JP2002229795A - Communication server and communication method with agent knowledge information by server - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow a user to search for another user having the same taste by her-himself or to search for another person having the same taste and maintain communications without directly associating with the person in a virtual community. SOLUTION: Agents serving as proxies of users maintain communications on a computer network in a computer network system. The compatibility in temperament between the users is evaluated and judged (compatibility in temperament evaluating/judging section 12) based on the user information held by the agents, thereby the user maintains communications with the other user to form a virtual community. When the user wants to search for another person having the same taste and have communications in the virtual community, the agent reflecting the characteristics and taste of the user communicates with the other agent, and the compatibility in temperament between the users is evaluated and judged based on the user information (knowledge information DB 15) held by the agents.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a communication server, a communication method using agent knowledge information by the server, a server program in which the method is programmed, and a recording medium on which the program is recorded.

[0002]

2. Description of the Related Art When a user tries to communicate with another person who likes his / her taste, he / she searches for a community that meets his / her preference and participates in the community, and within the community, a profile of community members, a bulletin board, and the like are displayed. By referencing, a member who has a hobby is judged and selected by himself. Then, they directly interact with the members via chat, mail, and the like. Each site collects information suitable for users to build a community and provides it free or at low cost. In addition, an interactive environment such as a chat or a forum is provided so that participants can send information by themselves.

In the above-mentioned online community, in order for the user to obtain a communication result between the agents as a pseudo-conversation, the contents of the conversation assumed in advance are prepared on the system side as a template (model), and the contents of the communication are prepared. In response to this, a method of selecting, assembling, and presenting it to the user has been adopted. In addition, in order for a user to perform notification within the community, the user is notified of information to all users within the community or is specified by designating a person to be notified. Further, in order for the user to input and register the attribute information of the user to the agent, each item of the user information is filled according to an information input form (form) at the time of user registration.

[0004]

According to the conventional method described above, since the agent which is the agent of the user is not used, the user himself / herself searches for another person having a preference, and directly contacts the other person. Communication is needed.
Also, in order for others to refer to preferences, the user must disclose his / her personal preference information. Furthermore, the system administrator and the service provider need to define and prepare a community that matches the user's preference.

On the other hand, in order to realize a pseudo-conversation of a human-like agent, only the contents of a conversation prepared in advance are simply assembled and presented, so that only those whose conversation contents are scarce and whose pattern is fixed to some extent are realized. Had the inconvenience of being unable to do so. In addition, even if notification (advertisement, question) is sent to all users in the community, information is sent to many users who are not interested in the notification content, so the notification content is effective after all. Are often discarded without being referred to, and when such a large number of occurrences occur, the value of the notification itself is reduced. Furthermore, even if it is desired to specify a person to whom information is to be transmitted, it is difficult to select and specify which users are interested in the content that the user wants to transmit. Generally, a user information input form (form) at the time of user registration has input items for a large number of users, and inputting them requires a lot of trouble and time for the user.

The present invention has been made in view of the above various circumstances. In a computer network system in which agents acting as agents of users communicate with each other on a computer network, user information held by agents is stored in the computer network system. Based on the evaluation and judgment of the compatibility between users, they communicate with other users, and by forming a virtual community, users search for users who have their own preferences and directly interact with others Communication server using agent knowledge information by the server, and a communication method using the agent knowledge information by the server, and a server program in which the method is programmed. Provide the recorded recording medium An object of the present invention is to.

In the communication between agents,
By using the knowledge information possessed by each agent, the parameters used in the communication between the agents can include contents having a personality (emotions related to words, correlation between words, etc.), and communication between the agents By adapting the result history according to the characteristics of the agent on the system side, the user can obtain a result as if the agent had a pseudo-conversation as a conversation history. If you want to notify the agent, the information is sent only to the agent who has a hobby with you based on the compatibility evaluation judgment based on the knowledge information of the agent, so that effective advertising, advertising, useful Q & A, friendly consultation etc. Communication server and agent by the server Communication method using preparative knowledge information, and the server program the method is programmed, also aims to provide a recording medium that the program is recorded.

Further, in communicating with another agent using the knowledge information, the agent accumulates learning about the word as a homework when an unknown word which does not exist in its own knowledge information comes out. By inquiring the user or registering the user arbitrarily, the user can enter information in an interactive manner with a user-friendly agent, inputting user information with less stress and acquiring new knowledge. Another object of the present invention is to provide a communication server capable of performing registration, a communication method using agent knowledge information by the server, a server program in which the method is programmed, and a recording medium in which the program is recorded.

[0009]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention is a communication server in a computer network system in which communication between user terminals is performed by communication between agents, wherein each of the agents has Compatibility evaluation determining means for evaluating the compatibility between the users based on knowledge information to be performed, and conversation processing means for communicating with other users by reflecting the evaluation result, I do.

[0010] In the communication server of the present invention, the knowledge information held by the agent includes words, relationships between words, emotions for the words and relationships between words, a degree of reaction to the emotions, and a user's degree of reaction. It is made up of profile information including preferences, and is constructed as an individual knowledge information database (DB).

Further, the communication server according to the present invention further comprises a common knowledge DB serving as a reference base for constructing the individual knowledge information DB, wherein the individual knowledge information DB has the common knowledge information DB as knowledge information. It is characterized by inheriting the relationship and emotion between words that have it.

Further, in the communication server according to the present invention, the compatibility evaluation determining means searches for a word that triggers a conversation based on the individual knowledge information database of the agent as the agent moves, and according to its own profile. A first knowledge information inquiring means for obtaining the degree of reaction of the emotion to the word, notifying the content to the agent encountered, and inquiring about the emotion to the word, having a synonymous relationship of the word, and the same meaning First knowledge information notifying means for obtaining a degree of emotional reaction to a word by searching an individual database held by the encountered agent and notifying the inquired agent;
It is characterized by having.

[0013] In the communication server of the present invention, the compatibility evaluation determining means may determine whether the synonymous relation of the word is found or the degree of emotional reaction is not found. And a second knowledge information inquiry means for inquiring about the inclusion relation of the word, and searching the individual knowledge information DB of the encountered agent regarding the inclusion relation of the word, and And a third knowledge information inquiring means for informing the inquiring agent of the degree of reaction to the emotion and inquiring about the emotion for the included word.

[0014] In the communication server of the present invention, the compatibility evaluation determining means acquires a degree of emotional reaction from the two agents and refers to the degree of reaction to obtain a favorable sensitivity based on a predetermined arithmetic expression. Compatibility evaluation means for performing calculation, and compatibility determination means for performing compatibility evaluation determination by comparing the calculated positivity with a preset threshold value, are provided.

In the communication server according to the present invention, the conversation processing means may include a communication log collecting means for collecting a communication log between the agents, and a conversation decoration pattern prepared in advance by referring to respective profiles of the agents. And a pseudo-conversation document notifying unit for notifying a user of the pseudo-conversation document by e-mail or web to obtain a corresponding pattern from the communication log and converting the communication log into a pseudo-conversation document in a readable format. It is characterized by the following.

Further, in the communication server according to the present invention, the conversation processing means obtains information transmission from the user and refers to its own knowledge information DB to search for at least one agent having a preference. Means, and information transmitting means for transmitting information to the searched agent, or obtaining information according to its own taste, and responding depending on the information.

Further, in the communication server of the present invention, the frequency at which the words of each of the individual knowledge information DBs are referred to in accordance with the communication between the agents is checked and compared with a predetermined threshold value. Forgetting means for deleting the word from the individual knowledge information DB.

Further, in the communication server of the present invention, in the communication between the agents, if a word inquired by another agent is unknown or does not have a degree of emotional reaction to the word, it corresponds to the word. Homework issuing means for accumulating words and issuing a response request to the user as homework when the accumulated words reach a predetermined amount is provided.

Further, in the communication server according to the present invention, if the queried word is an unknown word, it is checked whether or not the word exists in the common knowledge information DB. If the word does not exist in the DB, the word is added to the individual knowledge information DB of the agent.
First user interface means for encouraging the user to input an emotion for the word, and, when the word is present in the common knowledge information DB, inheriting the emotion for the word and individual knowledge information of the agent Registering in the DB and prompting the user to input the degree of the emotional reaction
And a user interface means.

In addition, in the communication server of the present invention, when the words are registered in the individual knowledge information DB, respective definitions of words having synonymous relations, inclusive relations of the words, and types of feelings for the words are provided. And a knowledge information DB defining means for interactively communicating with the user terminal.

According to the above configuration, when a user seeks another person having a preference in a virtual community and tries to communicate, an agent reflecting the characteristics and preferences of the user communicates with another agent, and the agent communicates with another agent. An evaluation of compatibility between users is performed based on the held user information. In this way, communication with other users is performed and a virtual community is formed. As a result, the user can search for a user having a preference without directly interacting with another person, and can seek out another person having a preference, thereby reducing the burden on the user and improving communication. be able to.

In the communication between agents, the knowledge information held by each agent is used, so that the parameters used in the communication between the agents include contents having a human touch (emotions related to words, correlation between words, etc.). The system can adapt the communication result history of the agent according to the characteristics of the agent, so that the user can obtain a result as if the agent had a pseudo-conversation as a conversation history. .

When an agent wants to notify certain information to another agent, the information can be notified only to an agent interested in the agent by a compatibility judgment method based on the knowledge information of the agent. Therefore, when the user wants to notify some information to another person, it is possible to notify the information to the user of the virtual community having a preference without explicitly specifying the person to notify. As a result, effective advertisement / publicity, useful Q & A, and friendly consultation can be performed. Further, when communicating with another agent using the knowledge information, the agent can accumulate learning about the word as a homework when an unknown word that does not exist in its own knowledge information appears. When the agent inquires of the user or the user arbitrarily registers with the agent, the user can input information in an interactive manner with the user-friendly agent, and thus input the user information with less stress. And registration of new knowledge is possible.

[0024] The "agent" here means
A program that runs autonomously on a computer system and resides on the system as a daemon process. A user who uses a computer registers an agent as user information such as his / her own preference and uses the agent as a proxy. The “virtual community” refers to a virtual group in which people having the same attribute (interest, purpose, affiliation, etc.) gather in a network system using a computer. In a virtual community, agents as agents of computer users form a virtual group.

In order to solve the above problems, the present invention provides:
A communication method using agent knowledge information by a communication server in a computer network in which communication is performed between user terminals by communication between agents, wherein the compatibility between the users is determined based on the knowledge information held by each agent. It is characterized in that evaluation is determined, and communication with another user is performed by reflecting the evaluation determination result.

Further, in the communication method using the agent knowledge information by the communication server according to the present invention, the compatibility between the users is evaluated and judged based on the knowledge information held by each agent, and the users having the same attribute are evaluated. To form a virtual community in which are gathered.

In order to solve the above-mentioned problems, the present invention provides:
A server program used for a communication server in a computer network in which communication is performed between user terminals by communication between agents, wherein the server program communicates with the users based on knowledge information held by each agent. The method includes a step of evaluating and determining compatibility, and a step of performing communication with another user by reflecting the result of the evaluation determination, wherein the steps are performed by a computer.

In the server program of the present invention, a step of examining a frequency at which a word registered as the knowledge information is referred to in the communication between the agents, and comparing the frequency with a predefined threshold value. Depending on the result, removing the word from the knowledge information, and causing the computer to execute each of the steps.

In the server program according to the present invention, in the communication between the agents, when a word inquired by another agent is unknown or does not have a degree of emotional reaction to the word, Storing a word, and issuing a response request as homework to the user when the stored word reaches a predetermined amount, and causing the computer to execute each of the steps. I do.

Further, in the server program of the present invention, when the queried word is an unknown word, a step of checking whether or not the word exists as common knowledge information; If the word does not exist as knowledge information, the word is registered as knowledge information constructed for each of the agents, and the step of prompting the user to input an emotion for the word is performed. If the word exists as the knowledge information, And registering the agent as knowledge information of the agent, and prompting the user to input the degree of reaction of the agent, and causing the computer to execute each of the steps.

In order to solve the above-mentioned problems, the present invention provides:
A computer-readable recording medium storing a server program used for a communication server in a computer network system in which communication is performed between user terminals by communication between agents, wherein the server program is held by each of the agents. A server program that includes a step of evaluating and determining compatibility between the users based on knowledge information, and a step of performing communication with another user by reflecting the result of the evaluation determination, and causing a computer to execute each of the steps Is a recording medium on which is recorded.

[0032]

FIG. 1 is a diagram showing an example of a connection configuration of a computer network system in which the present invention is realized, and is a diagram cited for explaining a mechanism of an online community. In FIG. 1, reference numeral 1 denotes a communication server according to the present invention. In a system in which agents acting as agents of users communicate with each other on a computer network 2, compatibility between users is determined based on user information held by the agents. By evaluating and determining, it is used to communicate with other users or to form a virtual community. FIG. 2 shows its internal configuration in a functionally developed manner.
The communication server 1 has a personal computer 3 and a PDA via a computer network 2 such as the Internet.
(Personal Digital Assistants) User terminals such as a terminal 4 and a mobile phone 5 that can be connected to the Internet are connected.

FIG. 2 is a block diagram showing the internal configuration of the communication server according to the present invention by developing its functions.
Each of the blocks enumerated below is specifically configured by a peripheral LSI including a CPU and a memory, and the CPU realizes a function as a block by reading out a program from the memory and sequentially executing the program.

The communication server 1 is roughly divided into functions, and comprises an agent management unit 11, a compatibility evaluation determination unit 12, a conversation processing unit 13, a knowledge information DB management unit 14, and a knowledge information DB 15. Agent management unit 11
Is a manager part that resides as a demona process of the communication server 1 and executes an agent process that receives an agent by performing management of a designated port, performs necessary processing, and responds. The compatibility evaluation determination unit 12 has a function of evaluating and determining compatibility between users based on knowledge information held by each agent, and the conversation processing unit 13 reflects the evaluation result of the compatibility evaluation determination unit 12 It has a function to communicate with other users.

The compatibility evaluation judging section 12 further includes first to third
, A first knowledge information notifying unit 124, a compatibility evaluation unit 125, and a compatibility determination unit 126. First knowledge information inquiry unit 12
1 searches for a word that triggers a conversation based on the individual knowledge information database 152 of the agent as the agent moves, obtains the degree of emotional reaction to the word according to his / her profile, and gives the agent A function of notifying the contents of the word and inquiring about feelings about the word. The first knowledge information notifying unit 124 obtains the synonymous relationship of the words and the degree of the emotional reaction to the words having the same meaning by searching the individual knowledge information database 152 of the agent that has been encountered. It has the function of notifying the agent who has made it.

The second knowledge information querying unit 122 further searches the agent inquiring if the synonym of the previous word is not found or the degree of emotional reaction is not found. It has the function to inquire. The third knowledge information inquiring unit 123 obtained by searching the individual knowledge information DB 152 of the agent who encountered the inclusive relation of the word, and asked about the degree of reaction to the emotion of the included parent-child word. It has the function of notifying the agent and inquiring about the feelings about the included words. Compatibility evaluation unit 12
Reference numeral 5 has a function of acquiring the degree of emotional reaction from both agents and calculating the positivity based on a predetermined arithmetic expression with reference to the degree of reaction. In addition, the compatibility determination unit 126 has a function of performing compatibility evaluation determination by comparing the positivity calculated by the compatibility evaluation unit 125 with a preset threshold.

The conversation processing unit 13 includes a pseudo conversation processing unit 130
And a word-of-mouth information input unit 135. The pseudo-conversation processing unit 130 updates parameters between agents by using knowledge information held by each agent in updating between agents.
Include human-like content such as emotions for words and correlations between words, and adapt the update result history of this agent according to the characteristics of the agent, so that the user can feel as if the agents had a pseudo-conversation. Such a result is obtained as a conversation history, and includes an update log collection unit 131, a document conversion unit 132, and a pseudo conversation document notification unit 133. The update log collecting unit 131 has a function of collecting a communication log between agents. The document converting unit 132 refers to each profile of the agent and obtains a corresponding pattern from a conversation decoration pattern 134 prepared in advance. To convert the communication log into a readable pseudo-conversation document. Pseudo conversation document notification unit 1
Reference numeral 33 has a function of notifying the user of the pseudo-conversation document by e-mail or web.

The word-of-mouth information input unit 135 wants to notify only an agent who has a hobby with the agent by the compatibility judgment based on the knowledge information of the agent when the agent wants to notify certain information to another agent. It is a means for notifying a user in a virtual community having a preference without instructing a partner to notify information and performing an effective advertisement, and is composed of an agent searching unit 136 and an information transmitting unit 137. Agent search unit 136
Obtains information transmission from the user and obtains his / her own individual knowledge information D
The information transmitting unit 137 has a function of searching for one or more agents that have a preference with reference to B152. The information transmission unit 137 transmits information to the searched agent or obtains information that matches its own preference. It has a function to respond depending on information.

The knowledge information held by the agent includes profile information including words and relationships between words, emotions for the words and relationships between words, the degree of reaction to the emotions (reaction value), and user preferences. And an individual knowledge information database (DB) 15
2 is constructed. In addition, a common knowledge information DB1 serving as a reference base for constructing the individual knowledge information DB 152
51, the individual knowledge information DB 152 inherits the relationship and emotion between words held by the common knowledge information DB 151 as knowledge information, thereby preventing unsolicited behavior between agents.

The individual knowledge information DB 152 grows with the passage of time. As will be described later, the individual knowledge information DB 152 appears in the form of inquiries and notifications through conversations with other agents.
The unknown words can be stored as homework, and the relations, emotions, and reaction values of the words can be obtained separately in an interactive form between the agent and the user. Also, some items of the agent profile entered or entered by the user at the time of initial registration or at any time are automatically registered as agent knowledge information. For items registered as an agent profile, user name and password must be entered, and age, hobby, gender, blood type, birthday,
Hometown and occupation are optional input items. In addition, an agent of a user who has joined from an existing web community site automatically registers knowledge items corresponding to characteristics of the joining web community site.

The knowledge information DB management unit 14
41 and a homework issuing unit 142. Forgetting processor 14
1 indicates each individual knowledge information D associated with communication between agents.
It has a function of checking the frequency of reference to the word B152 and comparing it with a predefined threshold, and deleting the word from the individual knowledge information DB 152 depending on the result. The homework issuing unit 142 accumulates the corresponding word if the word is unknown or does not have a degree of emotional reaction to the word with respect to the word queried by another agent in the communication between the agents. It has a function of issuing a response request to the user as homework when the number of the given words reaches a predetermined amount.

Further, if the inquired word is an unknown word, the lodging cost issuing unit 142 checks whether or not the word exists in the common knowledge information DB 151, and the word is stored in the common knowledge information DB 151. If the word does not exist, the word is registered in the individual knowledge information DB 152 of the agent, and if the word exists in the common knowledge information DB 151, the first user interface unit 143 prompts the user to input an emotion for the word. A second user interface unit is provided for inheriting the emotion for the word and registering it in the individual knowledge information DB 152 of the agent, and for prompting the user to input the degree of reaction of the emotion. In addition, the knowledge information DB
When registering the words in the individual knowledge information DB, the management unit 14 further defines the synonymous relationship of the words, the words in the inclusive relationship, and the definitions of the types of emotions for the words with the user terminal. It also has a knowledge information DB definition unit 145 that performs the dialogue between them. The knowledge information DB definition unit 145 also includes a function of setting and registering profile information.

FIGS. 3 to 10 are views cited for explaining the operation of the embodiment of the present invention shown in FIGS. 1 and 2.
A flowchart showing each procedure of agent definition, agent conversation mechanism, communication between agents, compatibility evaluation determination processing, knowledge information registration processing, forgetting processing, agent pseudo-conversation processing, and word-of-mouth information transmission is there. FIGS. 11 and 12 are also diagrams cited for explaining the operation of the embodiment of the present invention, and are diagrams showing an example of the correlation between words and the relationship between emotions and reaction values for words, respectively, and FIG. FIG. 3 is a diagram quoted for explaining the encounter of the present invention.

Hereinafter, the operation of the embodiment of the present invention will be described in detail with reference to FIGS. Prior to the description of the operation, as described above, the present invention is characterized in that communication between users is performed by determining compatibility between agents in a virtual community, as described above. It depends on the evaluation judgment method. In order to evaluate and judge the compatibility between agents, not only a simple comparison judgment based on the degree of coincidence and the number of coincidences of the attribute information of the user (profile of user's interest, affiliation, etc.) possessed by the agent, but also the agent has Knowledge information (emotion for words (like / dislike, etc.)
And the degree of emotion (reaction value) and the relationship between words).

Therefore, in the present invention, the knowledge information possessed by the agent is managed as the knowledge information DB 15 with the following configuration. That is, the knowledge information DB 15 has words and relationships between words (inclusion, synonym). Here, inclusion indicates the parent-child relationship of the word. Synonymous indicates a relationship having the same meaning between words. FIG. 11 shows, as an example, the correlation between words when the words are expanded into three layers.
Here, the hobbies include “music” and “cooking”, and “menu” and “menu” have a synonymous relationship. Furthermore, "cooking" and "France" are associated. Also,
The knowledge information DB 15 also has emotions (likes, dislikes, etc.) and degree values (reaction values) of reactions to the emotions with respect to the words and the relationships between the words. FIG. 11 also shows an example of the reaction value to the emotion. Here, 100 reaction values from likes (+) to dislikes (−) are set for the feeling of likes and dislikes, and are set by the user. Others are the same.

Knowledge information DB1 used in the present invention
5 is a common knowledge information DB 151 that is commonly used in the system.
And individual knowledge information DB1 that each agent has uniquely
52 types. The common knowledge information DB 151 is a reference base of the individual knowledge information DB 162, and the individual knowledge information D
B152 is a relationship between words in the common knowledge information DB 152,
Inherit emotions. This unifies the view of the world of each agent in the virtual community.

The definition of the agent follows the flowchart shown in FIG. That is, profile information definition (step S31), action setting information definition (step S31)
32), the processing is performed in the order of registration of profile information and action setting information in the individual knowledge information DB 152 (step S33). The information possessed by the agent includes knowledge information (words, relations between words, emotions, reaction values) and agent profiles that define the personality of the agent and action guidelines. The information included in the agent profile is as described above, and the action setting information of the agent includes aggressiveness, personality, whim, patience, and atmosphere. Positiveness indicates the degree of willingness to speak actively during conversation, and personality indicates the degree of willingness to notify or ask questions about the conversation. Whims also indicate the degree to which conversations are reasonably conducted or whether discussions are skipped.
Atmosphere indicates the degree of ease with which other agents can speak. These are defined by the user, and the profile information and action setting information defined here are stored in the knowledge information definition unit 1 in the knowledge information management unit 14.
It is recorded in the individual knowledge information DB 152 under the control of 45.

Next, the mechanism of communication by the agent will be described. As shown in FIG. 12, the virtual community includes imaginary towns (roads R, stores, and event venues C).
It is considered that agent B moves in the city. The action guidelines at that time are: (1) heading to an agent who is likely to talk to the surroundings, (2) heading to an object of interest existing in the city, and (3) wandering at random. In (1), the agent requests the city to search for agents present around him. If there is an agent who does not exist, the atmosphere (ease of talking)
To move. In (2), the agent having knowledge sets the agent whose response value is the largest among his or her own knowledge as his / her maximum interest, and among the objects of interest arranged in the city, the object of interest that matches his / her maximum interest. Move towards.
The objects of interest are arranged in towns covering all the first to third hierarchical levels of the words included in the common knowledge information DB 151. In (3), when there is no other agent around the ignorant agent, the agent randomly walks around the city without any purpose. The priority of (1) and (2) is (1)
Is determined by comparing the value of the atmosphere with the value of the reaction value with respect to the maximum interest of the user (2), and the larger value is given priority.

FIG. 4 is a flowchart showing a mechanism of conversation between agents. The agent acquires information on the city to which the agent moves, from a database (not shown), and moves the city along the map (steps S41 and S4).
2). Therefore, when the user encounters another agent and encounters another agent, it is determined whether to speak or speak, and the conversation is started according to the result (step S).
43-S46). If you did not meet the agent,
The destination is checked, it is confirmed that the destination has been reached, and conversation is started (steps S47 to S49). If the agent on the move encounters an agent to talk to, or encounters another agent on the way to (another agent or interest), he speaks from one of them based on the trigger. When the agent encounters the person he is trying to talk to, he speaks himself. If agents accidentally encounter each other during the movement, the atmosphere of the other agent is compared with the strength of the target object (the atmosphere of the headed object or the reaction value to the object of interest), and a new encounter is made. If the agent is large, he speaks to the person he encountered. If not, continue your move. On the other hand, if the other agent speaks, the same determination as above is performed,
Determine whether to respond. If you answer, start a conversation; otherwise, continue your movement. If the other party does not speak, it is determined in the same manner as to whether or not to speak.

Communication between agents using the knowledge information DB 15 follows the flowchart shown in FIG. The utterance pattern of the agent includes inquiry, notification, and notification (greeting). Which of these utterance patterns to use depends on
Determined by the nature of the profile information. The notification of the utterance pattern (greeting) is used when there is no word to be notified by itself in a special form of notification, and at the time of the first utterance (step S51). First, when the user speaks himself (step S52), the agent X searches for a word that triggers a conversation. Here, the word A based on one's preference
A response value of the emotion for the word A and the word A is retrieved from the individual knowledge information DB 152 held by the agent (step S5).
3). The agent X notifies the agent Y of the response value of the word A and the emotion, and inquires about the emotion of the agent Y regarding the word A (step S54).

Agent Y receives an inquiry from agent X, and agent Y searches synonymous relationship (alias) of word A from individual knowledge information DB 152. If a synonymous relationship is found, the agent Y searches the individual knowledge information DB 152 for a response value of emotion to the found word. When the response value of the emotion is found, the agent Y answers the response value of the emotion to the agent X (step S55). As a result, the agent Y calculates the favorable impression for the agent X (step S56). The calculation of the favorable sensitivity is performed by the compatibility evaluation determination unit 12, and details thereof will be described later with reference to a flowchart illustrated in FIG. When the answer to the question is obtained and the end condition is satisfied (step S5
In step 7), an end greeting is given to end the processing. If the end condition is not satisfied, the processing from step S52 is repeated.

On the other hand, Agent X is Agent Y
, And the agent X also calculates the favorable feeling for the agent Y. If no synonymous relationship is found, or if no emotion reaction value is found, agent Y becomes agent X
Is inquired about the inclusion relationship (parent) of the word A. Next, the agent X receives an inquiry from the agent Y. Then, the agent X searches the individual knowledge information DB 152 for the inclusion relationship (parent) of the word A. When the inclusion relation (parent) is found, the agent X stores the reaction value of the emotion for the found word A 'in the individual knowledge information DB1.
Search from 52. If the emotion response value is found, the agent X notifies the agent Y of the word A ′ and the emotion response value, and inquires about the emotion of the agent Y for the word A ′. Thereafter, the above processing is continued by replacing the word A with the word A ′. On the other hand, when the inclusion relation (parent) is not found or when the reaction value of the emotion is not found, the agent X notifies the agent Y of the end of the conversation. Then, the agent Y receives the notification of the conversation end from the agent X.

In the process of step S52, if the user does not speak and does not satisfy the termination condition (step S58), the agent receives a question from the partner (step S59), and Is answered if a valid answer is possible (step S6).
0, S61), and if not, it is recorded as homework (step S61). The processing of homework items will be described later with reference to the knowledge information input flow shown in FIG. And
Like the step S56, the positivity is calculated (step S6).
3).

The calculation of positivity follows the following procedure. That is, according to the flowchart shown in FIG. 6, first, the emotion reaction value is acquired from the knowledge information DB 152 (step S7).
1). Here, the type of emotion is checked (step S7).
2) If the emotion type is the same, the vector of the reaction value is further checked to see if the value is + or-, and if it is the same, the favorable impression is greatly increased (step S7).
4). The rate of increase in positivity is proportional to the magnitude of the value. on the other hand,
If the emotion types are the same and the reaction value vectors are different, the favorable impression is significantly reduced (step S75). The rate of decrease in favorable sensitivity is proportional to the magnitude of the value. In addition, as a result of checking the sign of the reaction value (step S76), if the types of emotions are different and the reaction value vectors are equal, the liking is slightly increased (step S77). If the emotion type is different and the reaction value vector is also different, the adjustment is made by slightly lowering the favorable impression.

Based on the result of the calculation of the positivity as described above, the evaluation of compatibility between agents is determined to be good when the positivity of both agents exceeds a threshold value set in advance by the system. As a result, the agent asks the user whether the user's mail address or chat ID may be notified to the partner agent, and executes the notification when both users understand the notification. Then, a direct exchange between the users is performed.

In the communication between the agents, if the agent does not know the word inquired by another agent or does not have a response value of the emotion to the word, the agent recognizes the word as an unknown word or a well-known word. Accumulate as missing words (words that know the word but have no response value). Here, we call this homework. When the accumulated amount of homework exceeds the threshold, the agent notifies the user of a homework answer request by e-mail. The user accesses the web page notified by e-mail, and performs Q & A with the agent in an interactive manner to register information (education) on words. The educated words are held in the individual knowledge information DB 152 as agent knowledge.
Further, the homework can be registered by the user in an information input form at an arbitrary timing.

Here, the registration operation of the knowledge information will be described with reference to FIG.
This will be described with reference to the knowledge information input flowchart shown in FIG. As described above, the homework issuing unit 142 accumulates, as a homework, unknown words or unknown words appearing in conversations between agents by the agent. Further, the agent determines whether the number of homework exceeds the threshold, and when the number exceeds the threshold, the agent notifies the user of a homework answer request by e-mail. Upon receiving the e-mail, the user accesses a web page for homework response and registers information in an interactive manner with an agent. At this time, the user interfaces 143 and 144 function.

In the flowchart shown in FIG. 7, first, the agent determines whether the homework is an unknown word or a word that is not well known (step S81). If the homework is an unknown word, and the user confirms the registration intention (step S82), and if there is a response after registration, it is further searched whether or not the word is in the common knowledge information DB 151 (step S82). Step S83). The corresponding word is common knowledge information D
If it is not in B151, the corresponding word is
2 and the emotion is registered in the first user interface unit 1
The registration is carried out in an interactive manner via 43. If the word is in the common knowledge information DB 151, the emotion of the word is inherited and registered in the individual DB (step S84). Further, the knowledge information definition unit 145 can also register the type of emotion, the reaction value of emotion, the inclusive relation, the synonymous relation, or the associative relation to be described later in an interactive manner (step S).
85-S89).

Note that the individual knowledge information DB 152 can have knowledge (words, word relationships, emotions) not existing in the common knowledge information DB 151 as unique knowledge. This is used to increase the individuality of individual agents. For example, the individual knowledge information DB 152 has an original word relationship called association. Associative is not inclusive or synonymous,
Some kind of relationship that connects words. For example, the inclusion relationship (parent) of “apple” is “fruit”, the synonymous relationship (alias) is “apple”, and the association relationship is “red”
For example, "round". This is used to extend the pseudo-conversation by communication between agents.
("Speaking of apples is red.")

Incidentally, the individual knowledge information DB 152 reduces knowledge by a mechanism called forgetting. This is used to prevent the knowledge of the individual knowledge information DB 152 held by each agent from becoming saturated in the virtual community and to reduce the information amount of the individual knowledge information DB 152. As shown in the flowchart of FIG. 8, first, a word is read from the individual knowledge information DB 135 (Step S).
91). The forgetting processor 141 checks whether or not the word has been referred to (step S92).
If the word read from No. 2 is not referred to, a comparison is made between the non-referenced period and the half-life (step S9)
3). Here, when it is determined that the non-referenced period> half-life, the storage intensity is reduced by half (step S9).
4). Further, a preset forgetting threshold is compared with the updated memory strength (step S95), and when the relationship of forgetting threshold <memory strength is established, the individual knowledge information DB
The corresponding word is deleted from 152 (step S96).
On the other hand, when the word read from the individual knowledge information DB 152 is referred to in the step of checking the reference frequency of the word (step S92), the storage strength is increased (step S97), and the number of references and the threshold are compared. When the relationship of reference count> threshold is established, the half-life is set long (step S99).

That is, the forgetting by the forgetting processing unit 141 is deleted from the individual knowledge information DB 152 when the storage strength of a certain word becomes 0. The value of the memory intensity decreases by half with the passage of time due to the half-life. When a word is referred to by communication between agents, the memory strength of the word increases. As the number of times a word is referred to by communication between agents increases, the half-life of the word increases. The memory strength is also used for word search of the individual knowledge information DB 152. That is, when there are a plurality of candidates as a result of a word search under a certain condition, a value having a higher storage strength is more likely to be stochastically selected as the first candidate.

By applying the above-described knowledge information management to communication between agents, the following mechanism can be realized. One is the realization of a pseudo-conversation of a human agent, and the other is the realization of word-of-mouth usage. In the former, the communication between agents is query (inquiry), inform (notification), iq (notification & inquiry), and refuse (deny). In this case, the contents (parameters) of the inquiry and the notification are as follows. That is, the type of inquiry is
Response value, inclusion (parent), inclusion (child), synonym, association, profile, greeting (start), and the type of notification is reaction value,
Parent word, child word, synonym, association, profile, greeting (start), greeting (end). At this time, the communication pattern is inquiry, notification or notification & inquiry, notification is notification or inquiry, notification & inquiry is notification or notification & inquiry. Data conversion is performed as if a human conversation was performed on the system side. The data conversion is performed according to the characteristics of the agents (male / female, young / middle-aged / old, polite / rough, from Tokyo / Kansai) set in the agent profile. The user can refer to the communication between the agents on the web page as the agent's pseudo conversation history. Further, the pseudo conversation history can be notified to the user by e-mail according to the user setting.

FIG. 9 shows the pseudo conversation process of the agent. In FIG. 9, first, the agent X communicates with the agent Y, and a log of the communication at that time is output to a communication log file (step S101). This communication log is collected by the communication log collecting unit 131. Next, a conversation decoration pattern to be used for conversion is obtained from the conversation decoration template file group 134 prepared in advance with reference to the agent profile for each of the agents X and Y (step S102). Document conversion unit 13
In step 2, the communication log is converted into a pseudo-conversation log in accordance with the obtained conversation decoration pattern (step S10).
3). The communication log converted in this case includes both agents X and Y. Then, the pseudo-conversation log converted into a readable format is registered in the web page via the pseudo-conversation document notification unit 133, and its URL (Uniform Resou
rce Locater) to the user. Further, the pseudo conversation log is transmitted to the user by mail according to the user setting (step S104). The user refers to the notified web page and refers to the pseudo conversation log. Or, refer to the pseudo conversation log by e-mail.

Next, word-of-mouth information transmission (advertisement /
Question) is explained. The conversation between the agents described above is realized by two agents meeting on the virtual community specifying each other and communicating with each other. On the other hand, in the virtual community, the agent can know the agent (s) that are compatible with itself by evaluating the compatibility based on the knowledge information of the agent. It can be performed. This means that the user of the agent can communicate indirectly via the agent with the preferred person without knowing the other users of the virtual community (without specifying the other party). . In this way, if the user wants to notify some information to another person, the information is sent only to other users of the virtual community who have a preference by requesting the information notification from the agent without explicitly specifying the other party. can do. As a result, information can be transmitted only to those who like it, and effective advertising, promotion,
Announcements, effective inquiries, consultations, confession, etc. can be made. When the user who receives the information makes a response to the information transmission (a request for providing detailed information to an advertising source, a response to a question, and the like), the user responds via the agent. Therefore, users can anonymously perform advertisements, Q & As, etc. without knowing each other directly.

FIG. 10 is a flowchart showing the word-of-mouth information transmission process. In FIG. 10, first, the user requests the agent for the content (advertisement, question, etc.) to transmit information. The requested agent refers to its own knowledge information via the agent search unit 136 (step S111), and searches for a plurality of agents (steps S112) that have a preference. The agent sends information to a plurality of agents (step S)
113). Other users obtain information (advertisements, questions, etc.) that matches their preferences. It responds to requests for detailed information to the advertising source, answers to questions, etc. as necessary.

A program for realizing the functions of the agent management unit 11, compatibility evaluation determination unit 12, conversation processing unit 13, and knowledge information management unit 14 in FIG.
The present invention is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read by a computer system constituting the community server 1, and the computer sequentially reads and executes the program. Is achieved. Also,
Here, the computer system includes an OS and hardware such as peripheral devices.

Further, if the computer system uses a WWW system, it also includes a homepage providing environment (or display environment). The computer-readable recording medium includes a floppy (registered trademark) disk, a magneto-optical disk, a ROM, and a CD.
-A storage medium such as a portable medium such as a ROM or a hard disk built in a computer system. Further, the computer-readable recording medium is a volatile memory (RA) in a computer system that is a server or a client when a program is transmitted through a network such as the Internet or a communication line such as a telephone line.
As shown in M), a program holding a program for a certain period of time is also included.

The above program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the "transmission medium" for transmitting a program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. Further, the program may be for realizing a part of the functions described above. further,
What can realize the above-mentioned function in combination with a program already recorded in the computer system, that is, a so-called difference file (difference program) may be used.

Although the embodiments of the present invention have been described in detail with reference to the drawings, the specific configuration is not limited to the embodiments, and includes a design and the like within a range not departing from the gist of the present invention. It is.

[0070]

As described above, according to the present invention, in a computer network system in which agents acting as agents of users communicate with each other on a computer network, the agents can communicate with each other based on the user information held by the agents. By evaluating the compatibility, it communicates with other users and forms a virtual community. By this, users search for users who prefer their own tastes and interact directly with others. Without this, it is possible to find other people who have a good taste and to communicate with each other. The user only needs to refer to the communication, compatibility evaluation, judgment process, and results of the agent's communication with other agents as appropriate, so that the burden on information exchange can be greatly reduced. When interaction between compatible agents progresses and reaches a certain threshold value, it is possible to provide a means for direct interaction between agent users, such as chat and mail exchange. In addition, the user can interact with another person who is interested without sharing his / her own personal information with another person, and the interested agents can naturally form a virtual community. This allows
The burden on the user during communication is greatly reduced.

According to the present invention, it is possible to realize a pseudo conversation between human agents. In other words, in communication between agents, by using the knowledge information held by each agent, the parameters used in communication between agents include contents with a human touch (emotions related to words, correlation between words, etc.). The system can adapt the history of the communication results of the agents according to the characteristics of the agents on the system side, so that the user can obtain a result as if the agents had a pseudo-conversation as the conversation history. Also, word-of-mouth information such as advertisements and questions can be transmitted. That is, when an agent wants to notify certain information to another agent, the information can be notified only to an agent interested in the agent by a compatibility determination method based on the knowledge information of the agent. Therefore, when the user wants to notify some information to another person, it is possible to notify the information to the user of the virtual community having a preference without explicitly specifying the person to notify. As a result, effective advertisement / publicity, useful Q & A, and friendly consultation can be performed. Further, it is possible to input knowledge information without stress for the user. That is, when communicating with another agent using knowledge information, if an unknown word not found in its own knowledge information comes out, the agent can accumulate learning about that word as homework. When the agent inquires of the user or the user arbitrarily registers with the agent, the user can input information in an interactive manner with the user-friendly agent, and thus input the user information with less stress. And registration of new knowledge is possible.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a connection configuration example of a computer network system in which the present invention is realized, and is a diagram cited for explaining a mechanism of an online community.

FIG. 2 is a block diagram showing the internal configuration of the communication server in FIG.

FIG. 3 is a flowchart cited for explaining the operation of the embodiment of the present invention shown in FIGS. 1 and 2;

FIG. 4 is a flowchart cited for explaining the operation of the embodiment of the present invention shown in FIGS. 1 and 2;

FIG. 5 is a flowchart cited for explaining the operation of the embodiment of the present invention shown in FIGS. 1 and 2;

FIG. 6 is a flowchart cited for explaining the operation of the embodiment of the present invention shown in FIGS. 1 and 2;

FIG. 7 is a flowchart cited for explaining the operation of the embodiment of the present invention shown in FIGS. 1 and 2;

8 is a flowchart cited for explaining the operation of the embodiment of the present invention shown in FIGS. 1 and 2. FIG.

FIG. 9 is a flowchart cited for explaining the operation of the embodiment of the present invention shown in FIGS. 1 and 2;

FIG. 10 is a flowchart cited for explaining the operation of the embodiment of the present invention shown in FIGS. 1 and 2;

FIG. 11 is a diagram cited for describing the operation of the embodiment of the present invention, and is a diagram illustrating an example of the correlation between words and the relationship between words and emotions and reaction values.

FIG. 12 is a diagram cited for explaining an operation of the embodiment of the present invention, and is a diagram cited for explaining an encounter between agents.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Communication server, 2 ... Computer network, 3 ... Personal computer, 4 ... PDA terminal, 5 ... Mobile telephone, 11 ... Agent management part, 12 ... Compatibility evaluation judgment part, 13 ... Conversation processing part, 14 ... Knowledge information DB management part , 1
5 ... Knowledge information DB, 121 ... First knowledge information inquiry section, 1
22: second knowledge information inquiry unit, 123: third knowledge information inquiry unit, 124: first knowledge information notification unit, 125: compatibility evaluation unit, 126: compatibility determination unit, 130: pseudo conversation processing unit,
131: communication log collection unit, 132: document conversion unit, 133
... Simulated conversation document notification unit, 134 ... Conversation decoration pattern file, 135 ... Word-of-mouth information input processing unit, 136 ... Agent search unit, 137 ... Information transmission unit, 141 ... Forgetting processing unit, 142 ... Homework issuing unit, 143 ... User Interface parts A, 144 ... User interface parts B, 145 ...
Knowledge information definition part, 151 ... common knowledge information DB, 152 ...
Individual knowledge information DB

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Tomohiko Ninomiya F-term in NTT Comware Co., Ltd. 1-9-1, Minami-ku, Tokyo 5B075 ND03 ND35 NK32 NK43 PQ36 UU08

Claims (19)

[Claims] 1. A communication server in a computer network system in which communication is performed between user terminals by communication between agents, wherein the compatibility between the users is evaluated and determined based on knowledge information held by each of the agents. A communication server comprising: compatibility evaluation determination means; and conversation processing means for communicating with another user by reflecting the evaluation result. 2. The knowledge information held by the agent includes profile information including words, relationships between words, emotions with respect to the words and relationships between words, degrees of reaction to the emotions, and user preferences. The communication server according to claim 1, wherein the communication server is constructed as an individual knowledge information database (DB). 3. A common knowledge DB serving as a reference base for constructing the individual knowledge information DB, wherein the individual knowledge information DB includes a relation between words and emotions which the common knowledge information DB has as knowledge information. 3. The communication server according to claim 2, wherein: 4. The compatibility evaluation determining means searches for a word that triggers a conversation based on an individual knowledge information database of the agent as the agent moves, and determines a reaction of emotion to the word according to its own profile. First knowledge information inquiring means for obtaining the degree, notifying the content of the agent to the encountered agent, and inquiring about the feeling for the word; synonymous relation of the word, and the degree of reaction of the feeling to the word having the same meaning. By searching the individual database of the encountered agent,
4. The communication server according to claim 1, further comprising: first knowledge information notifying means for notifying the inquired agent. 5. The compatibility evaluation determining means, if no synonymous relation of the word is found or if a degree of emotional reaction is not found, further includes the inclusive relation of the word to the inquired agent. And a second knowledge information inquiry means for inquiring about the inclusive relation of the word, obtained by searching the individual knowledge information DB of the encountered agent, and determining the degree of the reaction to the emotion of the included parent-child relationship with the emotion. 5. The communication server according to claim 4, further comprising: third knowledge information inquiry means for notifying an agent who has made an inquiry and inquiring about feelings regarding the included words. 6. The compatibility evaluation determining means obtains the degree of emotional reaction from both of the agents, and calculates the favorable sensitivity based on a predetermined arithmetic expression with reference to the degree of reaction. The communication server according to claim 4, further comprising: a compatibility determining unit configured to perform a compatibility evaluation determination by comparing the calculated favorable positivity with a preset threshold. 7. The conversation processing means refers to communication log collecting means for collecting a communication log between the agents, and a profile of each of the agents,
Document conversion means for obtaining a corresponding pattern from a conversation decoration pattern prepared in advance and converting the communication log into a readable pseudo-conversation document; The communication server according to claim 1, further comprising a notification unit. 8. The conversation processing means obtains information transmission from the user, refers to its own knowledge information DB, and searches for one or more agents that have a preference, and an agent search means; 4. The communication server according to claim 1, further comprising: an information transmitting unit that transmits information to an agent or obtains information that matches the user's preference, and responds depending on the information. 9. According to the communication between the agents, a frequency of referring to a word in each of the individual knowledge information DBs is checked and compared with a predetermined threshold value. 7. The communication server according to claim 1, further comprising a forgetting means for deleting the information from the DB. 10. The communication between the agents,
If a word inquired by another agent is unknown or does not have a degree of emotional reaction to the word, the corresponding word is accumulated and used when the accumulated word reaches a predetermined amount. 7. The communication server according to claim 1, further comprising homework issuing means for issuing an answer request as homework to a person. 11. If the queried word is an unknown word, it is checked whether the word exists in the common knowledge information DB. If the word does not exist in the common knowledge information DB, First user interface means for registering a word in the individual knowledge information DB of the agent and prompting the user to input an emotion for the word; if the word exists in the common knowledge information DB, 11. The system according to claim 10, further comprising second user interface means for inheriting and registering in the individual knowledge information DB of the agent, and for prompting the user to input the degree of the emotional reaction. Communication server. 12. When registering the word in the individual knowledge information DB, a definition of each word having a synonymous relationship, an inclusive relationship of the word, and a type of emotion with respect to the word, 12. The communication server according to claim 11, further comprising a knowledge information DB defining means for performing an interactive manner between the information servers. 13. A communication method using agent knowledge information by a communication server in a computer network system in which communication is performed between user terminals by communication between agents, the communication method being based on knowledge information held by each agent. A communication method using agent knowledge information by a communication server, comprising: evaluating and determining compatibility between users; and performing communication with another user by reflecting the evaluation determination result. 14. A virtual community in which users having the same attribute gather by evaluating compatibility between the users based on knowledge information held by each of the agents, and forming a virtual community. A communication method using agent knowledge information by the communication server described in. 15. A server program used for a communication server in a computer network system in which communication is performed between user terminals by communication between agents, wherein the users are connected to each other based on knowledge information held by each of the agents. A server program for causing a computer to execute a step of evaluating and determining compatibility of the computer and a step of performing communication with another user by reflecting the result of the evaluation determination. 16. A step of examining a frequency with which a word registered as the knowledge information is referred to in accordance with the communication between the agents, and comparing the frequency with a predefined threshold value, depending on the result, And a step of deleting the information from the knowledge information. 17. The communication between the agents,
When the word inquired by another agent is unknown or does not have a degree of emotional reaction to the word, the step of storing the corresponding word, and when the stored word reaches a predetermined amount. Issuance of a response request as homework to the user. 18. A method according to claim 18, wherein, when the queried word is an unknown word, a step of checking whether or not the word exists as common knowledge information; Registering the word as knowledge information constructed for each of the agents, and prompting the user to input an emotion for the word; if the word exists as the knowledge information, inheriting the emotion for the word; 16. The communication server according to claim 15, further comprising the steps of: registering as knowledge information of the agent; and prompting the computer to input the degree of reaction of the emotion. 19. A computer-readable recording medium storing a server program used for a communication server in a computer network system in which communication is performed between user terminals by communication between agents, wherein the knowledge held by each of the agents is provided. A recording medium recording a server program for causing a computer to execute a step of evaluating and determining compatibility between the users based on information, and a step of performing communication with another user by reflecting the evaluation result. .