, a message reception time, a sender name, and a relationship between messages, in addition to a message body composed of text, sound, image, or a combination thereof. The message data transmission / reception means may be a known means and is not particularly limited.
As a user interface of the client terminal device 30 used for transmission / reception of message data, for example, a browser screen 110 including a message display unit 111, a message input unit 112, a participant display unit 113, and a message history display unit 114 shown in FIG. Can be used. In this browser screen 110, the communication participant who wants to speak describes the message body as the content of the message in the message description part 112a and selects an icon representing an arbitrary message classification type from the message classification icon display part 112b. To do. As a result, message classification information is added to the message data, such as whether the message's main message is a new question or a question for another statement, whether it is an answer to a question, or an objection to another answer. This is a configuration that can be transmitted to the communication providing server 10.
In addition, as shown in the message display unit 111, the message data to be transmitted is drawn with a link line 111a on the message display unit 111 between the other message data, so Whether the message is a message is visually indicated, and the relationship information between the message data is described in the message data to be transmitted.
As a means for adding message classification type, relation information between message data, and other message information to the message body, it is related to a markup language (tag) tag such as XML (eXtensible Markup Language), and a message. It can be configured to be added to the text described as the main body.

  The time-series message data received by the message data transmission / reception means 11a, that is, the communication history, is stored as needed in the past log storage means 12a.

The history of communication saved by the past log saving unit 12a is read by the past log dividing unit 11b and divided into a set of message data related to a specific topic, that is, a topic unit. In the present specification or drawings, “topic” refers to the subject of communication that is spoken by transmitting and receiving message data regarding a certain matter.
As a method of dividing into topic units, known means such as a method of dividing the text document into topic units by morphological analysis and extracting a word that is the center of the document can be used. However, in message data conversations in a network communication system, there are many message data that only include “Yes”, “Yes.”, “No.”, etc., and only image data. It is possible to use not only the text data but also a method of dividing one or more elements selected from message classification types, message data transmission / reception times, interrelationships between message data, and communication elapsed time restrictions into a topic unit. preferable.
For example, as shown in the conceptual diagram of the topic unit dividing process shown in FIG. 4 and the topic unit dividing process flow shown in FIG. 5, the message classification type information included in the message data, the relationship between the message data, the preset communication elapsed time limit And can be divided into topic units.
Specifically, the message data is acquired (S100), and the linkage relationship between the message data is analyzed by referring to the relationship information data between the message data (S101). Among a series of message data having a chain relationship, one having the first message classification type “question” data is selected, and this is set as a division starting point (S102). Further, the chained message data is traced, and the message data having the next message classification type “question” data is set as the division end point. If the communication elapsed time limit (for example, set to 5 minutes) elapses before the message data having the next message classification type “question” data appears from the division start point, the communication elapsed time limit elapse point is branched. The end point is set (S103). A set of a series of message data belonging to the division start point to the division end point obtained by the above processing can be set as a topic unit (S104).
The past log dividing unit 11b stores the topic unit obtained by executing the above steps in the topic unit storing unit 12b.

Each topic unit divided by the past log dividing unit 11b and stored in the topic unit storage unit 12b is read out by the topic feature vector generation unit 11c, and the topic features are quantified, so that each topic unit A feature vector is generated. As the topic feature vector generating means, for example, as shown in the processing flow of FIG. 6, text data described as the message body of message data included in each topic unit is decomposed into words (S200). A known means such as a method based on morphological analysis can be used for this decomposition process.
Next, by calculating the weight of the decomposed word (a value indicating the degree related to the subject of the topic), a feature vector having the word weight included in each topic unit as a constituent element is calculated (S201). For calculating the word weight, a known method such as TF-IDF (Term Frequency-Inverse Document Frequency) can be used.
The feature vector of each topic unit obtained by the execution of the topic feature vector generation unit 11c is stored in the topic feature vector storage unit 12c in association with the topic unit for which the feature vector is generated.

The topic unit similarity search unit 11d measures the similarity between topic units by comparing the feature vectors of the topic units obtained by the topic feature vector generation unit 11c, and the topics stored in the topic feature vector storage unit 12c. Search for other topic units similar to a particular topic unit from the units.
That is, for example, as shown in FIG. 7, the topic unit similarity search unit 11d first receives a search key inputted by selecting a specific topic unit from the client terminal device 30 (S300), and stores topic unit feature vectors. The feature 12c is accessed to search for a feature vector for the specific topic unit (S301), and the feature vector and the feature vector of another topic unit are compared to search for a high similarity between topic units ( Then, it is displayed on the client terminal device 30 (S303). In the display, it is possible to display only a topic having the highest similarity or a plurality of topic units in descending order of similarity. The display content may be details (full text) of searched topic units, or a summary display of a plurality of candidates may be performed using a user interface for detailed output candidate display (FIG. 8) to be described later before detailed display. May be.
As a method for obtaining the similarity between feature vectors, that is, the similarity score, a known method such as a cosine function using an angle between vectors or an inner product of vectors can be used.

  The topic unit search user interface is configured such that the topic unit similarity search unit 11d executes processing by arranging a topic unit search request button 111b for each message data in the browser screen 110 shown in FIG. Can do. Specifically, when an arbitrary topic unit search request button 111b is pressed on the client terminal device 30 of the communication participant, a search request is transmitted to the topic unit similarity search means 11d and the topic unit search request button 111b is also sent. The topic unit to which the associated message data belongs is transmitted to the topic unit similarity search means 11d as a search key. As a result of the search according to the processing flow of FIG. 7, the extracted topic units are, for example, in the display example shown in FIG. Is displayed. When the communication participant selects one topic unit by referring to the headline (summary sentence generated by the topic unit summary sentence generating unit 11e described later) from the displayed detailed output candidates, for example, as shown in FIG. In such a display configuration, the topic unit selected in the detailed output candidate display is highlighted by changing the display color and similar to the desired topic unit by displaying message data around the topic unit. The flow may be displayed on the client terminal device 30 together with the information.

  The topic unit summary sentence generation unit 11e generates a topic unit communication summary sentence extracted by the topic unit search unit 11d. As a result, even if the topic unit contains a relatively large amount of message data, it is possible to easily understand what kind of content has been communicated in the topic unit by browsing the generated summary. Become. The topic unit summary sentence generation processing flow is, for example, as shown in FIG. 10, and the topic unit summary sentence generation unit 11e receives the topic unit summary sentence generation request retrieved from the topic unit similarity search unit 11d (S400). ), Dividing the text string in the message data body of the topic unit into words by using morphological analysis or the like (S402), and counting the frequency of appearance among words of a predetermined part of speech category (noun, etc.) A high-frequency one is discriminated as a keyword (S403), and all the message data including the keyword is presented, or only a predetermined number of characters from the head of the corresponding message data is presented to generate a topic unit summary sentence. (S404), to the topic unit similarity search means 11d.

By using the network communication system 1 configured as described above, for example, a history of communication performed on the same or similar topic in the past with respect to a communication related to a specific topic that is progressing in real time can be obtained. It can be provided as a network, can encourage active communication by communication participants, and can operate efficient conferences and symposiums. At the same time, it is possible to effectively use a log of communication performed using the network communication system 1.
The present embodiment can be realized by software processing using a general-purpose processor without using a dedicated device. For example, the processes of FIGS. 5, 6, 7, and 10 can be realized by using a computer program, which is executed by being introduced into a computer through a recording medium such as FD, MO, and CD-ROM and executed. The above embodiment can be realized.

FIG. 1 is a schematic configuration diagram of a network communication system according to an embodiment of the present invention. FIG. 2 is a schematic diagram of the configuration of the communication providing server according to the embodiment of the present invention. FIG. 3 is a browser display configuration diagram of 1 which is a user interface displayed on the client terminal device of the present invention. FIG. 4 is a diagram showing the concept of topic unit division processing of the present invention. FIG. 5 is a diagram showing a flow of topic unit division processing of the present invention. FIG. 6 is a diagram showing a topic unit feature vector generation processing flow of the present invention. FIG. 7 is a diagram showing a processing flow of topic unit similarity search according to the present invention. FIG. 8 is a diagram showing a window displaying one search candidate searched by the topic unit similarity search means of the present invention. FIG. 9 is a diagram showing one search result display window searched by the topic unit similarity search means of the present invention. FIG. 10 is a diagram showing a processing flow of topic unit summary sentence generation according to the present invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Network communication system 10 Communication provision server 20 Network 30 Client terminal device

Claims (10)

A control device comprising message data transmission / reception means; a storage device comprising past log storage means for accumulating a history of communication by message data transmitted from the client terminal device and received by the message data transmission / reception means; A network communication system in which a communication providing server configured by a plurality of client terminal devices capable of performing communication using the communication providing server are connected via a network,
The control device includes a past log dividing unit that divides a past log stored in the past log storage unit into topic units that are aggregates of message data for each topic, and the storage device includes the past log dividing unit. The communication providing server includes a topic unit similar to a specific topic based on a plurality of topic units stored in the topic unit storage unit. A network communication system comprising search means for searching and outputting and displaying on a client terminal device. The search means includes a topic feature vector generation means for generating a feature vector of each topic unit by quantifying the topic features of each topic unit stored in the topic unit storage means provided in the control device. The topic feature vector storage means for storing the topic feature vector generated by the topic feature vector generation means provided in the storage device, and the topic feature vector generation means provided in the control device. By comparing feature vectors of topic units, a topic unit similarity search for retrieving topic units similar to a specific topic from a plurality of topic units stored in the topic unit storage means and outputting and displaying them on a client terminal device The network communication system according to claim 1, comprising: means. The network communication system according to claim 1, wherein the control unit includes topic unit summary sentence generation means for extracting the topic unit summary. The past log dividing means refers to one or more factors selected from the message classification type assigned to the message data, the message data transmission / reception time, the interrelationship between the message data, and the communication elapsed time limit. 4. The network communication system according to claim 1, wherein a log division start point and / or a past log division end point is determined, and a past log is divided into topic units. A control device comprising message data transmitting / receiving means for transmitting / receiving message data between a plurality of client terminal devices connected via a network, and a communication history by message data transmitted from the client terminal device and received by the message data transmitting / receiving means And a storage device comprising a past log storage means for storing and storing as a past log,
The control device includes a past log dividing unit that divides a past log stored in the past log storage unit into topic units that are aggregates of message data for each topic, and the storage device includes the past log dividing unit. The client terminal device includes a topic unit storage unit that stores the topic unit obtained by the above, and searches for a topic unit similar to a specific topic based on a plurality of topic units stored in the topic unit storage unit. A communication providing server comprising search means for outputting and displaying on the server. The search means includes a topic feature vector generation means for generating a feature vector of each topic unit by quantifying the topic features of each topic unit stored in the topic unit storage means provided in the control device. The topic feature vector storage means for storing the topic feature vector generated by the topic feature vector generation means provided in the storage device, and the topic feature vector generation means provided in the control device. By comparing feature vectors of topic units, a topic unit similarity search for retrieving topic units similar to a specific topic from a plurality of topic units stored in the topic unit storage means and outputting and displaying them on a client terminal device 6. The communication providing server according to claim 5, further comprising: means. 7. The communication providing server according to claim 5, wherein the control unit includes topic unit summary sentence generation means for extracting the topic unit summary. The past log dividing means refers to one or more factors selected from the message classification type assigned to the message data, the message data transmission / reception time, the interrelationship between the message data, and the communication elapsed time limit. 8. The communication providing server according to claim 5, wherein a log division start point and / or a past log division end point is determined, and the past log is divided into topic units. The program for operating the communication provision server in any one of Claims 5-8. A computer-readable recording medium on which the program according to claim 9 is recorded.

Images