JP7019984B2 - Information presentation system, information presentation method and program - Google Patents

Description

The present invention relates to an information presentation system, an information presentation method and a program.

In recent years, with the enhancement of the Internet environment, social network services (hereinafter referred to as SNS) have become widespread, and it has become possible to easily communicate between a plurality of users using texts and images. .. For example, LINE (registered trademark) and Facebook (registered trademark) messenger are typical SNS applications. These SNSs not only exchange information between one-to-one users, but also share information transmitted and received between a large number of users in a predetermined group (dialogue between a plurality of users) among all users in the group. It also has a function.

There are various methods for presenting information sent and received to users in the above SNS, and in addition to the general display showing the chronological order in which messages (information) are exchanged, the time interval of exchange is intuitive. There is a presentation method in which a message is displayed according to the interval of display positions on the screen corresponding to the time interval to visually indicate the time-series change (see, for example, Patent Document 1).

Further, in a dialogue with a large number of people, there is a presentation method having an interface for presenting the contents of the statements made by the users in different groups in association with each other (see, for example, Patent Document 2).
Further, there is an electronic conference room system that records a dialogue content of each user and displays a message for each user in chronological order like a dialogue interface of LINE (registered trademark) when playing back (for example, a patent document). 3).

Japanese Patent No. 6064019 Japanese Patent No. 6062027 Japanese Patent No. 5949843

However, in the above-mentioned information presentation method, when a large number of people have a dialogue, the font of the message displayed on the display screen and the image of the balloon are often the same or similar, and the message of a user other than itself is displayed. There is a problem that it is difficult to intuitively distinguish on the above display screen.

Further, in the information presentation method, when viewing the message exchange between users in the group constructed in the SNS, after a certain amount of message exchange is performed between other users, the user participates in this exchange from the middle. There are also users. If this mid-career user does not view and confirm all past message exchanges, it will detect messages that he or she wants to view, that is, that he or she is "interested" or "related" to. There is a problem that it is difficult.

The present invention has been made in view of such a situation, and is interested in the font of a message, the image of a balloon, etc., which are presumed to be "interested" by the user or "related" to the user. / Information presentation that can be highlighted (highlighted) compared to the image of a message with low relevance, and messages that are of interest / relevance to users other than itself can be intuitively perceived and easily distinguished. Provide systems, information presentation methods and programs.

In order to solve the above-mentioned problems, the information presentation system of the present invention has a dialogue when a plurality of users belonging to a predetermined dialogue group use each terminal and have a dialogue between the respective users via a network. It is an information presentation system that controls the display of dialogue data that each of the users wants to refer to from the data history, and has a speech storage unit that stores dialogue data of each of the users belonging to the dialogue group, and the user's speech storage unit. A browsing-oriented information estimation unit that estimates browsing-oriented information, which is a score indicating the degree of interest / association of each user, from the contents of each dialogue data for each dialogue data by a model corresponding to each of the users . The user obtains the score in accordance with the characteristic information of at least the speech information indicating the speech of the user in the dialogue data and the behavior information indicating the operation of the terminal of the user in the dialogue data. Highlight data that determines the highlight information indicating the display form of the dialogue data by associating the model construction unit generated by machine learning for each user with the viewing-oriented information of each of the dialogue data of the user for each user. It is characterized by including an extraction unit and a data display control unit that displays each of the dialogue data in the display form indicated by the highlighting information on the display screen of the terminal.

In the information distribution system of the present invention, the model building unit generates a similar subgroup based on the attribute information of the user in the dialogue group, generates the model for the similar subgroup, and the browsing-oriented information estimation unit generates the model. It is characterized in that the model of the similar subgroup corresponding to the attribute information of the user is used for estimating the score of the user who newly joins the dialogue group.
In the information presentation system of the present invention, the emphasized data extraction unit presents in a display form in which the user's perception becomes higher as the browsing-oriented information becomes higher, as compared with the dialogue data having a lower browsing-oriented information. It is characterized by determining the highlighting information.

The information presentation system of the present invention further includes a data analysis unit that performs natural language processing of dialogue data stored as a history for each user and extracts feature information of the dialogue data from at least the meaning and arrangement of words. It is characterized in that the browsing-oriented information estimation unit obtains the browsing-oriented information of each of the dialogue data for each user in response to the feature information.

In the information presentation system of the present invention, when the browsing-oriented information estimation unit estimates the browsing-oriented information, not only the dialogue data but also at least the dialogue data of other users in the same dialogue group as the user. The browsing-oriented information is obtained using behavioral information including the number of responses, the number of characters in the dialogue message in the response, the time until the response, and the presence / absence or combination of expansion of dialogue data of other users. It is characterized by estimating.

In the information presentation system of the present invention, when the browsing-oriented information estimation unit estimates the browsing-oriented information, not only the dialogue data but also at least the address and gender of another user in the same dialogue group as the user. It is characterized in that the browsing-oriented information is estimated by using the attribute information including any one or a combination of each of, age and occupation.

In the information presentation system of the present invention, the data display control unit displays the dialogue data by compression (reduction), expansion, font size, display color, blinking, or a combination as a display form for emphasizing the dialogue data. It is characterized by doing.

In the information presentation system of the present invention, the data display control unit presents contents other than texts of audio, still images, and moving images attached to the dialogue data as a display form for emphasizing the dialogue data. It is characterized by that.

In the information presentation method of the present invention, when a plurality of users belonging to a predetermined dialogue group use each terminal and have a dialogue between the respective users via a network, the user's history of dialogue messages is recorded. It is an operation method of the information presentation system that controls the display of the dialogue message that each wants to refer to, and the browsing-oriented information estimation unit is the speech storage unit in which the dialogue data of each user belonging to the dialogue group is stored. A browsing-oriented information estimation process that estimates browsing-oriented information, which is a score indicating the degree of interest / association of each user, from the contents of each dialogue data for each dialogue data by a model corresponding to each of the users . The model building unit obtains the score corresponding to at least the characteristic information of the utterance information indicating the utterance of the user in the dialogue data and the behavior information indicating the operation of the terminal of the user in the dialogue data. The model construction process in which the model is generated by machine learning for each user and the emphasis data extraction unit correspond to the viewing-oriented information of each of the dialogue data of the user for each user and display the dialogue data. The highlight data extraction process for determining the highlight information indicating the form, and the data display control in which the data display control unit displays each of the dialogue data on the display screen of the terminal according to the display form indicated by the highlight information. It is characterized by including processes.

In the program of the present invention, when a plurality of users belonging to a predetermined dialogue group use each terminal and have a dialogue between the respective users via a network, each of the above users from the history of the dialogue message It is a program that operates a computer as an information presentation system that controls the display of a dialogue message to be referred to, and the computer is used as a user in an utterance storage unit in which dialogue data of each user belonging to the dialogue group is stored. A browsing-oriented information estimation means that estimates browsing-oriented information, which is a score indicating the degree of interest / association of each user from the contents of the dialogue data, for each dialogue data by a model corresponding to each of the users, at least the user . The model for obtaining the score corresponds to the characteristic information of the utterance information indicating the utterance in the dialogue data and the behavior information indicating the operation of the terminal of the user in the dialogue data. A model building means generated by learning, a highlight data extraction means for each user, which corresponds to the viewing-oriented information of each of the dialogue data of the user and determines highlight information indicating a display form of the dialogue data. It is a program for functioning as a data display control means for displaying each of the dialogue data in the display form indicated by the highlighting information on the display screen of the terminal.

As described above, according to the present invention, a message font or an image of a balloon that is presumed to be "interested" by the user or "related" to the user is compared with an image of another message. To provide an information presentation system, information presentation method and program capable of intuitively perceiving a message having an interest / relevance in a message of a user other than itself and making it easy to distinguish. Can be done.
Further, according to the present invention, the above-described configuration makes it intuitive, as opposed to a display in which a message presumed to be "not interested" or "irrelevant (or low)" is intuitively perceptible. "Interested" or "related" by making a display that is not perceptible (for example, a more subdued or refrained display for "interested" or "related" messages). The message can also be intuitively perceptible.

It is a block diagram which shows the structural example of the dialogue system using the information presentation system by one Embodiment of this invention. It is a figure which shows the presentation example of the emphasis presentation performed to the terminal apparatus 20 by the information presentation system 10 in this embodiment. It is a figure which shows the configuration example of the information presentation system 10 in the dialogue system 100 of FIG. It is a figure which shows the structural example of the utterance history table stored in the utterance storage part 108 in this embodiment. It is a figure which shows the configuration example of the user attribute table stored in the attribute storage part 109 in this embodiment. It is a figure which shows the configuration example of the user action table stored in the action storage part 110 in this embodiment. It is a figure which shows the structural example of the action table stored in the action storage part 110 in this embodiment. It is a flowchart which shows the operation example of the dialogue system using the information presentation system of this embodiment.

The present invention provides a method for providing information that facilitates grasping the contents of dialogue messages exchanged between a large number of users in a group dialogue and identifying the speaker of the dialogue in the use of the dialogue system of the SNS. Here, in the present invention, browsing-oriented information scoring interests and relationships between any user in the group and other users in the same group is obtained, and the browsing-oriented information is other than a predetermined threshold value or more. Highlight the display of the user's dialogue message compared to other dialogue messages.
In addition, a message whose browsing-oriented information is less than the above threshold value is compressed (reduced) so as to be less noticeable (modest or restrained) compared to a message whose browsing-oriented information is more than the threshold value, and is viewed-oriented. Messages with information above the threshold value are compared with messages with more browsing-oriented information below the threshold value to be emphasized.
Here, the above-mentioned browsing orientation indicates the intention of the user who wants to refer to (browsing) dialogue data including words and the like having each of his / her own preference and relevance.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a configuration example of a dialogue system using an information presentation system according to an embodiment of the present invention. In FIG. 1, the dialogue system 100 includes an information presentation system 10, a terminal device 20, and a communication network 30.
The dialogue system 100 shown in FIG. 1 is used for transmitting / receiving and presenting dialogue data in a dialogue group composed of a plurality of users. Each of the terminal devices 20 is a terminal owned by each user who uses the dialogue system 100, and is a terminal used by a user who requests transmission / reception and presentation of information. In the present embodiment, the communication network 30 is, for example, the Internet, a LAN (Local Area Network), or the like.

In the present embodiment, the terminal device 20 is, for example, a general-purpose computer, a personal computer, a tablet terminal, a smartphone, or the like.
Further, the terminal device 20 inputs the voice spoken by the user by its own input unit (for example, a microphone), converts it into digital format voice data, and uses its own transmission unit and the communication network 30. It is transmitted as speech information to the information presentation system 10.

The information presentation system 10 presents the content based on the utterance information supplied from the terminal device 20 of a certain user to each terminal device 20 of all the users belonging to the dialogue group. Here, the information presentation system 10 is a server system that presents the contents to the terminal device 20, and is, for example, a general-purpose computer or a personal computer.
Further, the information presentation system 10 converts the voice data in the utterance information received from the terminal device 20 into text (character data), writes it in a storage unit (described later), and stores it.

Then, the information presentation system 10 obtains browsing-oriented information for each user based on the text stored in the storage unit, scoring interests and relationships with other users in the same group, and is based on the browsing-oriented information. The predetermined content is emphasized and presented to each user's terminal device 20.

Further, in the above configuration, the information presentation system 10 has been described as performing a process of converting voice data into text, but the terminal device 20 converts voice data into text, and the text is used as utterance information in the information presentation system 10. It may be configured to send.
Further, the information presentation system 10 transmits the voice data acquired from the terminal device 20 to an external server (not shown) having a conversion mechanism for converting the voice data into text, and acquires the converted text from this external server. It may be configured as follows.

Further, in the above configuration, the information presentation request from the user is input by the microphone included in the terminal device 20, but the present invention is not limited to this configuration.
For example, any device or sensor that can convey the user's intention may be used. For example, the character information may be input by a keyboard or a touch panel, or the information indicated by the user's gesture may be input by a camera provided in the terminal device 20.
In addition to the above, images that convey the user's intentions (emoticons, pictograms, LINE (registered trademark) stamps, etc.) and video content are configured to be input by the camera, or are saved in advance without going through a sensor. The structure may be such that the moving image content is used.

Further, the dialogue system 100 in the present embodiment is not limited to use only for dialogue between users (humans). For example, it may be a conversation including a chatbot in which a computer, which is a kind of an automatic answering system, a question answering system, and a dialogue system, extracts or generates a response content. In the case of these systems, the message presented to the terminal device 20 receives the content from the response generation unit inside the information presentation system 10 or the response generation unit installed outside, and with the user belonging to the group for the content. The degree of relevance of the above may be calculated and emphasized. Here, since chatbots are a well-known technique, detailed explanations are omitted.

FIG. 2 is a diagram showing an example of presentation of emphasis presentation performed by the information presentation system 10 in the present embodiment to the terminal device 20. In each of FIGS. 2A and 2B below, the utterance content of the user who owns the terminal device 20 is presented on the right side of the figure, and the utterance content of another user belonging to the dialogue group is presented on the left side of the figure. Is presented.

FIG. 2A shows the size of the speech bubble of the utterance (dialogue message) expanded or compressed according to the level of the score obtained by the information presentation system 10. Here, the compressed dialogue message can be expanded by clicking the balloon part or operating other buttons (as will be described later, the action of the user who performs this expansion or compression is also scored in the storage unit as action information. It is written and stored as history information for calculation).

FIG. 2B shows the color of the utterance (dialogue message) displayed on the display screen of the terminal device 20 and the color, font, and font size of the characters according to the level of the score obtained by the information presentation system 10. Is changed and presented. The color of the balloon and the color and font of the characters can be customized by the user in addition to the initial values (default values) in the system (similar to the above, the user's behavior due to the setting change in the customization). It may also be configured as behavioral information, which is written and stored in the storage unit as history information for score calculation by means.

Further, in the present embodiment, the method of optimized presentation in which the content (for example, the balloon of the dialogue message) is emphasized and presented on the display screen of the terminal device 20 corresponding to the browsing-oriented information is given, but the display method is described. Not limited to these. For example, by adding underlines and borders to the characters in the balloon, blinking the characters in the balloon, and replacing the pictograms with other contents such as videos or still images as a substitute for the characters in the balloon, the score of browsing-oriented information can be obtained. Accordingly, the display control of the content on the display screen of the terminal device 20 may be performed.

FIG. 3 is a diagram showing a configuration example of the information presentation system 10 in the dialogue system 100 of FIG. The information presentation system 10 includes an acquisition control unit 101, a data analysis unit 102, a model construction unit 103, a browsing-oriented information estimation unit 104, a similar browsing-oriented user estimation unit 105, an emphasis data extraction unit 106, a data display control unit 107, and an utterance storage. Each of the unit 108, the attribute storage unit 109, and the action storage unit 110 is provided, and the optimized presentation of the utterance content described with reference to FIG. 2 is controlled.

The acquisition control unit 101 is an external input interface for acquiring utterance information from the terminal device 20, and outputs the input utterance information to the data analysis unit 102. Further, the acquisition control unit 101 may be configured to include not only the function of inputting the utterance information from the terminal device 20 but also the function of directly inputting the utterance information from the keyboard, the microphone, various sensors and the like.

The data analysis unit 102 in the utterance information such as keyword extraction and vectorization from the utterance information so that the utterance information supplied from the acquisition control unit 101 can be used in the model construction unit 103 and the browsing-oriented information estimation unit 104. Numerical data as feature information (so-called feature quantity) is obtained. At this time, the data analysis unit 102 converts the utterance information (text data) into information that can be handled by a computer by a method of natural language processing, and uses machine learning technology for the relationships of the obtained words and the like. , Extract the feature information of the utterance information. For example, as natural language processing and machine learning technology, the above-mentioned features including numerical values and symbols indicating the co-occurrence and similarity of dialogue contents by using methods such as keyword extraction by tr-idf and vectorization using word2vec. Information may be extracted (written in the utterance log table of the utterance storage unit 108, which will be described later, and stored as historical data used to generate a model).

The model construction unit 103 stores mathematical formulas (functions) and rules used for calculating the score as browsing-oriented information. Here, the model building unit 103 is a model for calculating a score indicating interest and relevance to the user's utterance by using the information of each of the utterance storage unit 108, the attribute storage unit 109, and the action storage unit 110, which will be described later. Build (formulas and rules).
In the present embodiment, as an example of the model building unit 103, there is a configuration in which a model for holding feature information in which a score is given to a keyword having a high interest or relevance to a user is held as a database. Then, the browsing-oriented information estimation unit 104 compares the feature information extracted from the input utterance information with the feature information in the database, and performs a process of calculating the score as the browsing-oriented information.

Further, the model building unit 103 may be configured to generate a model to be used for estimation from feature information by, for example, a method of machine learning (learning with supervision, learning without supervision, reinforcement learning, etc.). In this case, the model construction unit 103 generates a basis function for score estimation (such as a linear regression function) as a model by using the feature information of each of the utterance storage unit 108, the attribute storage unit 109, and the action storage unit 110. .. Then, the model construction unit 103 uses the numerical values of the feature information of each of the utterance storage unit 108, the attribute storage unit 109, and the action storage unit 110 as teacher data, and optimizes each model of the user belonging to the group by machine learning. Perform processing.

Here, the model in the above description generates both a model for each user in the dialogue group and a model for each similar subgroup consisting of users having similar viewing-oriented orientations in the dialogue group. Is also good. The model for each user is used to search for utterance information corresponding to each user's browsing intention.
On the other hand, a model of a similar subgroup is prepared because, for example, a user who newly joins the group cannot generate a model for each user because there is little dialogue with each user in the group. Then, in the user classification, a model of a similar subgroup having the most similar tendency may be selected from the overall models, and the score of the browsing-oriented information may be obtained. Here, the user classification of the newly joined user is performed by searching for a similar subgroup including users having similar attribute data by using the attribute data of the user in the attribute storage unit 109. Then, for newly joined users, the extracted similar subgroup model is used for score estimation.

Further, each time each user makes a new utterance, the model building unit 103 updates the corresponding model based on the new utterance information. Here, the update frequency, that is, the update cycle can be arbitrarily set by the administrator of the information presentation system 10.
By this update, the browsing intention for each utterance content of the user is clarified, the accuracy of selecting the utterance information is improved, and the utterance information that the user wants to browse can be extracted.
Further, the model building unit 103 generates a plurality of models different for each user, for example, a browsing-oriented score, a content display preference score, an attribute information score, and the like, and these are used as a single model. Alternatively, it may be used as a combined synthetic model and configured to calculate the score.

The browsing-oriented information estimation unit 104 uses the utterance information obtained from the data analysis unit 102 and the model of the model building unit 103 to view-oriented information of each utterance information (degree of interest / relevance of the utterance content to be browsed). The score is estimated as (information indicating).
Further, the browsing-oriented information estimation unit 104 uses the above-mentioned preference model as the browsing-oriented information to obtain a score regarding the preference for the display form of the content on the terminal device 20, and asks the data display control unit 107 to obtain a score. And output.

The highlighted data extraction unit 106 corresponds to the score of the browsing-oriented information obtained by the browsing-oriented information estimation unit 104, and uses a method of highlighting and presenting each content (such as a presentation example shown in FIG. 2) as an optimized presentation method. decide. Further, the emphasized data extraction unit 106 not only changes the text according to the score obtained by the browsing-oriented information estimation unit 104, but also adds information on the preference of the display control of the content, and combines the contents that can be referred to inside and outside. It may be a configuration to be presented. As described above, as a substitute for the characters in the balloon, the pictogram may be replaced with other contents such as a moving image or a still image, and may be presented on the display screen of the terminal device 20 as a display form (presentation form). Here, as the score of the browsing-oriented information increases, the display form of the emphasized presentation that is easy for the user to perceive is determined as the emphasized presentation information as compared with the utterance information having a lower score.
Then, the emphasized data extraction unit 106 combines each utterance information with the emphasized presentation information for each utterance information (whether or not the emphasized presentation is performed, and if the emphasized presentation is performed, the method thereof), and the data display control unit 107 is combined with the emphasized presentation information. Output to.
Here, for example, a threshold value may be set for the score of browsing-oriented information, and the score above this threshold value may be highlighted.
That is, the emphasized data extraction unit 106 is easier for the user to perceive as a display of a message (dialogue message) of another user whose browsing-oriented information is equal to or higher than the threshold value, as compared with a message whose browsing-oriented information is less than the above set value. It is determined as the emphasized presentation information to be presented in the emphasized form.
Further, the emphasized data extraction unit 106 is more inconspicuous as the display form of the message of the user whose browsing-oriented information is less than the above threshold value, which is harder for the user to perceive as compared with the message whose browsing-oriented information is the set value or more. It is determined as emphasized presentation information that is presented in a compressed (reduced) form such as (modest or restrained).

The data display control unit 107 generates information presentation and, if necessary, processed content based on the highlight presentation information (optimized presentation information) input from the highlight data extraction unit 106, and the generated content is used as a terminal. It is transmitted to the device 20.
At this time, the data display control unit 107 performs processing processing on the display form of the content according to the highlighted presentation information from the data display control unit 107 (for example, the change processing shown in FIG. 2 is performed).

For example, the data display control unit 107 corresponds to the highlighted presentation information and has a high score of the user's browsing-oriented information, that is, with respect to the utterance content (message) of the browsing-oriented content, various effects (blinking) are applied to the text-only content. , Font, color change), and replace images and other content such as still images and videos with text. On the other hand, the data display control unit 107 hides the content of the message by compressing the utterance content having a low score of the user's browsing-oriented information as shown in FIG.

Further, the data display control unit 107 sets an explicit item (such as "Do you like this display? Yes / No") of the terminal device 20 in order to acquire the evaluation of each user for the display content of the utterance content. It may be configured to be displayed on the display screen to encourage the user to evaluate.
Then, when the response from the user to the explicit item is supplied from the terminal device 20, the acquisition control unit 101 writes and stores the response to the action storage unit 110.

The data display control unit 107 can arbitrarily change the display form of the content displayed along with the highlighted presentation information according to the operation information of the terminal device 20 (expansion of compressed message or font). Changes etc.). Then, when the changes are made by these users, the acquisition control unit 101 writes and stores these change processes in the action storage unit 110 as action information in the application.
In addition, each of the acquisition control unit 101, the data analysis unit 102, and the data display control unit 107 obtains acquired utterance content, analysis results, and content information (characters, voice, still images, moving images, etc.) used in the emphasized presentation. It is written and stored in the utterance storage unit 108.

FIG. 4 is a diagram showing a configuration example of the utterance log table stored in the utterance storage unit 108 in the present embodiment. Here, the utterance information will be described as a text data message. The utterance log table in FIG. 4 has at least a time, a user ID (identification), a message ID, a message body, and a corresponding message ID for each record. The time indicates the time when the message was sent. The user ID is identification information that identifies the user who sent the message. The message ID is identification information that identifies the message. The message body is the text data of the message. The corresponding message ID is identification information that identifies the message transmitted immediately before the message of the message ID.

In FIG. 4, for example, "I saw you!" Of the message ID "M21701101000003" is a reply to the corresponding message ID "M2171010010001" in the same record. The text of this corresponding message ID (message ID) "M2017010010001" is "Have you seen XX yet?". That is, the user with the user ID "U002" replies "I saw it!" To the message "Have you seen it yet?" Which is a question from the user with the user ID "U001".

From the above-mentioned utterance log table, the correspondence relationship of information and messages such as the number of dialogues between each user indicated by the user ID, the number of dialogue characters, the content of the dialogue, the length of time the dialogue was performed, and the response time are derived and combined. It may be configured to be memorized. Here, the number of dialogues indicates the number and frequency of simple dialogues independent of the amount (number of characters) of dialogue sentences. The number of dialogue characters indicates the number of characters in the text data in the message. The response time is obtained as a difference in time, is the time required to respond to a message from another user, and indicates each of the shortest, longest, and average values.

That is, a message having a high score of browsing-oriented information has a large number of dialogues and a large number of dialogue characters, and the response time is short. On the other hand, a message having a low score of browsing-oriented information has a small number of dialogues and a small number of dialogue characters, and has a long response time.
In addition to the index used as the score of the browsing-oriented information described above, the score may be configured based on the interrelationship between the number of characters and the interval of the response time.
The browsing-oriented information estimation unit 104 estimates the preference score for each of the messages that are the utterance information of other users by the model that estimates the score corresponding to the preference by using the information in the above-mentioned utterance log table. ..

Further, the model building unit 103 generates a model for estimating the score of the browsing-oriented information for each user by using the text data of the message for each user in the utterance log table.
Then, the browsing-oriented information estimation unit 104 inputs feature information such as text data which is utterance information of another user by using the model for each user, and obtains and gives a score for each of the utterance information.

FIG. 5 is a diagram showing a configuration example of a user attribute table stored in the attribute storage unit 109 in the present embodiment. The user attribute table in FIG. 5 has at least columns for the user ID, the age, gender, occupation, affiliation group, unmarried / married, child-bearing / non-children, etc. of the user indicated by the user ID for each record. .. The user ID is identification information that identifies the user. Age indicates the age of the user. Gender indicates whether the user is male or female. The occupation indicates the occupation of the user (for example, office worker, student, housewife, housewife, etc.). The affiliation group indicates which dialogue group the corresponding user belongs to. Unmarried / married indicates whether the user is unmarried or married. Child presence / absence indicates whether or not the user has children.

Further, in addition to those shown in FIG. 5, a column such as geographical information (whereabouts) such as birthplace, place of residence (address), place of work, hobbies, intentions, etc. may be provided. The information of this user attribute information is obtained by the response of the questionnaire survey to the user. In addition to this, if the user permits, it is used as information for obtaining the user's preference score, so that the web browsing history or application usage history that can be acquired from the user's terminal device 20 may be described. good. For example, the model building unit 103 forms a similar subgroup using the information in the user attribute table described above, and generates a model corresponding to the similar subgroup using the information in the utterance log table of the users of the similar subgroup. do.

FIG. 6 is a diagram showing a configuration example of a user action table stored in the action storage unit 110 in the present embodiment. The user action table in FIG. 6 has at least a time, a user ID, an action ID, an action ID performed immediately before, a message ID, ... For each record. The time indicates the time when the user took an action. The user ID is identification information that identifies the user who has caused the action. The action ID is identification information that identifies the type of action performed by the user. The immediately preceding action ID is identification information indicating the type of the action immediately performed by the user indicated by the user ID. The message ID is identification information that identifies the message.

FIG. 7 is a diagram showing a configuration example of an action table stored in the action storage unit 110 in the present embodiment. The action table in FIG. 7 has columns for action ID and action for each record. The action ID is identification information that identifies the type of action performed by the user. The action indicates a name indicating the type of action.

FIG. 8 is a flowchart showing an operation example of the dialogue system using the information presentation system of the present embodiment. The operation of the flowchart of FIG. 8 is started, for example, when a user joins a group to which the user belongs and has a dialogue with another user. In the operation explanation of the flowchart of FIG. 8 below, each of the model for each user and the model for each similar subgroup are generated as already described in the model construction unit 103 as a model for obtaining the score of the browsing-oriented information. , Perform in the accumulated state.

Step S1:
The browsing-oriented information estimation unit 104 determines whether or not the dialogue in the group to which the user belongs is the first time, that is, whether or not the user participates in the dialogue in the group for the first time. At this time, the browsing-oriented information estimation unit 104 refers to the utterance log table of the utterance storage unit 108, and determines whether or not it is the first participation depending on whether or not the user ID of the participating user is in the utterance log table. ..
Then, when the user ID of the participating user is in the utterance log table, the browsing-oriented information estimation unit 104 determines that this user is not the first participant, and proceeds to step S2.
On the other hand, if the user ID of the participating user is not in the utterance log table, the browsing-oriented information estimation unit 104 determines that this user is the first participant and proceeds to the process in step S3.

Step S2:
The browsing-oriented information estimation unit 104 reads a model (individual model) for obtaining a score of browsing-oriented information corresponding to this user from the model construction unit 103.

Step S3:
The browsing-oriented information estimation unit 104 requests the model construction unit 103 to extract similar subgroups corresponding to this user.
As a result, the model construction unit 103 extracts a similar subgroup consisting of users similar to the first-participating user from the attribute information of the first-participating user, and notifies the browsing-oriented information estimation unit 104.
Then, the browsing-oriented information estimation unit 104 reads the model (overall model) corresponding to the notified similar subgroup from the model construction unit 103.

Step S4:
After reading the model (individual model or overall model), the browsing-oriented information estimation unit 104 is in a standby state (input standby) waiting for the utterance information to be supplied. That is, the browsing-oriented information estimation unit 104 waits for the analysis result of the utterance information to be supplied from the data analysis unit 102.

Step S5:
The browsing-oriented information estimation unit 104 determines whether the utterance information has been input, or whether the action information, which is information on whether some action (screen operation, etc.) of the user has been performed instead of the utterance information, has been input. At this time, if the utterance information is input (yes), the browsing-oriented information estimation unit 104 advances the process to step S6. On the other hand, the browsing-oriented information estimation unit 104 advances the process to step S6 when the behavior information is input, that is, when the utterance information is not input (no).

Step S6:
The data analysis unit 102 performs a process of analyzing newly received utterance information (dialogue message received when the application for dialogue is started), and browse-oriented information estimation unit 104 for feature information of each utterance information. Output to. Here, the data analysis unit 102 analyzes each of the text data of the dialogue message, and outputs the feature information of the utterance content to the browsing-oriented information estimation unit 104 as the analysis result.

As a result, the browsing-oriented information estimation unit 104 calculates the score of the browsing-oriented information of each utterance information by using the feature information supplied from the data analysis unit 102 and the model read from the model building unit 103.
Then, the emphasized data extraction unit 106 emphasizes presentation information (optimized presentation information) of each utterance information according to the score of each utterance information, or the combination of the score and various memory information (utterance, attribute, action). Ask for. Here, the various storage information refers to the user stored in each of the utterance log table, the user attribute table, and the user action table stored in each of the utterance storage unit 108, the attribute storage unit 109, and the action storage unit 110. Information.

That is, the emphasis data extraction unit 106 may add information regarding the preference of each user to the display to the emphasis presentation information.
The emphasis data extraction unit 106 transmits a combination of utterance information and emphasis presentation information to the user's terminal device 20.
As a result, the terminal device 20 displays the utterance information on the display screen in response to the highlighted presentation information, as shown in the example of FIG.

Step S7:
The browsing-oriented information estimation unit 104 determines whether or not the calculation of the score of the emphasized presentation information for all the utterance information is completed. At this time, the browsing-oriented information estimation unit 104 ends the process when the calculation of the score of the emphasized presentation information for all the utterance information is completed. On the other hand, if the browsing-oriented information estimation unit 104 has not completed the calculation of the score of the emphasized presentation information for all the utterance information, the process proceeds to step S9.
Here, the model building unit 103 corresponds to the information newly written in the utterance log table of the utterance storage unit 108 and the user action table of the action storage unit 110, and updates the model for each user and the model for each similar subgroup. Perform processing.

Step S8:
The acquisition control unit 101 writes and stores the acquired user action information in the user action table of the action storage unit 110, and advances the process to step S2.

Step S9:
The acquisition control unit 101 writes and stores the acquired utterance information in the utterance log table of the utterance storage unit 108, and advances the process to step S2.

As described above, according to the present embodiment, the display of the terminal device 20 in which the user himself / herself is presumed to be “interested” or “related” to himself / herself (the score of the browsing-oriented information is higher than the threshold value). Since the font of the message on the screen and the image of the balloon are highlighted in comparison with the image of other messages, the message is relevant from the messages of other users other than the user himself, that is, the message is highly intuition-oriented. Can be easily identified intuitively.
Further, according to the present embodiment, it is presumed that the user himself / herself is "not interested" or "not related" to the user himself / herself (the score of the browsing-oriented information is less than the threshold value), and the message on the display screen of the terminal device 20 is displayed. In order to make the image of the font or balloon less noticeable than the image of the message that the user himself is "interested" or "related" to himself, the message of the user other than the user himself. Among them, it is possible to make it difficult to intuitively discriminate messages that are not related, that is, that the user has a low viewing intention. As a result, the display of the message that the user himself / herself is presumed to be “interested” or “related” to the user himself / herself is emphasized.
Further, according to the present embodiment, it is possible to display a message on the display screen of the terminal device 20 by reflecting the interest of the user and the relationship between the user and other users of the members in the dialogue group. By emphasizing the effect of "quickly collecting and grasping necessary information (information with a high tendency to browse)" for each user and the message of users with similar tastes, "easy to identify the user who is interacting". It can provide the effect of "transformation".

By recording a program for realizing the function of the information presentation system of FIG. 1 in the present invention on a computer-readable recording medium, and having the computer system read and execute the program recorded on the recording medium, the program is executed. You may perform a process of highlighting and presenting the spoken information on the display screen of the terminal device. The term "computer system" as used herein includes hardware such as an OS and peripheral devices.
Further, the "computer system" shall also include a WWW system provided with a homepage providing environment (or display environment). Further, the "computer-readable recording medium" refers to a portable medium such as a flexible disk, a magneto-optical disk, a ROM, or a CD-ROM, and a storage device such as a hard disk built in a computer system. Furthermore, a "computer-readable recording medium" is a volatile memory (RAM) inside a computer system that serves as a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. In addition, it shall include those that hold the program for a certain period of time.

Further, the program may be transmitted from a computer system in which this program is stored 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 above program may be for realizing a part of the above-mentioned functions. Further, a so-called difference file (difference program) may be used, which can realize the above-mentioned function in combination with a program already recorded in the computer system.

100 ... Dialogue system 10 ... Information presentation system 20 ... Terminal device 30 ... Communication network 100 ... Dialogue system 101 ... Acquisition control unit 102 ... Data analysis unit 103 ... Model construction unit 104 ... Browsing-oriented information estimation unit 106 ... Emphasis data extraction unit 107 … Data display control unit 108… Speaking storage unit 109… Attribute storage unit 110… Action storage unit

Claims (10)

When a plurality of users belonging to a predetermined dialogue group use each terminal and have a dialogue between each user via a network, display of dialogue data that each of the users wants to refer to from the history of dialogue data. It is an information presentation system that controls
An utterance storage unit that stores dialogue data of each user belonging to the dialogue group, and
Browsing-oriented information estimation that estimates browsing-oriented information, which is a score indicating the degree of interest / association of each user, from the contents of each dialogue data of the users for each dialogue data by a model corresponding to each of the users. Department and
The user obtains the score in accordance with the characteristic information of at least the utterance information indicating the utterance in the dialogue data of the user and the behavior information indicating the operation of the terminal of the user in the dialogue data. A model construction unit generated by machine learning for each,
A highlight data extraction unit that corresponds to the viewing-oriented information of each of the dialogue data of the user for each user and determines the highlight information indicating the display form of the dialogue data.
An information presentation system comprising a data display control unit for displaying each of the dialogue data in the display form indicated by the highlighting information on the display screen of the terminal. The model building unit
A similar subgroup is generated from the attribute information of the user in the dialogue group, and the model for the similar subgroup is generated.
The browsing-oriented information estimation unit
The model of the similar subgroup corresponding to the attribute information of the user is used for estimating the score of the user who newly joins the dialogue group.
The information presentation system according to claim 1. The highlighting data extraction unit determines the highlighting information so as to present it in a display form in which the user's perception becomes higher as compared with the dialogue data having the lower browsing-oriented information as the browsing-oriented information becomes higher. The information presentation system according to claim 1 or 2 . Each user further has a data analysis unit that performs natural language processing of dialogue data stored as a history and extracts feature information of the dialogue data from at least the meaning and arrangement of words.
The present invention according to any one of claims 1 to 3, wherein the browsing-oriented information estimation unit obtains the browsing-oriented information of each of the dialogue data for each user in response to the feature information. The information presentation system described. When the browsing-oriented information estimation unit estimates the browsing-oriented information, not only the dialogue data but also at least the number of responses to the dialogue data of other users in the same dialogue group as the user, the dialogue in the response. A claim characterized in that the browsing-oriented information is estimated using behavioral information including any one or a combination of the number of characters in the message, the time until the response, and the presence / absence of expansion of dialogue data of other users. The information presentation system according to any one of items 1 to 4 . When the browsing-oriented information estimation unit estimates the browsing-oriented information, not only the dialogue data but also at least any of the whereabouts, gender, age, and occupation of other users in the same dialogue group as the user. The information presentation system according to any one of claims 1 to 5 , wherein the browsing-oriented information is estimated by using the attribute information including the combination. The data display control unit is characterized in that, as a display form for emphasizing the dialogue data, the dialogue data is displayed by any one or a combination of compression (reduction), expansion, font, font size, display color, and blinking. The information presentation system according to any one of claims 1 to 6 . The data display control unit is characterized in that, as a display form for emphasizing the dialogue data, content other than the text of each text of the voice, the still image, and the moving image attached to the dialogue data is presented. The information presentation system according to any one of claims 7 . When a plurality of users belonging to a predetermined dialogue group use each terminal and have a dialogue between the respective users via a network, the display of the dialogue message that each of the users wants to refer to from the history of the dialogue message. It is an operation method of the information presentation system that controls.
A score of the degree to which the browsing-oriented information estimation unit indicates the interest / relationship of each user from the content of each dialogue data of the user in the speech storage unit in which the dialogue data of each user belonging to the dialogue group is stored. The browsing-oriented information estimation process, which estimates the browsing-oriented information for each dialogue data by the model corresponding to each of the users, and the browsing-oriented information estimation process.
The model building unit obtains the score corresponding to at least the characteristic information of the utterance information indicating the utterance in the dialogue data of the user and the behavior information indicating the operation of the terminal of the user in the dialogue data. A model construction process in which the model is generated by machine learning for each user,
A highlight data extraction process in which the highlight data extraction unit corresponds to the viewing-oriented information of each of the dialogue data of the user for each user and determines the highlight information indicating the display form of the dialogue data.
An information presentation method comprising a data display control process in which a data display control unit displays each of the dialogue data in the display form indicated by the highlighting information on the display screen of the terminal. When a plurality of users belonging to a predetermined dialogue group use each terminal and have a dialogue between the respective users via a network, the display of the dialogue message that each of the users wants to refer to from the history of the dialogue message. A program that operates a computer as an information presentation system that controls
The computer
From the content of each dialogue data of the user in the utterance storage unit in which the dialogue data of each user belonging to the dialogue group is stored, the browsing-oriented information which is a score indicating the interest / relationship of each user is obtained. A browsing-oriented information estimation means that estimates for each dialogue data by a model corresponding to each of the users .
The user obtains the score in accordance with the characteristic information of at least the utterance information indicating the utterance in the dialogue data of the user and the behavior information indicating the operation of the terminal of the user in the dialogue data. Model building means generated by machine learning for each,
A highlight data extraction means for each user, which corresponds to the browsing-oriented information of each of the dialogue data of the user and determines the highlight information indicating the display form of the dialogue data.
A program for functioning as a data display control means for displaying each of the dialogue data in the display form indicated by the highlighting information on the display screen of the terminal.

Images