JP2014109998A - Interactive apparatus and computer interactive method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique for continuing attractive interaction with users.SOLUTION: An interactive apparatus 100 comprises: a topic extraction unit 31 for executing first processing for extracting one utterance content data as a topic word, based on the number of pieces of utterance content data indicating identical contents from an interaction record table 34 in which utterance content data of each user is individually stored; a message generation unit 32 for generating topic presentation message data including the topic word extracted by the topic extraction unit 31; and a data output unit 33 for outputting the topic presentation message data generated by the message generation unit 32.

Description

  The present invention relates to interactive computer interface technology.

  Voice interactive computer interfaces are beginning to be offered. In this interface, the user's utterance is recognized as a voice, processing corresponding to the utterance is executed, and the processing result is returned from the computer to the user. As a result, the user can operate the computer in a natural interactive manner.

  The following Patent Document 1 proposes a chat system using the Internet. This system recognizes the age, sex, hobbies, preferences, etc. of each user by registering the remark text acquired from the user terminal for each user and searching the registered remark text. Further, Patent Document 1 below provides a specific information to a person having a specific hobby and preference, and searches for a user who is interested in a movie of a specific genre using the user DB. We propose to include topics related to new movies of the genre in conversations with users who have been confirmed by search.

JP 2004-145541 A

  According to the above-described proposed method, it is considered that a specific topic related to a specific hobby or a specific preference is provided to a user who is determined to have a specific hobby or a specific preference. Specifically, a topic related to a new movie of the genre can be provided to a user who is determined to be interested in a movie of a specific genre.

  However, since there is a possibility that a specific hobby or a specific preference cannot be determined for each user only by the user's speech history, the above-described topic is provided to all users in the above-described proposed method. Not necessarily. This is considered to be because the above-described proposal method is only intended to provide such topics as part of the promotion activities of products and services.

  The present invention has been made in view of such circumstances, and provides a technique for continuing an attractive dialogue with a user.

  Each aspect of the present invention employs the following configurations in order to solve the above-described problems.

  The first aspect relates to an interactive apparatus. The dialogue apparatus according to the first aspect extracts one statement content data as a topic word based on the number of statement content data indicating the same content from a dialogue record table storing each statement content data of each user. A topic extraction unit that executes processing, a message generation unit that generates topic distribution message data including topic words extracted by the topic extraction unit, and a data output unit that outputs topic distribution message data generated by the message generation unit, Have.

  The second aspect relates to a computer interaction method executed by at least one computer. In the computer interaction method according to the second aspect, one message content data is extracted as a topic word based on the number of message content data indicating the same content from a dialog record table storing each user's message content data. Executing one process, generating topic message message data including the topic word extracted by executing the first process, and outputting the generated topic message message data.

  Another aspect of the present invention may be a program that causes at least one computer to realize the configuration of the first aspect, or a computer-readable recording medium that records such a program. May be. This recording medium includes a non-transitory tangible medium.

  According to each aspect described above, it is possible to provide a technique for continuing an attractive dialogue with a user.

It is a figure which shows notionally the process structural example of the dialogue apparatus which concerns on embodiment of this invention. It is a conceptual diagram which shows the structural example of the speech dialogue server apparatus in 1st Embodiment. It is a figure which shows notionally the process structural example of the speech dialogue server apparatus in 1st Embodiment, and a client apparatus. It is a figure which shows the example of a dialogue recording table. It is a figure which shows the example of a preference category table. It is a flowchart which shows the operation example of the speech dialogue server apparatus regarding the production | generation process of a topic swing message. It is a flowchart which shows the operation example of the speech dialogue server apparatus regarding the production | generation process of a topic swing message when a topic word is acquired by 1st process. It is a flowchart which shows the operation example of the speech dialogue server apparatus regarding the production | generation process of a topic swing message when a topic word is acquired by 2nd process. It is a flowchart which shows the operation example of the speech dialogue server apparatus regarding the update of a preference category table. It is a figure which shows notionally the process structural example of the speech dialogue server apparatus in 2nd Embodiment, and a client apparatus. It is a flowchart which shows the operation example of the speech dialogue server apparatus in 2nd Embodiment regarding the production | generation process of a topic swing message. It is a figure which shows notionally the process structural example of the speech dialogue server apparatus in a modification, a client apparatus, and a dialogue recording apparatus. It is a figure which shows the example of the dialogue recording table in a modification.

  Embodiments of the present invention will be described below. In addition, embodiment mentioned below is an illustration and this invention is not limited to the structure of the following embodiment.

  FIG. 1 is a diagram conceptually illustrating a processing configuration example of an interactive apparatus 100 according to an embodiment of the present invention. As shown in FIG. 1, the dialogue apparatus 100 converts one statement content data into a topic word based on the number of statement content data indicating the same content from the dialogue record table 34 that stores each statement content data of each user. The topic extraction unit 31 that executes the first processing extracted as the message, the message generation unit 32 that generates the topic message data including the topic word extracted by the topic extraction unit 31, and the topic generation generated by the message generation unit 32 A data output unit 33 for outputting message data.

  The dialogue apparatus 100 has, for example, the same hardware configuration as that of the voice dialogue server apparatus 1 in the detailed embodiment to be described later. Is realized. The dialogue record table 34 may be realized by the dialogue apparatus 100 or may be realized by another computer.

  The computer interaction method according to the embodiment of the present invention includes the following steps executed by at least one computer, such as the interaction device 100 shown in FIG. That is, according to the computer interaction method, a first message content data is extracted as a topic word based on the number of message content data indicating the same content from a dialog record table 34 storing each user's message content data. Executing the process, generating topic message data including the topic word extracted by executing the first process, and outputting the generated topic message message data.

  As described above, the content data of each user is stored in the dialogue record table 34. The data stored in the dialogue record table 34 corresponds to the speech history of each user. Here, the utterance content data is at least a part of the utterance text data of the user, and is, for example, a noun or a noun phrase indicating the utterance content of the user. The utterance text data may be sentence data obtained by recording and recognizing the utterance voice uttered by the user, or by the user or another person using an input device such as a keyboard or a touch panel. It may be input text data.

  In the present embodiment, one message content data is extracted from the dialog record table 34 based on the number of message content data indicating the same content stored in the dialog record table 34. Therefore, the extracted utterance content data is a word or a word string indicating a topic or a trend that appears in common during the utterances of a plurality of users.

  In this embodiment, topic message data including the extracted message content data as a topic word is generated, and the topic message data is output. Topic message data indicates a message for the conversation device 100 to talk to the target user, and the speech content data extracted as described above is used as the topic word.

  Therefore, according to the present embodiment, it is possible to output a topic message based on recent topics and trends. For example, “Tokyo Skytree” is extracted as a topic word, and a topic message such as “By the way, Tokyo Skytree looks like a topic” is output. However, the output form of topic message data is not limited. For example, topic message message data is sent to a terminal used by the user and is output as a voice from the terminal. Further, the topic message message data may be displayed on a display device included in the user terminal, or may be printed on a printing device. Further, the topic message message data may be output as voice or display from the interactive device 100.

  As described above, according to the present embodiment, it is possible to make each user a hot topic or trend related to a trend that appears during the speech of a plurality of users, and thus it is possible to continue an attractive dialogue with the user. .

  Hereinafter, the details of the above-described embodiment will be described. Below, 1st Embodiment and 2nd Embodiment are illustrated as detailed embodiment. Each of the following detailed embodiments is an example in the case where the above-described dialogue apparatus 100 and the computer dialogue method are applied to a cloud-type voice dialogue server apparatus. However, the dialogue apparatus 100 and the computer dialogue method described above are not limited to application to a voice dialogue system that uses voice as an input / output form, and can be applied to apparatuses or systems having various input / output forms.

[First Embodiment]
〔Device configuration〕
FIG. 2 is a conceptual diagram illustrating a configuration example of the voice conversation server device (hereinafter simply referred to as a conversation server) 1 in the first embodiment. The dialogue server 1 is communicably connected to the client device 2 used by the user and the information providing site 5 via the communication network 3. The communication network 3 is a public network such as the Internet, a WAN (Wide Area Network), a LAN (Local Area Network), a wireless communication network, or the like. In the present embodiment, the communication mode among the dialogue server 1, the client device 2, and the information providing site 5 is not limited.

  The information providing site 5 is a general computer, and is a server device that provides various types of information such as a WEB server and a database server. The present embodiment does not limit the hardware configuration and software configuration of the information providing site 5.

  The dialog server 1 performs a dialog with the user through the client device 2. Specifically, the dialogue server 1 acquires the user's comment text data from the client device 2 and performs processing based on the comment text data. Further, the dialogue server 1 sends the topic message message data to the client device 2 and outputs the topic message from the client device 2.

  As shown in FIG. 2, the dialogue server 1 includes a CPU (Central Processing Unit) 10, a memory 11, an input / output interface (I / F) 12, and a communication device that are mutually connected via a bus 15 as a hardware configuration. 13 etc. The memory 11 is a random access memory (RAM), a read only memory (ROM), a hard disk, a portable storage medium, or the like. The input / output I / F 12 is connected to a device that accepts an input of a user operation such as a keyboard and a mouse, and a device that provides information to the user such as a display device and a printer. The communication device 13 communicates with other nodes via the communication network 3. Note that the hardware configuration of the dialog server 1 is not limited.

  The client device 2 is a general user terminal having a user interface that enables user input and output to the user, such as a PC (Personal Computer), a tablet terminal, a smartphone, a mobile phone, and a car navigation device. The client device 2 may be a robot having a communication function with a human. Hereinafter, the client device 2 is simply referred to as a client 2. As shown in FIG. 2, the client 2 includes a CPU 10, a memory 11, an input / output I / F 12, a communication device 13, and the like that are connected to each other via a bus 15. The input / output I / F 12 includes an input device (not shown) that receives input of a user operation such as a keyboard and a mouse, a voice input unit 17 such as a microphone that acquires a user's speech, a speaker that outputs voice, and the like And an output device (not shown) for providing information to the user such as a display device and a printer. Note that the hardware configuration of the client 2 is not limited.

[Processing configuration]
FIG. 3 is a diagram conceptually illustrating a processing configuration example of the dialogue server 1 and the client 2 in the first embodiment. In FIG. 3, the same reference numerals as those in FIG. 1 are attached to the same processing units as those of the interactive apparatus 100 shown in FIG.

<Client device>
The client 2 includes a voice recognition unit 21, a voice synthesis unit 22, and the like. Each of these processing units is realized, for example, by executing a program stored in the memory 11 by the CPU 10. In addition, the program is installed from a portable recording medium such as a CD (Compact Disc) or a memory card or another computer on the network via the input / output I / F 12 or the communication device 13 and is stored in the memory 11. It may be stored.

  The speech recognition unit 21 generates speech text data corresponding to the speech speech by executing speech recognition processing on the speech speech of the user input and recorded through the speech input unit 17. The message text data is sent to the dialogue server 1 together with the sender identification information (denoted as sender ID). The transmission source ID may be information for identifying the client device 2, information for identifying the user, or both of them. The comment text data is character data in which a voice uttered by the user is converted into text. In the present embodiment, a known technique may be used for the voice recognition process, and the voice recognition process itself and various voice recognition parameters used in the voice recognition process are not limited.

  The voice synthesizer 22 performs voice synthesis processing on topic message data sent from the dialogue server 1 and outputs the voice of the message via the voice output unit 18. The voice output from the voice output unit 18 is, for example, a voice that reads a message included in the topic message data. In the present embodiment, a known method may be used for the speech synthesis process, and the speech synthesis process itself is not limited.

<Voice dialogue server device>
As shown in FIG. 3, the dialogue server 1 includes a dialogue control engine 25, a WEB server processing unit 26, and the like. Each of these processing units is realized, for example, by executing a program stored in the memory 11 by the CPU 10. The program may be installed from a portable recording medium such as a CD or a memory card or another computer on the network via the input / output I / F 12 or the communication device 13 and stored in the memory 11. Good.

  The WEB server processing unit 26 realizes a general WEB server function. For example, the WEB server processing unit 26 enables bidirectional communication between the client 2 and the voice conversation server device 1, execution of a servlet on the WEB server, and the like. The present embodiment does not limit the specific processing of the WEB server processing unit 26.

  The dialog control engine 25 is a processing unit that realizes a dialog with the user through the client 2. As shown in FIG. 3, the dialogue control engine 25 includes a topic extraction unit 31, a message generation unit 32, a data output unit 33, a dialogue record table 34, a speech data acquisition unit 35, a category determination unit 36, a learning processing unit 37, A preference category table 38, a storage processing unit 39, and the like are included. Each of these processing units is realized, for example, by executing a program stored in the memory 11 by the CPU 10. Note that the topic extraction unit 31, the message generation unit 32, the data output unit 33, and the dialogue recording table 34 are the same as the processing units of the dialogue device 100 shown in FIG. The description will focus on the content different from each processing unit.

  Further, the dialogue control engine 25 determines the topic generation message data generation timing and the transmission destination of the topic distribution message data. For example, the dialogue control engine 25 starts generating topic message data when generating a dialogue message that is not expected to be spoken by the user, unlike a question or confirmation, such as a greeting or a dialogue message that provides certain information. To do. Along with the determination of the start, the dialogue control engine 25 causes the topic extraction unit 31, the message generation unit 32, and the data output unit 33 to operate. Further, the dialogue control engine 25 determines the destination of the dialogue message as described above as the destination of the topic message data. Hereinafter, the user (transmission source ID) determined as the transmission destination of the topic message data may be referred to as a target user. Note that the present embodiment does not limit the generation and transmission timing of topic message data.

  The utterance data acquisition unit 35 acquires the user's utterance text data and the transmission source ID transmitted from the client 2 via the WEB server processing unit 26. In the user's remark text data acquired by the remark data acquisition unit 35, a word that affirms a topic that has been shaken such as “I want to hear”, “what it is” (hereinafter, referred to as a topic affirmative word), or Response utterance text data including words (hereinafter referred to as topic negation words) that deny a topic that has been shaken such as “don't listen” and “don't need” are included.

  Note that the comment text data to be processed by the dialogue control engine 25 in the present embodiment may be limited. For example, the conversation server 1 processes the utterance text data that is not handled in the process described in this embodiment by another process.

  The category determination unit 36 determines the category of the comment text data acquired by the comment data acquisition unit 35. For example, the category determination unit 36 determines a category based on the keyword included in the comment text data by referring to a table holding a plurality of correspondence relationships between the keyword and the category. The category determination unit 36 may perform part-of-speech decomposition processing on the comment text data in order to extract keywords. Note that this embodiment does not limit the category determination method itself.

  The storage processing unit 39 stores at least one part of the utterance text data acquired by the utterance data acquisition unit 35 and the category determined by the category determination unit 36 in the dialogue recording table 34 together with the transmission source ID and the recording time. To do. At least a part of the comment text data stored in the dialogue record table 34 is expressed as comment content data. For example, when the comment text data “Tell me about today's Ichiro” is acquired, “Today's Ichiro” is stored in the dialog record table 34 as the statement content data.

  In addition, when the storage processing unit 39 transmits any dialogue message data to the user's utterance text data acquired by the utterance data acquisition unit 35, the storage processing unit 39 interactively displays response content data indicating the content of the dialogue message data. Store in the recording table 34. At this time, the storage processing unit 39 stores the response content data in a record including the corresponding message content data.

  FIG. 4 is a diagram illustrating an example of the dialogue record table 34. The dialogue record table 34 stores a user's speech history. As shown in FIG. 4, the dialogue recording table 34 in the present embodiment stores a transmission source ID, a recording time, a category, message content data, and response content data for each message text data. The sender ID is data acquired by the message data acquisition unit 35 together with the message text data, and the recording time indicates the time when the target record is stored. The message content data means data stored in the dialog record table 34 in each message text data. The response content data means data stored in the dialogue record table 34 in the dialogue message data generated by the dialogue server 1 for the message text data and output from the client device 2.

  FIG. 5 is a diagram illustrating an example of the preference category table 38. The preference category table 38 stores, for each user, the degree of preference for each category corresponding to the content of each user's speech. The transmission source ID shown in FIG. 5 is data acquired by the message data acquisition unit 35 together with the message text data of the category. The preference level indicates the degree to which the user corresponding to the transmission source ID likes the corresponding category or is interested in the corresponding category. In the present embodiment, the preference level is indicated by a numerical value of 1 or more, and the magnitude of the numerical value indicates the height of the preference level.

  The learning processing unit 37 adds a record to the preference category table 38 based on the user's comment text data, and further updates the preference level of the preference category table 38. Specifically, the learning processing unit 37 includes the category when the category determined by the category determining unit 36 regarding the comment text data acquired by the comment data acquiring unit 35 is not stored in the preference category table 38. The record is added to the preference category table 38. At this time, the learning processing unit 37 adds, to the record to be added, the transmission source ID acquired by the utterance data acquisition unit 35, the category determined by the category determination unit 36, and the utterance acquired by the utterance data acquisition unit 35. The preference level corresponding to the contents of the text data is included. For example, when the comment text data indicates a question message about the category, the learning processing unit 37 determines that the preference level is 2 because the user is considered interested in the category.

  In addition, the learning processing unit 37 acquires the user's response message text data in the message data acquisition unit 35 for the topic message output from the client device 2 based on the topic message message data transmitted from the dialogue server 1. In this case, the target record is added to the preference category table 38 or the preference level of the target record in the preference category table 38 is increased or decreased according to the response content of the response message text data. Specifically, the learning processing unit 37 updates the preference level of the record that matches the transmission source ID corresponding to the response message text data and the category related to the topic message. For example, when the response message text data indicates an affirmative response, it is considered that the user has shown interest in the topic word category of the topic message, so the preference level of the record is incremented by one. On the other hand, when the response message text data indicates a negative response, it is considered that the user did not show interest in the category of the topic word of the topic message, so the preference level of the record is decremented by 1.

  The topic extraction unit 31 acquires the topic word by executing the first process described above or the second process described below in the interactive apparatus 100. As described above, the first process determines a topic word indicating a hot topic or trend based on the remarks of a plurality of users. A 2nd process determines the topic word along with a user's preference. In the second process, the topic extraction unit 31 specifies one category from the target user categories stored in the preference category table 38 based on the preference level, and this specified in the dialog record table 34 is specified. Based on the number of records including statement content data indicating the same content of the category, one message content data of the category is extracted as a topic word from the dialogue record table 34.

  The topic extraction unit 31 selectively executes the first process and the second process so that the first process and the second process are executed at a predetermined rate. This predetermined ratio is held in advance by the topic extraction unit 31. For example, the execution ratio of the first process is set to 15% and the execution ratio of the second process is set to 85% as the predetermined ratio. This predetermined ratio may be set to a value preferred by the user, and the value is not particularly limited. The predetermined ratio may be set for each user. As described above, it is desirable that the first process and the second process are selectively executed, but this embodiment does not exclude a form in which only one of the first process and the second process is executed.

  Further, in the first process, the topic extraction unit 31 has each statement content data set indicating the same content in the dialogue record table 34 corresponding to each rank of the number of statement content data in each statement content data set. The speech content data may be selectively acquired as the topic word from the dialogue recording table 34 so that each is selected at a predetermined ratio. The topic extraction unit 31 holds a predetermined selection ratio in advance for each rank of the number of pieces of speech content data in each speech content data set. For example, the probability that the speech content data set having the largest number of speech content data is selected is 50%, the probability that the next largest speech content data set is selected is 25%, and the probability that the next largest speech content data set is selected. 12.5%, the selection ratio (selection probability) of each rank is held. However, the topic extraction unit 31 may always select a speech content data set having the largest number of speech content data without using such a selection ratio.

  Further, the topic extraction unit 31 may selectively specify a category so that the category is specified from the preference category table 38 at a selection ratio corresponding to each rank of the degree of preference in the second process. . In this case, the topic extraction unit 31 holds in advance a predetermined selection ratio for each rank of the preference level. For example, the selection ratio (selection probability) of each rank is held such that the probability that the category with the highest degree of preference is specified is 50%, and the probability that the category with the next highest degree of preference is specified is 25%. . However, the topic extraction unit 31 may always specify the category having the highest degree of preference without using such a selection ratio.

  Further, in the second process, the topic extraction unit 31 converts each statement content data set indicating the same content in the statement content data of the specified category stored in the dialogue record table 34 into each statement content data. One message content data of the specified category is selectively acquired as a topic word from the dialogue record table 34 so that each message is selected at a predetermined ratio corresponding to each rank of the message content data number of the set. May be. This predetermined ratio is the same as described above.

  The message generation unit 32 generates topic distribution message data including the topic word acquired by the topic extraction unit 31.

  The data output unit 33 transmits the topic message message data generated by the message generation unit 32 to the client 2 of the target user.

[Operation example]
First, a computer interaction method according to the first embodiment will be described with reference to FIGS. 6A, 6B, and 6C. FIGS. 6A, 6B, and 6C are flowcharts showing an example of the operation of the dialogue server 1 related to the process of generating the topic message.

  When the conversation server 1 decides to transmit the topic message to the target user (transmission source ID), it reads information on the topic affirmative word and the topic negative word (S61). Information about the topic affirmative word and the topic negative word is held in advance by the dialogue server 1, for example.

  The dialogue server 1 generates a random number between 1 and 100 (S62). The dialogue server 1 determines whether or not the value of the random number is 15 or less (S63). Each process of (S62) and (S63) is a predetermined ratio of the first process and the second process (the execution ratio of the first process is 15% and the execution ratio of the second process is 85%). It is a process for executing each.

  When the value of the random number is 15 or less (S63; YES), the dialogue server 1 executes the process shown in FIG. 6B, that is, the first process, and when the value of the random number is larger than 15 (S63; NO), The process shown in 6C, that is, the second process is executed.

  FIG. 6B is a flowchart illustrating an operation example of the dialogue server 1 related to the topic exchanging message generation processing when the topic word is acquired by the first processing. The dialogue server 1 extracts from the dialogue record table 34 the top 10 record sets of the number of records among the record sets whose statement content data indicates the same content (S71). Here, there is a possibility that the content data of a plurality of users is included in the extracted record set.

  Subsequently, the dialogue server 1 generates a random number, and specifies a record set in the order of the number of records corresponding to the random number from the plurality of record sets extracted in (S71) (S72). By specifying using this random number, for example, the probability that the record set with the maximum number of records is specified is 50%, and the probability that the record set with the next largest number of records is specified is 25%. , A record set is selectively identified.

  The dialogue server 1 determines whether or not the category of the record set specified in (S72) is tourism information (S73). When the category is tourism information (S73; YES), the dialogue server 1 extracts response content data from the record set (S75). On the other hand, when the category is not tourism information (S73; NO), the dialogue server 1 requests the information providing site 5 that provides news information by using the statement content data of the record set as a search keyword, and performs the search. The resulting news information is acquired (S74). Since the tourist information has a smaller degree of change with time than other information, the previous response content data is extracted as shown in (S75). On the other hand, the latest news information is acquired from the information providing site 5 as shown in (S74) for categories other than tourism information.

  The dialogue server 1 acquires the statement content data of the specified record set as a topic word (S76). The dialogue server 1 generates topic message data including the topic word, and transmits the topic message data to the client device 2 (S77).

  As a result, the topic message is output from the client device 2 by voice. For example, when the topic word “Tokyo Skytree (registered trademark)” included in the remarks of many users is acquired (S76), the topic message “Tokyo Skytree looks like a topic” is voiced. Is output. The user makes a response utterance including a topic affirmative word such as “I want to hear” or a topic negative word such as “I do not want to hear” for the topic message. The response utterance is converted into response utterance text data by being recorded and voice-recognized by the client device 2.

  The dialogue server 1 acquires the response message text data from the client device 2 (S78). The dialogue server 1 determines whether the topic affirmative word or the topic negative word read in (S61) is included in the response utterance text data (S79). When the response utterance text data does not include a topic affirmative word or a topic negative word (S79; NO), the dialogue server 1 determines that the topic has changed, and ends the process shown in FIG. 6B.

  Furthermore, the dialogue server 1 determines whether or not a topic affirmative word is included in the response message text data (S79; YES, S80). When the response server text data includes an affirmative word (S79; YES, S80; YES), the dialog server 1 receives the news information data acquired in (S74) or the response content data acquired in (S75). Is generated, and the dialog message data is transmitted to the client apparatus 2 (S82). As a result, a dialogue message having the same content as the response content in the previous dialogue stored in the dialogue record table 34 is transmitted to the client 2, and the response content is output by voice in the client 2.

  At this time, the dialogue server 1 adds 1 to the preference level of the corresponding record in the preference category table 38 (S83). This is because the user is interested in the topic message output in (S77). On the other hand, when the response utterance text data does not include the topic affirmative word (S79; YES, S80; NO), the dialogue server 1 decrements the preference level of the corresponding record in the preference category table 38 by 1 (S81). This is because the user did not show interest in the topic message output in (S77). Details of the update process of the preference category table 38 will be described later.

  FIG. 6C is a flowchart illustrating an operation example of the dialogue server 1 related to the processing of generating a topic message when the topic word is acquired by the second processing. The dialogue server 1 extracts the top 10 records of the preference level of the target user from the preference category table 38 (S91). Here, the target user is a user determined as a transmission destination of the topic message data, and is identified by a transmission source ID in the present embodiment.

  The dialogue server 1 generates a random number, and specifies the category of the record of the preference ranking corresponding to the random number among the plurality of records extracted in (S91) (S92). By specifying using this random number, for example, the record at a predetermined ratio for each preference rank, such as the probability that the record with the highest degree of preference is specified is 50%, and the probability that the record with the next highest degree of preference is specified is 25%. Are selectively identified.

  Subsequently, the dialogue server 1 extracts, from the dialogue record table 34, the top 10 record sets of the number of records in each record set in which the statement content data of the category specified in (S92) indicates the same content. (S93). Here, as in the case of FIG. 6B, there is a possibility that the content data of a plurality of users is included in the extracted record set.

  Subsequently, the dialogue server 1 generates a random number, and specifies a record set in the order of the number of records corresponding to the random number from the plurality of record sets extracted in (S93) (S94). Since each process after (S94) is the same as each process after (S72) shown by FIG. 6B, description is abbreviate | omitted here.

  According to FIG. 6C, for example, a category “professional baseball player” having a high preference for the target user is identified (S92), and the topic word “Ichiro” included in the remarks of many users regarding this category “professional baseball player”. "Is acquired (S98), a topical message saying" Ichiro seems to be a topic "is output by voice.

  The computer interaction method in the present embodiment is not limited to the examples shown in FIGS. 6A, 6B, and 6C. In the above example, the number of record sets extracted from the dialogue record table 34 is set to 10 (S71) and (S93), and the number of records extracted from the preference category table 38 is also set to 10 ( S91). However, the record set and the number of records may be set to different numbers, or may be set to a number other than 10.

  Next, the learning method of the category preference degree in the first embodiment will be described with reference to FIG. FIG. 7 is a flowchart showing an operation example of the dialogue server 1 regarding the update of the preference category table 38. This learning method is implemented in (S81) and (S83) shown in FIG. 6B, (S103) and (S105) shown in FIG. 6C, and before the user's comment text data is stored in the dialogue record table 34. Or later.

  At this time, the dialogue server 1 has acquired the target transmission source ID, the target category, and the preference level update information (S111). In the scenes of (S81) and (S83) shown in FIG. 6B, the target transmission source ID is the transmission source ID specified as the transmission destination of the topic message data in (S77), and the target category is (S72 ) Is the category of the record set specified by Also, the preference level update information is -1 in the scene of (S81) and +1 in the scene of (S83). In the scenes of (S103) and (S105) shown in FIG. 6C, the target transmission source ID is the transmission source ID corresponding to the target user acquired in (S91), and the target category is specified in (S92). Category. Also, the preference level update information is −1 in the scene of (S103) and +1 in the scene of (S105). Furthermore, in a scene where the user's comment text data such as “Tell me about Ichiro” is stored in the dialogue record table 34, the sender ID of the comment text data is the target sender ID, and the category of the comment text data is the target. For example, the preference level update information is set to +2.

  The dialogue server 1 extracts a record corresponding to the target transmission source ID and the target category from the preference category table 38 (S112). When the corresponding record exists in the preference category table 38 (S113; YES), the dialogue server 1 updates the preference level of the extracted record based on the preference level update information (S117).

  Here, when the updated preference value is not 1 or more (S118; NO), the dialogue server 1 resets the updated preference value to 1 (S119). Thereby, according to this embodiment, the preference degree stored in the preference category table 38 is kept at 1 or more. This is because the purpose of this embodiment is to obtain a high degree of preference for a category. However, it is not always necessary to prevent the preference level from becoming 0 or less. In this case, in (S91) of FIG. 6C, only one or more preference levels need be extracted, and (S118), (S119), and (S114) are omitted from FIGS.

  On the other hand, when the corresponding record does not exist in the preference category table 38 (S113; NO), the dialogue server 1 further determines whether the preference level update information indicates an increasing direction (S114). When the preference level update information indicates an increasing direction (S114; YES), the dialogue server 1 adds a record including the target transmission source ID, the target category, and the preference level corresponding to the preference level update information to the preference category table 38. (S115). The preference level corresponding to the preference level update information means a preference level in which a value indicated by the preference level update information is set.

[Operation and Effect of First Embodiment]
As described above, in the first embodiment, the first process or the second process is executed in order to extract the topic word to be included in the topic message message data from the dialogue record table 34. In the first process, one utterance content data is extracted as a topic word from the utterance content data of a plurality of users stored in the dialogue recording table 34 based on the number of utterance content data indicating the same content. According to this first process, as described above, it is possible to acquire a hot topic or trending topic word that appears during the speech of a plurality of users, and in turn, a topic message including such a topic word. Can be provided to the user.

  On the other hand, in 1st Embodiment, the preference degree for every user's category is learned based on each user's comment text data. In this learning, the preference level of the category of the topic word in the topic message is stored in the preference category table 38 based on the user's response utterance to the topic message provided to the user by the dialogue server 1 via the client device 2. Reflected. Further, the preference level of the category related to the utterance is reflected in the preference category table 38 based on the user's other utterances. According to this learning, for each user, it is possible to acquire a category that has interest and interest and a degree of preference thereof.

  In addition, in the second process described above, one category for the target user is specified based on the preference level of each category learned for each user, and the specified category stored in the dialogue record table 34 is specified. One piece of comment content data is acquired as a topic word based on the number of pieces of comment content data indicating the same content from among the plurality of users' content data of the category. According to this second process, it is possible to acquire a topic word indicating a hot topic or trend for a category that the user likes, and to provide a user with a topic message including such a topic word. it can.

  Therefore, according to the first embodiment, a hot topic that does not depend on the user's own preference or a hot topic that conforms to the user's own preference can be given to the user. Will be triggered. That is, according to the first embodiment, an attractive dialogue with the user can be continued.

  Further, in the first embodiment, the first process and the second process for acquiring attractive topic words are selectively executed at a predetermined rate. Thereby, it can prevent that the same topic word set is repeatedly provided with respect to each user, and a fresh topic can be shaken for each user each time. Therefore, since the possibility that the user desires to continue the conversation increases, according to the first embodiment, it is possible to continue a more attractive conversation with the user.

  In order to prevent such monotony of the topic, in the first embodiment, each of a plurality of record sets (remark content data sets) of extraction candidates stored in the dialogue record table 34 is a record in each record set. One message content data is selectively acquired as a topic word from the dialogue recording table 34 so as to be selected at each predetermined ratio corresponding to the number rank. For the same purpose, in the first embodiment, a category is selectively specified from the preference category table 38 at each predetermined ratio corresponding to each rank of the preference level. Thereby, it can prevent that the same topic word set is repeatedly provided with respect to each user, and a fresh topic can be shaken for each user each time.

[Second Embodiment]
In the second embodiment, it is possible to schedule topic allocation timing. Hereinafter, the dialogue server 1 in the second embodiment will be described focusing on the content different from the first embodiment. In the following description, the same contents as those in the first embodiment are omitted as appropriate.

[Processing configuration]
FIG. 8 is a diagram conceptually illustrating a processing configuration example of the dialogue server 1 and the client 2 in the second embodiment. The dialogue server 1 in the second embodiment further includes a schedule processing unit 41 and a schedule table 42 in addition to the configuration of the first embodiment. The schedule processing unit 41 and the schedule table 42 are also realized by executing a program by the CPU 10 in the same manner as other processing units.

  The schedule table 42 stores timing information and a category about each user's topic swing. The information stored in the schedule table 42 may be information obtained by the user operating the client device 2 based on an input screen or the like and acquired from the client device 2, or from a portable recording medium or the like. Information acquired via the input / output I / F 12 may be used.

  In the timing information, a time and a period such as date and time may be set, or a specific message generated at the same time may be set. For example, as the timing information, 8:00 am every day, 7:00 pm on December 24, 2012, the same time as the first greeting message of each day, and the like can be set. However, the present embodiment does not limit the specific content of the timing information, and information that can identify a certain timing may be set in the timing information.

  The schedule processing unit 41 acquires, as the topic word, one piece of content data of the target user corresponding to the category stored in the schedule table 42 to the topic extraction unit 31 at the timing indicated by the timing information stored in the schedule table 42. The message generator 32 generates topic message data, and the data output unit 33 outputs topic message data.

  The topic extraction unit 31 uses the category specified by the schedule processing unit 41, and based on the number of records in the dialog recording table 34 including the statement content data indicating the same content of the specified category. From 34, the third process of specifying one piece of content data of the category as a topic word is executed. The topic extraction unit 31 executes the third process when executed based on the schedule table 42, and selects the first process and the second process as in the first embodiment otherwise. Run it.

[Operation example]
Hereinafter, a computer interaction method according to the second embodiment will be described with reference to FIGS. 6A, 6B, 6C, and 9. FIG. FIG. 9 is a flowchart showing an operation example of the dialogue server 1 according to the second embodiment, related to the topic generation message generation process. In the second embodiment, the dialogue server 1 executes the operation shown in FIG. 9 at the timing based on the schedule table 42, and at other timings, as in the first embodiment, FIG. 6A, FIG. 6B, and FIG. The operation indicated by 6C is performed.

  At the timing based on the schedule table 42, the category specified from the schedule table 42 is acquired. In this case, the dialog server 1 extracts from the dialog record table 34 the top 10 record sets of the number of records in each record set in which the content data of the designated category indicates the same content (S121). . Here, there is a possibility that the content data of a plurality of users is included in the extracted record set.

  Since each process after (S122) is the same as each process after (S72) shown by FIG. 6B, description is abbreviate | omitted here.

[Operation and Effect of Second Embodiment]
As described above, in the second embodiment, topic allocation timing information is stored in the schedule table 42 for each user, and topic allocation message data is generated and output at the timing indicated by the timing information. Further, the schedule table 42 stores the category of the topic to be shaken along with the timing information, and the topic content message data generated at the specified timing includes the message content data of the category specified in the schedule table 42. Included.

  Therefore, according to the second embodiment, it is possible to provide a topical message including an attractive topic word suitable for the timing at a timing and category according to the user's age and life style as well as the preference. .

[Modification]
In each of the above-described embodiments, an example has been shown in which the dialogue server 1 acquires news information from the information providing site 5 connected via the communication network 3. However, the news information may be acquired from a database held in the dialogue server 1. In this case, the dialogue server 1 may determine the acquisition destination of the news information from the information providing site 5 and the database in the dialogue server 1.

  Further, in each of the above-described embodiments, the example in which the utterance text data acquired by the utterance data acquisition unit 35 is generated from the voice uttered by the user is shown. However, the comment text data may be text data input by a user operation via the input device of the client 2 such as a keyboard or a touch panel. Further, the output form of the dialogue message data is not limited to the voice output, and may be displayed on a display device provided in the client 2 or may be printed on a printing device connected to the client 2. Good. Furthermore, the dialog message data may be output in two or more output forms.

  Further, in each of the above-described embodiments, an example in which the client 2 has the voice recognition unit 21 and the voice synthesis unit 22 is shown, but the dialog server 1 may have a voice recognition unit 21 and a voice synthesis unit 22. In this case, voice data is exchanged between the dialogue server 1 and the client 2. Further, both the dialogue server 1 and the client device 2 may be realized by a single computer, and a topical message may be output from a display device or an audio output unit included in the computer itself.

  Further, in each of the above-described embodiments, as shown in FIG. 3, the dialog server 1 has all the above-described processing units. However, each of the above-described processing units is performed on a plurality of computers including the dialog server 1. It may be realized. For example, among the processing units described above, a processing unit that records user's remark text data, a processing unit that learns user's preferences, and a processing unit that generates an interactive message such as topical message data are set on different computers. You may make it prepare. In this example, a dialog recording device 70 that can communicate with the dialog server 1 is further provided.

  FIG. 10 is a diagram conceptually illustrating a processing configuration example of the dialog server 1, the client 2, and the dialog recording device 70 in the modification. In this modification, the dialog control engine 25 of the dialog server 1 includes a utterance data acquisition unit 35, a topic extraction unit 31, a message generation unit 32, and a data output unit 33. The dialog recording device 70 includes a category determination unit 36, A learning processing unit 37, a preference category table 38, a storage processing unit 39, a dialogue recording table 34, and a communication processing unit 71 are included. The communication processing unit 71 acquires message text data from the dialog server 1, extracts a record that meets the request from the dialog record table 34 and the preference category table 38 in response to the request from the dialog server 1, and stores the record in the dialog server Send to 1.

  In each of the above-described embodiments, all the utterance content data in the dialogue recording table 34 is the subject of the topic word, but the utterance content data that is the subject of the topic word may be limited. FIG. 11 is a diagram illustrating an example of the dialogue record table 34 in the modification. The dialogue record table 34 in the modified example further stores a topic flag in addition to the contents of FIG. The topic flag is a flag for discriminating whether or not the content of the speech is suitable for the topic. In the example of FIG. 4, the topic flag is set to 1 in a record of speech content data suitable for topic swing, and is set to 0 in a record of speech content data unsuitable for topic swing.

  The storage processing unit 39 in this modification example determines the topic flag corresponding to the comment text data based on the category determined by the category determination unit 36 when storing the user's comment text data in the dialogue recording table 34. . For example, the topic flag corresponding to the comment text data of the category “weather” or “stock price” is set to 0, and the topic flag corresponding to the comment text data of other categories is set to 1. This is because “weather” and “stock price” are difficult to lead to topic development and are not suitable for topic swing.

  The topic extraction unit 31 in the present modification excludes records for which the dialogue flag is set to 0 when extracting message content data as topic words from the dialogue record table 34. Similarly, also in (S71) in FIG. 6B, (S93) in FIG. 6C, and (S121) in FIG. 9, records in which the dialogue flag is set to 0 are excluded from extraction targets.

  According to such a modification, only the content of the speech that is suitable for the topic swing is extracted as the topic word among the speech of the plurality of users, so that it is possible to provide a topic swing message that is highly likely to attract the user. .

  In addition, in the some flowchart used by the above-mentioned description, although several process (process) is described in order, the execution order of the process performed by each embodiment is not restrict | limited to the order of the description. In each embodiment, the order of the illustrated steps can be changed within a range that does not hinder the contents. Moreover, each above-mentioned embodiment and each modification can be combined in the range with which the content does not conflict.

  Each of the above-described embodiments and modifications may be specified as in the following supplementary notes. However, each embodiment and each modification are not limited to the following description.

(Appendix 1)
A topic extraction unit that executes a first process of extracting one piece of comment content data as a topic word based on the number of pieces of comment content data indicating the same content from a dialogue record table that stores each piece of user content data;
A message generator for generating topic message data including topic words extracted by the topic extractor;
A data output unit that outputs topic message data generated by the message generation unit;
A dialogue apparatus comprising:

(Appendix 2)
The dialog record table stores each user's message content data and categories corresponding to the message content,
The topic extraction unit includes the first process, or
From the preference category table that stores the preference level for each category corresponding to the content of the speech of the target user by learning using the speech history of the target user, one category is identified based on the preference level, and the dialog record A second process of extracting one piece of comment content data of the specified category as the topic word from the dialog record table based on the number of pieces of comment content data indicating the same content of the specified category in the table; The interactive apparatus according to attachment 1, which is executed.

(Appendix 3)
The topic extraction unit selectively executes the first process and the second process so that the first process and the second process are executed at a predetermined rate.
The interactive apparatus according to attachment 2.

(Appendix 4)
In the second process, the topic extraction unit is configured such that each statement content data set indicating the same content in the statement content data of the specified category stored in the dialogue record table is each statement content data set. One of the specified category content data is selectively extracted as the topic word from the dialogue record table so that the content is selected at each predetermined ratio corresponding to each ranking of the number of the speech content data.
The interactive apparatus according to appendix 2 or 3.

(Appendix 5)
A message data acquisition unit that acquires user response message text data and transmission source identification data accompanying the output of the topic message data,
Based on the sender identification data acquired by the message data acquisition unit, the category related to the topic message data, and the preference update information corresponding to the response content of the response message text data acquired by the message data acquisition unit. A learning processing unit for updating the preference category table;
The interactive apparatus according to any one of appendices 2 to 4, further comprising:

(Appendix 6)
A remark data acquisition unit that acquires remark text data and transmission source identification data of the user;
A category determination unit for determining a category of the comment text data acquired by the comment data acquisition unit;
Based on the sender identification data acquired by the remark data acquisition unit, the category determined by the category determination unit, and the preference update information corresponding to the content of the remark text data acquired by the remark data acquisition unit A learning processing unit for updating the preference category table;
The interactive apparatus according to any one of appendices 2 to 5, further comprising:

(Appendix 7)
One of the target users corresponding to the category stored in the schedule table is sent to the topic extraction unit at the timing indicated by the timing information stored in the schedule table for storing the timing information and the category for each user. One speech content data is extracted as the topic word, the message generator generates the topic message data, and the data output unit outputs the topic message data.
The interactive apparatus according to any one of appendices 2 to 6, further comprising:

(Appendix 8)
In the first process, the topic extraction unit is configured so that each comment content data set indicating the same content in the dialogue record table corresponds to each rank of the number of comment content data in each comment content data set. Extracting the content data of the speech selectively as the topic word from the dialogue record table so that each of them is selected in proportion;
The interactive apparatus according to any one of appendices 1 to 7.

(Appendix 9)
The dialogue record table stores a topic flag indicating whether or not the content of the speech is suitable for topicing together with the content data of each user.
The topic extraction unit excludes the comment content data corresponding to the topic flag indicating that it is not suitable for topic swing from the extraction target as the topic word,
The interactive apparatus according to any one of appendices 1 to 8.

(Appendix 10)
In a computer interaction method executed by at least one computer,
Executing a first process of extracting one piece of content data as a topic word based on the number of pieces of speech content data indicating the same content from the dialogue record table storing the content of each user.
Generating topic message data including topic words extracted by the execution of the first process;
Outputting the generated topic message message data;
A computer interaction method comprising:

(Appendix 11)
The dialog record table stores each user's message content data and categories corresponding to the message content,
From the preference category table that stores the preference level for each category corresponding to the content of the speech of the target user by learning using the speech history of the target user, one category is identified based on the preference level, and the dialog record A second process of extracting one piece of comment content data of the specified category as the topic word from the dialogue record table based on the number of pieces of statement content data indicating the same content of the specified category in the table; Selectively executing the first process and the second process so that the first process is executed at a predetermined rate;
The computer interaction method according to appendix 10, including the above.

(Appendix 12)
In the second process, the message content data sets each indicating the same content in the message content data of the specified category stored in the dialog record table is the number of message content data in each message content data set. One of the specified category content data is selectively extracted as the topic word from the dialogue record table so that the selected content ratio is selected at a predetermined ratio corresponding to each rank of
The computer interaction method according to appendix 11.

(Appendix 13)
Acquire user response message text data and sender identification data accompanying the output of the topic message data,
Updating the preference category table based on the acquired sender identification data, the category relating to the topic message data, and the preference update information corresponding to the response content of the acquired response message text data;
The computer interaction method according to appendix 11 or 12, further comprising:

(Appendix 14)
Obtain the user's message text data and sender identification data,
Determining a category of the acquired comment text data;
Updating the preference category table based on preference update information corresponding to the content of the acquired sender identification data, the determined category, and the acquired message text data;
The computer interaction method according to any one of appendices 11 to 13, further including:

(Appendix 15)
In the first process, each of the message content data sets indicating the same content in the dialog record table is selected at each predetermined ratio corresponding to each rank of the message content data numbers in each message content data set. As described above, the message content data is selectively extracted as the topic word from the dialog record table.
15. The computer interaction method according to any one of appendices 10 to 14.

(Appendix 16)
The dialogue record table stores a topic flag indicating whether or not the content of the speech is suitable for topicing together with the content data of each user.
The extraction as the topic word of the statement content data excludes the statement content data corresponding to the topic flag indicating that it is not suitable for topic swing from the extraction target as the topic word.
The computer interaction method according to any one of appendices 10 to 15.

(Appendix 17)
In a program for causing at least one computer to execute computer interaction processing,
The computer interaction process is:
Executing a first process of extracting one piece of content data as a topic word based on the number of pieces of speech content data indicating the same content from the dialogue record table storing the content of each user.
Generating topic message data including topic words extracted by the execution of the first process;
Outputting the generated topic message message data;
A program that includes that.

(Appendix 18)
The dialog record table stores each user's message content data and categories corresponding to the message content,
The computer interaction process is:
From the preference category table that stores the preference level for each category corresponding to the content of the speech of the target user by learning using the speech history of the target user, one category is identified based on the preference level, and the dialog record A second process of extracting one piece of comment content data of the specified category as the topic word from the dialogue record table based on the number of pieces of statement content data indicating the same content of the specified category in the table; Selectively executing the first process and the second process so that the first process is executed at a predetermined rate;
The program according to appendix 17, including the above.

(Appendix 19)
In the second process, the message content data sets each indicating the same content in the message content data of the specified category stored in the dialog record table is the number of message content data in each message content data set. One of the specified category content data is selectively extracted as the topic word from the dialogue record table so that the selected content ratio is selected at a predetermined ratio corresponding to each rank of
The program according to appendix 18.

(Appendix 20)
The computer interaction process is:
Acquire user response message text data and sender identification data accompanying the output of the topic message data,
Updating the preference category table based on the acquired sender identification data, the category relating to the topic message data, and the preference update information corresponding to the response content of the acquired response message text data;
The program according to appendix 18 or 19, further including:

(Appendix 21)
The computer interaction process is:
Obtain the user's message text data and sender identification data,
Determining a category of the acquired comment text data;
Updating the preference category table based on preference update information corresponding to the content of the acquired sender identification data, the determined category, and the acquired message text data;
The program according to any one of appendices 18 to 20, further including:

(Appendix 22)
In the first process, each of the message content data sets indicating the same content in the dialog record table is selected at each predetermined ratio corresponding to each rank of the message content data numbers in each message content data set. As described above, the message content data is selectively extracted as the topic word from the dialog record table.
The program according to any one of appendices 17 to 21.

(Appendix 23)
The dialogue record table stores a topic flag indicating whether or not the content of the speech is suitable for topicing together with the content data of each user.
The extraction as the topic word of the statement content data excludes the statement content data corresponding to the topic flag indicating that it is not suitable for topic swing from the extraction target as the topic word.
The program according to any one of appendices 17 to 22.

1 Voice dialogue server device (dialogue server)
2 Client device (client)
5 Information service site 10 CPU
DESCRIPTION OF SYMBOLS 11 Memory 17 Voice input part 18 Voice output part 21 Voice recognition part 22 Speech synthesizer 25 Dialogue control engine 26 WEB server processing part 31 Topic extraction part 32 Message generation part 33 Data output part 34 Dialog record table 35 Speech data acquisition part 36 Category Determination unit 37 Learning processing unit 38 Preference category table 39 Storage processing unit 41 Schedule processing unit 42 Schedule table 100 Interactive device

Claims (15)

A topic extraction unit that executes a first process of extracting one piece of comment content data as a topic word based on the number of pieces of comment content data indicating the same content from a dialogue record table that stores each piece of user content data;
A message generator for generating topic message data including topic words extracted by the topic extractor;
A data output unit that outputs topic message data generated by the message generation unit;
A dialogue apparatus comprising: The dialog record table stores each user's message content data and categories corresponding to the message content,
The topic extraction unit includes the first process, or
From the preference category table that stores the preference level for each category corresponding to the content of the speech of the target user by learning using the speech history of the target user, one category is identified based on the preference level, and the dialog record A second process of extracting one piece of comment content data of the specified category as the topic word from the dialog record table based on the number of pieces of comment content data indicating the same content of the specified category in the table; The interactive apparatus according to claim 1, which is executed. The topic extraction unit selectively executes the first process and the second process so that the first process and the second process are executed at a predetermined rate.
The interactive apparatus according to claim 2. In the second process, the topic extraction unit is configured such that each statement content data set indicating the same content in the statement content data of the specified category stored in the dialogue record table is each statement content data set. One of the specified category content data is selectively extracted as the topic word from the dialogue record table so that the content is selected at each predetermined ratio corresponding to each ranking of the number of the speech content data.
The interactive apparatus according to claim 2 or 3. A message data acquisition unit that acquires user response message text data and transmission source identification data accompanying the output of the topic message data,
Based on the sender identification data acquired by the message data acquisition unit, the category related to the topic message data, and the preference update information corresponding to the response content of the response message text data acquired by the message data acquisition unit. A learning processing unit for updating the preference category table;
The interactive apparatus according to any one of claims 2 to 4, further comprising: A remark data acquisition unit that acquires remark text data and transmission source identification data of the user;
A category determination unit for determining a category of the comment text data acquired by the comment data acquisition unit;
Based on the sender identification data acquired by the remark data acquisition unit, the category determined by the category determination unit, and the preference update information corresponding to the content of the remark text data acquired by the remark data acquisition unit A learning processing unit for updating the preference category table;
The interactive apparatus according to any one of claims 2 to 5, further comprising: One of the target users corresponding to the category stored in the schedule table is sent to the topic extraction unit at the timing indicated by the timing information stored in the schedule table for storing the timing information and the category for each user. One speech content data is extracted as the topic word, the message generator generates the topic message data, and the data output unit outputs the topic message data.
The interactive apparatus according to any one of claims 2 to 6, further comprising: In the first process, the topic extraction unit is configured so that each comment content data set indicating the same content in the dialogue record table corresponds to each rank of the number of comment content data in each comment content data set. Extracting the content data of the speech selectively as the topic word from the dialogue record table so that each of them is selected in proportion;
The dialogue apparatus according to any one of claims 1 to 7. The dialogue record table stores a topic flag indicating whether or not the content of the speech is suitable for topicing together with the content data of each user.
The topic extraction unit excludes the comment content data corresponding to the topic flag indicating that it is not suitable for topic swing from the extraction target as the topic word,
The dialogue apparatus according to any one of claims 1 to 8. In a computer interaction method executed by at least one computer,
Executing a first process of extracting one piece of content data as a topic word based on the number of pieces of speech content data indicating the same content from the dialogue record table storing the content of each user.
Generating topic message data including topic words extracted by the execution of the first process;
Outputting the generated topic message message data;
A computer interaction method comprising: The dialog record table stores each user's message content data and categories corresponding to the message content,
From the preference category table that stores the preference level for each category corresponding to the content of the speech of the target user by learning using the speech history of the target user, one category is identified based on the preference level, and the dialog record A second process of extracting one piece of comment content data of the specified category as the topic word from the dialogue record table based on the number of pieces of statement content data indicating the same content of the specified category in the table; Selectively executing the first process and the second process so that the first process is executed at a predetermined rate;
The computer interaction method according to claim 10, comprising: In the second process, the message content data sets each indicating the same content in the message content data of the specified category stored in the dialog record table is the number of message content data in each message content data set. One of the specified category content data is selectively extracted as the topic word from the dialogue record table so that the selected content ratio is selected at a predetermined ratio corresponding to each rank of
The computer interaction method according to claim 11. Acquire user response message text data and sender identification data accompanying the output of the topic message data,
Updating the preference category table based on the acquired sender identification data, the category relating to the topic message data, and the preference update information corresponding to the response content of the acquired response message text data;
The computer interaction method according to claim 11 or 12, further comprising: Obtain the user's message text data and sender identification data,
Determining a category of the acquired comment text data;
Updating the preference category table based on preference update information corresponding to the content of the acquired sender identification data, the determined category, and the acquired message text data;
The computer interaction method according to claim 11, further comprising: In a program for causing at least one computer to execute computer interaction processing,
The computer interaction process is:
Executing a first process of extracting one piece of content data as a topic word based on the number of pieces of speech content data indicating the same content from the dialogue record table storing the content of each user.
Generating topic message data including topic words extracted by the execution of the first process;
Outputting the generated topic message message data;
A program that includes that.

Images