Dialogue data generation device, dialogue data generation method, program, and dialogue data generation system

Abstract

A dialogue data generation device 1 includes a scenario identification unit 30 and a dialogue generation unit 40. The scenario identification unit 30 identifies a scenario E for generating dialogue data W related to a book B on the basis of inputted identification information IB for the book B. The dialogue generation unit 40 generates, on the basis of the identified scenario E, dialogue data W corresponding to at least one from among: reaction dialogue R related to a reaction to the content of the book B; quiz dialogue Q related to a quiz about the content of the book B; and chat dialogue C that uses the content of the book B as a starting point but does not have a direct relationship with the content of the book B.

Description

Dialogue data generation device, dialogue data generation method, program, and dialogue data generation system

[0001] The present disclosure relates to a book review dialogue technology.

[0002] Expressing thoughts about texts in books and textbooks (hereinafter collectively referred to as "books"), discussing them with others, and answering questions based on the content are thought to be effective in enriching vocabulary and expressions (hereinafter collectively referred to as "terminology") and helping knowledge to be solidified. In addition, enjoyable conversations are thought to stimulate new reading desire and help establish reading habits.

[0003] Systems that allow people to discuss and post their impressions of books between one another, such as word-of-mouth posting on book purchasing sites and posting on message boards, exist.

[0004] As systems for dialogue between a human and a system, there are, for example, a task-oriented dialogue system as shown in Non-Patent Document 1, and a chat dialogue system as shown in Non-Patent Document 2.

[0005] Mayumi Matsubara et al., "Topic introduction considering user characteristics in task-oriented dialogue systems," Information Processing Society of Japan Transactions on Digital Practice (Web), Vol. 4, p. 3, July 15, 2023, <URL: https: / / www.ipsj.or.jp / dp / contents / publication / 55 / TR0403-13.html> Empirical analysis of training strategies of transformer-based Japanese chat systems. Hiroaki Sugiyama, Masahiro Mizukami, Tsunehiro Arimoto, Hiromi Narimatsu, Yuya Chiba, Hideharu Nakajima, Toyomi Meguro, 2022 IEEE Spoken Language Technology Workshop (SLT), 685-691, 2022.

[0006] However, when using a task-oriented dialogue system, the system may give answers that do not match the content of the book, making it impossible to achieve a reliable dialogue about impressions. When using a casual dialogue system, even if quizzes are posed during the dialogue, the questions may not accurately correspond to the content of the book. A dialogue system that only responds with the same response patterns can easily make users bored. As a result, these issues make it difficult to have an enjoyable dialogue with the system and to motivate users to read new books.

[0007] Therefore, the present disclosure has been made to solve the above problem, and aims to provide a technology that can lead to a new motivation to read.

[0008] In order to solve the above problem, a dialogue data generation device according to one embodiment of the present disclosure includes a scenario identification unit that identifies a scenario for generating dialogue data related to a book based on input book identification information, and a dialogue generation unit that generates dialogue data corresponding to at least one of an impression dialogue related to impressions of the book's contents, a quiz dialogue related to a quiz about the book's contents, and a casual conversation that starts from the book's contents and has no direct relationship to the book's contents, based on the identified scenario.

[0009] According to the present disclosure, a scenario for generating dialogue data related to a book is identified based on the input book identification information, and dialogue data corresponding to at least one of impression dialogue, quiz dialogue, and casual conversation dialogue is generated based on the identified scenario, which can lead to new motivation to read.

[0010] FIG. 1 is a diagram showing an example of the functional configuration of a dialogue data generation system according to a first embodiment. FIG. 2 is a diagram showing an example of implementation of an input device 2 and an output device 3. FIG. 3 is a diagram showing an example of a processing flow of the dialogue data generation system according to the first embodiment. FIG. 4 is a diagram showing an example of the functional configuration of a first dialogue generation unit 41. FIG. 5 is a diagram showing an example of a processing flow of the first dialogue generation unit 41. FIG. 6 is a diagram showing examples of terms used according to age and grade. FIG. 7 is a diagram showing an example of the configuration of a scenario E according to age and grade. FIG. 8 is a diagram showing an example of the configuration of a first dialogue database DB3. FIG. 9 is a diagram showing example information for a quiz dialogue Q. FIG. 10 is a diagram showing an example of the configuration of scenario E. FIG. 11 is a diagram showing an example of the configuration of scenario E. FIG. 12 is a diagram showing another example of the configuration of scenario E. FIG. 13 is a diagram showing another example of the configuration of scenario E. FIG. 14 is a diagram showing another example of the configuration of scenario E. FIG. 15 is a diagram showing another example of the configuration of scenario E. FIG. 16 is a diagram showing an example of a dialogue between a user and a system when the dialogue data generation system 100 is used. FIG. 17 is a diagram illustrating an example of the functional configuration of a computer.

[0011] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. Components having the same functions are designated by the same numbers, and duplicated descriptions will be omitted.

[0012] <<First Embodiment>> <Dialogue Data Generation System 100> As shown in FIG. 1, the dialogue data generation system 100 according to this embodiment includes a dialogue data generation device 1 and an input device 2. ｉ (i=1, 2, 3, ..., m) and output device 3 ｉ(i=1, 2, 3, ..., n), a user information database DB1, a scenario database DB2, a first dialogue database DB3, a dialogue history database DB4, and a book database DB5. Each database (DB1 to DB5) may be provided outside the present system. The maximum number of input devices 2 is m, similar to the input interface units 10 described later, but any number may be used as long as it is equal to or less than the number of input interface units 10. The maximum number of output devices 3 is n, similar to the output interface units 80 described later, but any number may be used as long as it is equal to or less than the number of output interface units 80. Hereinafter, the input devices 2 ｉ and output device 3 ｉ When no particular distinction is required, they may be referred to as "input device 2" and "output device 3," respectively.

[0013] Fig. 2 is a diagram showing an example of the implementation of the input device and the output device. In Fig. 2, only the input device 2 and the output device 3 of the dialogue data generation system 100 are shown, and the others are not shown. The input device 2 and the output device 3 are placed on a specific desk, but the desk is also not shown.

[0014] In FIG. 2, input device 2 １ (hereinafter also referred to as the "first input device") is an RFID (Raid Frequency Identification) reader. １ (hereinafter also referred to as the "first input interface unit"). ２ (hereinafter also referred to as the "second input device") is provided in the input interface unit 10. ２ (hereinafter also referred to as the "second input interface unit"). ３ (hereinafter also referred to as the "third input device") is provided in the input interface unit 10. ３ (hereinafter also referred to as the "third input interface unit"). ３ｂ is microphone 2 ３ Input device 2 is connected to ４ (hereinafter also referred to as the "fourth input device") is the input interface unit 10.４ (hereinafter also referred to as the "fourth input interface unit"). Each input device 2 may be connected to the dialogue data generation device 1 either by wire or wirelessly. In FIG. 2, a wireless connection is illustrated.

[0015] In FIG. 2, output device 3 １ (hereinafter also referred to as the "first output device") is connected to the output interface unit 80. １ (hereinafter also referred to as the "first output interface unit"). ２ (hereinafter also referred to as the "second output device") is connected to the output interface unit 80. ２ (hereinafter also referred to as the "second output interface unit"). ３ (hereinafter also referred to as the "third output device") is an output interface unit 80. ３ (hereinafter also referred to as the "third output interface unit").

[0016] (Outline of Processing of Dialogue Data Generation System 100) The dialogue data generation system 100 performs the dialogue data generation method of this embodiment by carrying out the processing flow illustrated in FIG.

[0017] The dialogue data generation system 100 includes, for example, a display 3 １ The process starts when the start button displayed on the screen is pressed (step S1).

[0018] A user of the system (user U) inserts, for example, a library card into a barcode reader 2. ２ By reading the user information, the user is identified from within the user information database DB1 (step S2).

[0019] Display 3 １ , or Speaker 3 ２ The dialogue data generation system 100 asks the user U about the age, gender, etc., and ３ , Touch Panel 2 ４ The user U is prompted to input an answer from the above (step S3).

[0020] By prompting the user U to select a book, the user U can read the selected book (book B) using, for example, an RFID reader 2. １ This allows the RFID reader 2 １ The identification information (identification information IB) attached to book B is read, and the identification information IB of book B is input to the dialogue data generation system 100 (step S4).

[0021] Based on the input identification information IB, the dialogue data generation device 1 identifies a scenario ("scenario E" described later) from among multiple scenarios in the scenario database DB2 for generating dialogue data regarding book B ("dialogue data W" described later) (step S5).

[0022] Based on the identified scenario E, the dialogue data generation device 1 generates dialogue data (dialogue data W) corresponding to at least one of an impression dialogue (impression dialogue R) related to impressions on the contents of book B, a quiz dialogue (quiz dialogue Q) related to a quiz on the contents of book B, and a casual conversation (chat dialogue C) originating from the contents of book B and not directly related to the contents of book B. ｉ (i=1, 2, 3, . . .)) is generated (steps S6A, S6B, S6C). ｉ The contents of the dialogue data W are output from the output device (output device 3). ｉ Since it does not require any special identification, it will be referred to as "dialogue data W."

[0023] Finally, the dialogue data generation device 1 prompts the user U to answer a questionnaire, and the user U answers if necessary (step S7), and the processing of the system ends.

[0024] It should be noted that steps S1, S2, S3 and S7 are optional functions and may be provided as needed.

[0025] (Dialogue data generation device 1) The dialogue data generation device 1 identifies a scenario E for generating dialogue data W from among a plurality of scenarios in the scenario database DB2 based on the identification information IB of the book B input from the input device 2. Based on the identified scenario E, the dialogue data generation device 1 generates dialogue data W corresponding to at least one of an impression dialogue R, a quiz dialogue Q, or a casual conversation dialogue C about the book B.

[0026] The dialogue data generation device 1 includes an input interface unit 10 ｉ (i=1, 2, 3, . . . , m), an input / output information management unit 20, a scenario identification unit 30, a dialogue generation unit 40, a level change unit 50, a configuration change unit 60, a book determination unit 70, and an output interface unit 80. ｉ (i=1, 2, 3, ..., n). When the dialogue data generation system 100 has each database (DB1 to DB5) within the system, the database may be provided within the dialogue data generation device 1. The book determination unit 70 and the book database DB5 are optional functions and may be provided as needed. Hereinafter, the input interface unit 10 ｉ and an output interface unit 80 ｉ When no particular distinction is required, they may be written as "input interface unit 10" and "output interface unit 80", respectively.

[0027] (Input Interface Unit 10) The input interface unit 10 converts information input from the input device 2 into a predetermined format as needed.

[0028] Taking FIG. 2 as an example, the input interface unit 10 １ RFID reader 2 １ The input interface unit 10 receives the identification information IB of the book B that has been read by the input interface unit 10. ２ is barcode reader 2 ２ The user's identification information IU is read by the button 2. ３ｂ While is pressed, microphone 2 ３ acquires the speech of the user U. ３ , Mike 2３ The input interface unit 10 converts the acquired voice into predetermined text information. ４ Touch Panel 2 ４ The button selection information read by the input / output information management unit 20 is converted into information such as ON / OFF information for a specific button. For example, if only one button is displayed, when that button is selected, the information is converted into information indicating that it has been selected. If multiple buttons are displayed, the information is converted into information indicating which button was selected. Each piece of information received or converted is sent to the dialogue generation unit 30 via the input / output information management unit 20.

[0029] (Input / Output Information Management Unit 20) The input / output information management unit 20 transmits information received from the input interface unit 10 to the dialogue generation unit 40. The input / output information management unit 20 transmits dialogue data W received from the dialogue generation unit 40 to the corresponding output interface unit 80. The input / output information management unit 20 controls the timing of transmission to the output interface unit 80 so that images, sounds, actions, etc. output from each output device 3 can operate in synchronization. The output information management unit 20 may have a function to control the timing of transmission to the dialogue generation unit 40 so that the dialogue generation unit 40 can efficiently process the information input from each input interface unit 10. The input / output information management unit 20 may have a function to temporarily store the received information.

[0030] (Scenario Identification Unit 30) The scenario identification unit 30 identifies a scenario (scenario E) corresponding to book B in the scenario database DB2 based on the input identification information IB of book B. The identified scenario E is sent to the dialogue generation unit 40.

[0031] (Dialogue Generation Unit 40) The dialogue generation unit 40 generates dialogue data W corresponding to at least one of an impression dialogue R, a quiz dialogue Q, and a chat dialogue C based on the identified scenario E. The generated dialogue data W is transmitted to the input / output information management unit 20. The dialogue generation unit 40 has a first dialogue generation unit 41 and a second dialogue generation unit 42.

[0032] (First Dialogue Generation Unit 41) The first dialogue generation unit 41 acquires information corresponding to the content of book B from a predetermined database, and generates dialogue data W corresponding to the impression dialogue R or the quiz dialogue Q. The first dialogue generation unit 41 references information in at least one of the following databases: user information (user information K) of user U in user information database DB1, information for each book in first dialogue database DB3, information on dialogues with user U up until the generation of the dialogue data W in dialogue history database DB4 (dialogue information CI), and book information on book B (book information J) in book database DB5, and generates dialogue data W corresponding to the impression dialogue R or the quiz dialogue Q based on the identified scenario E.

[0033] 4, the first dialogue generation unit 41 has an utterance induction unit 411, an utterance acquisition unit 412, an utterance understanding unit 413, a template selection unit 414, and an utterance generation unit 415. The first dialogue generation unit 41 performs the dialogue generation method for the impression dialogue R and the quiz dialogue Q of this embodiment by carrying out the processing flow illustrated in FIG.

[0034] The utterance eliciting unit 411 induces utterances based on the utterance types (templates) described in the scenario E (step S411). For example, the utterance eliciting unit 411 generates dialogue data W, which is an utterance eliciting utterance to encourage utterances from the user U, such as "What did you think of that story?", and transmits the dialogue data W to the input / output information management unit 20.

[0035] The speech acquisition unit 412 acquires speech by an operation command such as Wait:input[CONTENT_STORY][] (step S412).

[0036] The speech understanding unit 413 understands the speech (step S413). The speech understanding unit 413 mainly performs the following steps (a) to (c).

[0037] (a) What should be understood from the utterance acquired by the utterance acquisition unit 412 is acquired from the currently active command line of scenario E. For example, a command such as Wait:input[CONTENTS_SENTIMENT][POS / NEG] is used to determine whether the content of the acquired [CONTENTS_SENTIMENT] is positive or negative.

[0038] (b) A value of the understanding target is obtained from the obtained user utterance. In the above example, the value of the understanding target is the information of [CONTENTS_SENTIMENT].

[0039] (c) When the object of understanding is identification information IB such as an ISBN, if the identification information IB is available in the first dialogue database DB3, it is stored.

[0040] The template selection unit 414 determines the branching of the IF statement in the scenario E according to the content of the acquired understanding target, and selects the template at the branching destination as the template for the next dialogue data W (step S414).

[0041] The utterance generation unit 415 generates an utterance (step S415). The utterance generation unit 415 determines whether the selected template contains a replacement tag such as [TITLE] or [POS_CONTENT], and if a replacement tag is present, replaces the replacement tag with information contained in the corresponding replacement tag in the first dialogue database DB3. The replaced template is output as dialogue data W.

[0042] After step S415, the process returns to step S412, for example, and executes the commands instructed in scenario E while selecting a template for the next utterance according to the branching of the IF statement described in scenario E and the content of the template utterance, until it reaches the stage of obtaining an utterance from user U (Wait:Input).

[0043] The first dialogue generation unit 41 may generate the dialogue data W by referring to a "connecting question to LLM" (see FIG. 8) that connects to the second dialogue generation unit 42.

[0044] (Second Dialogue Generation Unit 42) The second dialogue generation unit 42 generates dialogue data W corresponding to the chat dialogue C using a predetermined trained language model. The second dialogue generation unit 42 receives a command to generate dialogue data W for the second dialogue generation unit 42 described in a scenario E such as "ForLLM," and generates the dialogue data W that becomes the chat dialogue C. The second dialogue generation unit 42 is, for example, a trained large-scale language model. The chat dialogue C is generated using, for example, the technology of Non-Patent Document 2 described above. The duration and number of dialogues (number of dialogue exchanges) conducted by the second dialogue generation unit 42 are provided by the dialogue generation unit 40.

[0045] (Level Changing Unit 50) The level changing unit 50 changes the difficulty level of the content of the dialogue data W generated by the dialogue generation unit 40 based on the input identification information IU of the user U or dialogue data with the user U (dialogue information CI) up until the time the dialogue data W was generated. For the change, for example, multiple documents of different levels are prepared in the scenario database DB2. Alternatively, multiple paraphrases of different difficulty levels may be prepared in the same scenario E. As shown in FIG. 6 , if the level changing unit 50 determines that the user U frequently uses terms such as "angry," "smiling," and "happy," the level changing unit 50 causes the first dialogue generation unit 41 to generate dialogue data W in which the expressions (phrases) are changed to ones that are appropriate for the content at a preschool child's level. Similarly, if it is determined that the user U frequently uses terms such as "school," "tired," "fragrant," and "angry," the level is determined to be that of a lower elementary school level; if it is determined that the user U frequently uses terms such as "overpass," "dust cloud," and "joy, anger, sorrow, and happiness," the level is determined to be that of a junior high school level; if it is determined that the user U frequently uses terms such as "restoration," "hesitation," and "despair," the level is determined to be that of a high school level; if it is determined that the user U frequently uses terms such as "sorrow, pain from separation," "pathetic," and "wisdom," the level is determined to be that of a university student or above. The level change unit 50 causes the first dialogue generation unit 41 to generate dialogue data W with expressions (phrases) in accordance with the content of the determined level. The level change unit 50 may also cause the second dialogue generation unit 42 to generate dialogue data W with expressions (phrases) in accordance with the content of the determined level, similar to the first dialogue generation unit 41. For example, the dialogue data W may be generated by specifying a condition, such as generating responses for a specified age group. In addition, the dialogue data W may be generated by restricting the use of terms that are beyond the level of the user U or prohibited words, or by simply paraphrasing the generated dialogue, and the terms used for each user U may be individually changed to generate the dialogue data W.

[0046] 6 can be used not only to determine the level of the user U, but also as a list of terms that the dialogue generation unit 40 actively uses or restricts when generating dialogue data W. In this way, when the grade level or age of the user U is known from the user information K or the dialogue between the actual user U and the dialogue data generation system 100, it is possible to impose restrictions such as mainly or frequently using terms that are expected to be learned in that grade.

[0047] The level of user U may be determined not by school year, but by age, after estimating the level of difficulty using technology such as that described in Reference 1. (Reference 1): Sanae Fujita, Tetsuo Kobayashi, Yasuhiro Minami, Hiroaki Sugiyama. "Target Age Estimation of Texts for Children," Cognitive Science, Vol. 22, No. 4, pp. 604-620, 2015. Online ISSN: 1881-5995, Print ISSN: 1341-7924

[0048] (Configuration Modification Unit 60) The configuration modification unit 60 modifies the configuration of the scenario E based on the input identification information IU of the user U or the dialogue data (dialogue information CI) with the user U up until the time of generating the dialogue data W. For example, as shown in FIG. 7 , the configuration of the scenario E is changed depending on the grade level, age, etc., as follows: for 2-3 year olds, only quiz dialogue Q; for 4-5 year olds, "quiz dialogue Q and impression dialogue R," for 6-8 year olds, "quiz dialogue Q, impression dialogue R (lower elementary school level) and chat dialogue C," for 9-11 year olds, "quiz dialogue Q, impression dialogue R (upper elementary school level) and chat dialogue C," for junior high school students, "quiz dialogue Q, impression dialogue R (junior high school level) and chat dialogue C," for high school students, "quiz dialogue Q, impression dialogue R (high school level) and chat dialogue C," and for adults, "quiz dialogue Q, impression dialogue R (university student level or above) and chat dialogue C." In addition, the configuration change unit 60 is not necessarily limited to changing the configuration of the scenario E based on the age or grade indicated in the user information K; for example, it may determine an appropriate configuration based on the dialogue situation of the user U and change to the appropriate configuration midway through the dialogue with the user U.

[0049] (Book Determination Unit 70) The book determination unit 70 determines recommended books (recommended books M) for the user U based on the user information K of the user U or dialogue data with the user U (dialogue information CI) up until the generation of the dialogue data W. For example, the recommended books are determined based on information such as the user's grade, age, preferences, books already read, and recommended books for that age as user information K. Alternatively, if the book currently being read can be identified from the dialogue information CI, this information may also be taken into consideration. Books that the user U has already read may be given a lower priority for selection as recommended books M. Information on the determined recommended books M is transmitted to the first dialogue generation unit 41, and is used as one piece of information when the first dialogue generation unit 41 generates the dialogue data W.

[0050] (Output Interface Unit 80) The output interface unit 80 converts the generated dialogue data W received from the input / output information management unit 20 into a predetermined format. The converted information is output from each output device 3 connected to the dialogue data generation device 1.

[0051] Taking FIG. 2 as an example, the output interface unit 80 １ The output interface unit 80 converts the dialogue data W into a predetermined format required for video output on a display or the like. ２ sends the dialogue data W to speaker 3 ２ The output interface unit 80 converts the audio into a predetermined format required for audio output. ３ The dialogue data W is sent to robot 3. ３ The converted data is then transmitted to the connected output device 3. As described above, the input / output information management unit 20 controls the timing of transmission to the output interface unit 80 so that the images, sounds, and actions output from each output device 3 are synchronized. ３ For example, when making an utterance that elicits a response such as "uh-huh," the robot 3 may perform a specific movement. ３ When making an utterance to encourage the user to speak, the robot 3 ３When saying "Come again," robot 3 ３ When making an utterance that does not involve a specific movement, the robot 3 ３ Alternatively, the robot may be made to perform a randomly selected pattern from a combination of several possible movements.

[0052] (User Information Database DB1) The user information database DB1 is a database that stores information about the user U (user information K). The user information K is information that can identify the user U, such as a user ID (identification information IU), name, gender, age, information on books read, books borrowed, etc. The user information K may be configured so that information obtained by the user U using the dialogue data generation system 100 is added as needed. The identification information IU may be a library card number or a number assigned by the dialogue data generation system 100.

[0053] (Scenario Database DB2) The scenario database DB2 is a database having multiple scenarios E, each consisting of a structure and a sentence template for the dialogue generation unit 40 to generate dialogue data W. The scenarios E may be from the same book, but may have different structures. Multiple scenarios E may be prepared, even if they have the same structure, by changing the terminology according to the age or grade of the user U. The scenarios E may be associated with the child's development (grade, age, etc.), and both the structure and the terminology may be changed. Details of the scenarios E will be described later.

[0054] (First dialogue database DB3) The first dialogue database DB3 is a database that stores information that is referenced when generating dialogue data W for the impression dialogue R or the quiz dialogue Q. The first dialogue database DB3 stores information that is referenced in the scenario E, such as book identification information IB, book impressions, and quizzes.

[0055] 8 is a diagram showing an example of the configuration of the first dialogue database DB3. Two types of books are shown here: one with identification information IB "000-978-4-03206070-6," the title "The Crow's Bakery," and the author "Kako Satoshi," and the other with identification information IB "978-4879817600," the title "A Day in the Cafeteria," and the authors "Otsuka Nao (author) and Ishiyama Azusa (illustrator)." The first dialogue database DB3 has question templates about what the user U likes and dislikes about book B in order to induce the user U to express his or her impressions of book B.

[0056] As an example of a template for a question about a favorite part, we have "I was impressed by [POS_CONTENT]. Do you agree?". Information corresponding to [POS_CONTENT] is stored as "The part where the children's snack bread became popular" for the book "The Crow's Bakery" and "The part where a lot of vegetables are fried all at once" for the book "A Day in the Cafeteria."

[0057] An example of a template for questions about dislikes is "Also, [NEG_CONTENT] was impressive. What do you think?". Information corresponding to [NEG_CONTENT] includes "When the bakery became busy taking care of the baby, the bakery became deserted" from the book "The Crow's Bakery" and "The sight of everyone eating their school lunch with relish" from the book "A Day in the Cafeteria."

[0058] The first dialogue database DB3 has information such as "Oh, speaking of [TITLE], what about [QUESTION_FOR_LLM]?" as an example of a template for a connecting question (connecting question) for transitioning to the second dialogue generation unit 42. When the first dialogue generation unit 41 generates the dialogue data W, information on the title of each book B ([TITLE] in FIG. 8) is incorporated into [TILET]. The first dialogue database DB3 has, as information corresponding to [QUESTION_FOR_LLM], "Do you have a favorite bread?" for the book "The Crow's Bakery" and "What would you be happy to have served at lunch?" for the book "A Day in the Cafeteria."

[0059] The first dialogue database DB3 also has quiz information used in the quiz dialogue Q. In the present disclosure, a quiz is a type of question that primarily has a correct answer. For example, as shown in FIG. 9 , the book "Gurigura" has information corresponding to [QUIZ], such as "What was being cooked in the big frying pan?". It also has information such as "Castella, omelette, and dorayaki" as answer options. In the example of FIG. 9 , a rule is established that the first option is the correct answer. When generating the quiz dialogue Q, the first dialogue generation unit 41 randomly rearranges the information in the answer options to generate the quiz dialogue Q.

[0060] (Dialogue History Database DB4) The dialogue history database DB4 stores the content of the dialogue (dialogue information CI) between the user U and the dialogue data generation system 100. A user ID such as identification information IU is required. However, if a user ID is not available, a session ID, a cookie, or the like may be used. Corresponding to such an identifier, information such as the date and time of the dialogue, the dialogue information CI as text information, and the book selected by the user may also be stored.

[0061] (Book Database DB5) The book database DB5 is a database of books recommended by the book determination unit 70, and is a database that contains at least a set of identification information IB and bibliographic information such as titles. It may also hold information such as the author, artist, publisher, ISBN (which may be the same as the identification information IB), plot summary, book content, reviews, target age group, genre information, recommendation level (which may also be flag information indicating whether or not to recommend), etc.

[0062] (Scenario E) As shown in Figure 10, scenario E is composed of, for example, State, which indicates the transition state of each operation command, and L1 and L2, which indicate the command, and basically, processes related to the generation of dialogue data W are carried out in order from the top row to the bottom row. Examples of each command are as follows (i) to (v). These are merely examples, and the content of the commands may be changed as needed, or the number of commands may be increased or decreased.

[0063] (i) "System:" is a command to generate dialogue data W. The first dialogue generation unit 41 generates the text written after this command as dialogue data W. If the text contains a replacement tag, such as [TITLE], the first dialogue generation unit 41 obtains information corresponding to the replacement tag from any database or from information obtained from the user U, replaces the replacement tag with the obtained information, and then generates the dialogue data W.

[0064] (ii) "Wait: INPUT" is a command to acquire the next utterance. The next process is suspended (waits) until an utterance from the user U or information such as identification information IB is input. The acquired information is stored in the tag written after INPUT.

[0065] (iii) "IF:" is a command that determines whether the input information is a conditional branch. If the condition shown after this command is met, the dialogue data W described in L2 is generated.

[0066] (iv) "TO:" is a command to transition (jump) to a specified State.

[0067] (v) “ForLLM” is a command for transferring the generation process of the dialogue data W to the second dialogue generation unit 42 .

[0068] <Dialogue data generation method (1)> Since the overall flow of the dialogue data generation system 100 has been explained in Figure 3, here we will use two types of scenario E with different configurations and mainly explain the processing performed by the scenario identification unit 30, dialogue generation unit 40, level change unit 50, configuration change unit 60, and book determination unit 70 of the dialogue data generation device 1.

[0069] 10 and 11 show a single scenario E divided into two, which are made up of states S0 to S15. To distinguish it from Figures 12 to 15, which will be described later, the scenario E in Figures 10 and 11 will be referred to as "scenario E1."

[0070] Scenario E1 is an example of a scenario for generating dialogue data W relating to an impression dialogue R and a chat dialogue C. In the explanation of scenario E1 and scenario E2 described later, the input device 2 and output device 3 of the dialogue data generation system 100 will be explained assuming that the devices shown in FIG.

[0071] S0 to S1 of scenario E1 (FIG. 10) are states that determine book B, which is the subject of the dialogue.

[0072] S0: <start>corresponds to step S411 in FIG. 5, and the user U is sent a book to be interacted with by the RFID reader 2. １ The first dialogue generation unit 41 generates dialogue data W such as "Let's talk about your impressions of the book you read" and "Please place the book you read on the stand and read it (please put the book you read on the stand)." Here, the hiragana notation in "( )" indicates that the utterance is to be displayed on the display 3. １ It is used when outputting to a message such as "Hiragana". This allows even children who miss something or cannot read kanji to have a smooth conversation. Below, we will omit the explanation of the hiragana notation in Figures 10 to 15.

[0073] 5, and acquires the identification information IB. Step S412 is a step corresponding to acquiring an utterance, and in S1: GetISBN, the identification information IB is acquired as an input from the user U. The first dialogue generation unit 41 receives the identification information IB from the RFID reader 2. １ The process waits until book B is placed on top (until other information IB is acquired from book B). The information of the read identification information IB is held for use in the review dialogue R.

[0074] S1:forBook corresponds to steps S413 to S415 in FIG. 5 and generates dialogue data W of the user U's reaction to the book selection. If the identification information IB of the selected book B is in the first dialogue database DB3 (IF:hasISBN), the first dialogue generation unit 41 acquires the [TITLE] information corresponding to book B in the first dialogue database DB3 and replaces the acquired information with the [TITLE] portion of the template "You read [TITLE]!" to generate dialogue data W. If the identification information IB and [TITLE] information corresponding to book B can be acquired from, for example, the user information database DB1 or the dialogue history database DB4, they may be acquired from those databases. As a process of the first dialogue generation unit 41, even if the corresponding information is not available in the specified database, if the information exists in another database, the first dialogue generation unit 41 can acquire the information from the database containing that information, as is the case with other processes in this disclosure.

[0075] If the identification information IB of book B is not in the first dialogue database DB3 (i.e., if the necessary information is not available in any database; the same applies below) (IF:notHasISBN), the first dialogue generation unit 41 generates dialogue data W saying, "You've read that book!"

[0076] S1: JudgeRecommendation generates dialogue data W as a reaction to a book that the dialogue data generation system 100 has recommended in the past or a book that the user U has read in the past, based on the user information database DB1 or the dialogue history database DB4. If the book is a book that the user U has recommended in the past (IF: inRecommendedBook), the first dialogue generation unit 41 generates dialogue data W saying, "I'm glad you read the book I recommended!" If the book is a book that the user U has read in the past (IF: inReadBook), the first dialogue generation unit 41 generates dialogue data W saying, "I'm glad you read it again and again!"

[0077] S4, S7, and S8 of scenario E1 (FIGS. 10 and 11) are states that ask about impressions of book B.

[0078] S4: If Book B is in the first dialogue database DB3 (IF:hasISBN), the first dialogue generation unit 41 generates dialogue data W asking, "What did you think of the story?" If Book B is not in the first dialogue database DB3 (IF:notHasISBN), the first dialogue generation unit 41 generates dialogue data W asking, "What was the story about?" The answered content is stored as [CONTENT_STORY].

[0079] S7: In ContentsStory2, if book B selected by user U is in the first dialogue database DB3 (IF:hasISBN), the first dialogue generation unit 41 generates dialogue data W such as "Hmm, I see! That's right!" and "I thought [POS_CONTENT] was impressive. Do you agree?" In this case, the first dialogue generation unit 41 generates dialogue data W by replacing the [POS_CONTENT] portion of the template with the [POS_CONTENT] information obtained from the first dialogue database DB3. If book B is not in the first dialogue database DB3 (IF:notHasISBN), the first dialogue generation unit 41 generates dialogue data W such as "Oh, I see! What did you think?" The obtained answer is stored as [CONTENT_STORY].

[0080] S8: In ContentsStory3, if book B is in the first dialogue database DB3 (IF:hasISBN), the first dialogue generation unit 41 generates dialogue data W such as "That's right," and "Also, [NEGS_CONTENT] was impressive. What about you?" In this case, the first dialogue generation unit 41 replaces the [NEGS_CONTENT] portion of the template with the [NEGS_CONTENT] information obtained from the first dialogue database DB3 to generate dialogue data W. If book B is not in the first dialogue database DB3 (IF:notHasISBN), the first dialogue generation unit 41 generates dialogue data W such as "I see. Thank you. Is there anything else that stands out in your memory?" The obtained answer is stored as [CONTENT_STORY].

[0081] In S8: ContentsStory4, if book B is in the first dialogue database DB3 (IF:hasISBN), the first dialogue generation unit 41 generates dialogue data W including "Yes, yes," and "Oh, speaking of [TITLE], what about [QUESTION_FOR_LLM]?" In this case, the first dialogue generation unit 41 replaces [TITLE] in the template with the [TITLE] information acquired from the first dialogue database DB3, and also replaces [QUESTION_FOR_LLM] in the template with the [QUESTION_FOR_LLM] information acquired from the first dialogue database DB3, thereby generating dialogue data W. [QUESTION_FOR_LLM] is a connection question to the second dialogue generation unit 42. If book B is not in the first dialogue database DB3 (IF:notHasISBN), the first dialogue generation unit 41 generates dialogue data W including "Thank you very much. If you don't mind, let's talk a little more!" and "What are your hobbies?" The obtained answer is stored as [ANSWER_FORQ]. The information in [ANSWER_FORQ] is used as information for the casual conversation dialogue C generated by the second dialogue generation unit 42.

[0082] S9 of scenario E1 (FIG. 11) is a state in which a casual conversation takes place.

[0083] S9: ForLLM generates dialogue data W as a chat dialogue C by the second dialogue generation unit 42. Using the above-mentioned [ANSWER_FORQ] as input information, the second dialogue generation unit 42 generates dialogue data W as the chat dialogue C. The second dialogue generation unit 42, which is a trained model, generates dialogue data W as the chat dialogue C, and a dialogue with the user U is carried out. The number of times and duration of this dialogue back-and-forth are instructed by the dialogue generation unit 40 and are executed in accordance with the content.

[0084] S13 and S15 of scenario E1 (FIG. 11) are ending states.

[0085] In S13: Thanks, the first dialogue generation unit 41 generates dialogue data W including "It was fun chatting with you! Please come again!" and "Finally, please answer the questionnaire!", and ends the dialogue (S15: <end>).

[0086] 12 to 15 show four scenarios E, which are different from scenario E1, and are composed of states S0 to S15. Scenario E in FIGS. 12 to 15 will be referred to as "scenario E2."

[0087] Scenario E2 is an example of a scenario for generating dialogue data W relating to a quiz dialogue Q and an impression dialogue R, and is a scenario including a dialogue regarding a recommended book in the impression dialogue R.

[0088] In the following explanation, in order to avoid duplication, the explanation of the wording of the dialogue data W generated by the dialogue data generation device 1 will be limited to a part, and the explanation will focus on the characteristic differences from Figures 10 to 11 and commands that were not used in Figures 10 to 11.

[0089] S0 to S1 of scenario E2 (FIG. 12) are states that make the user aware of book B, which is the subject of the dialogue.

[0090] In this State, similar to scenario E1, the first dialogue generation unit 41 generates dialogue data W such as "Let's talk about your impressions of the book you read" to identify book B. In S1:GetISBN, the system waits for the input of identification information IB. However, if the identification information IB of book B cannot be recognized in S1:Greeting (IF:notHasISBN), the system generates dialogue data W saying "Thank you for bringing the book!" and returns to S1:GetISBN (TO:GetISBN). That is, RFID reader 2 １ The user will have to wait until book B is placed in the

[0091] Like S1:JudgeRecommendation in scenario E1, S1:JudgeRecommendation generates predetermined dialogue data W if the book is a recommended book (IF:RecommendedBook) or a book that has been read in the past (IF:inReadBook). If either condition is met, the process transitions (jumps) to S7:ContentsStory2 (TO:ContentsStory2). S7:ContentsStory2 is a process that generates dialogue data W that asks about impressions of the book. In other words, in scenario E2, if the book is a recommended book or a book that has been read repeatedly, the process moves to the impression dialogue R without conducting the quiz dialogue Q.

[0092] S2 (FIG. 13) is a state in which a quiz dialogue is carried out.

[0093] S2: EhonHyoshi asks what is depicted on the cover of the picture book. The answer of user U is stored as [HYOSHI_WHAT].

[0094] S2: In EhonHyoshi2, if book B is in the first dialogue database DB3 (IF:hasISBN), the first dialogue generation unit 41 generates dialogue data W saying, "Hmm, yeah. I look like [HYOSHI_WHAT]!" At this time, the [HYOSHI_WHAT] in the template is replaced with the acquired [HYOSHI_WHAT] information to generate the dialogue data W. Regardless of whether book B is in the first dialogue database DB3, the first dialogue generation unit 41 generates dialogue data W saying, "Do you like the way it looks?" The obtained answer is saved as [HYOSHI_IMPRESSION] and a determination is made as to whether the content is positive or negative.

[0095] S2: In EhonHyoshi3, if the acquired [HYOSHI_IMPRESSION] is positive (POS), the first dialogue generation unit 41 generates dialogue data W saying "That's nice!" If the [HYOSHI_IMPRESSION] is negative (NEG), the first dialogue generation unit 41 generates dialogue data W saying "I see."

[0096] S4, S5, S7, and S8 of scenario E2 (FIGS. 13 and 14) are states in which an impression dialogue is carried out.

[0097] In S4: ContentsStory1, in order to ask for opinions on Book B, the first dialogue generation unit 41 generates dialogue data W saying, "I'd like to read this book. What kind of story is it about?" The obtained answer is stored as [CONTENT_STORY]. If Book B is in the first dialogue database DB3 (IF:hasISBN), the first dialogue generation unit 41 generates dialogue data W saying, "Oh, I see! I'm sure other friends also said [POS_CONTENT] was good." At this time, the dialogue data W is generated by replacing [POS_CONTENT] in the template with the information of [POS_CONTENT] obtained from the first dialogue database DB3. If Book B is not in the first dialogue database DB3 (IF:notHasISBN), the first dialogue generation unit 41 generates dialogue data W saying, "Hmm, I see. I see! Other friends also recommended it."

[0098] In S5: ContentsPosNeg, the first dialogue generation unit 41 generates dialogue data W of "Do you like that story?" The obtained answer is stored as [CONTENTS_SENTIMENT], and it is determined whether the content is positive or negative.

[0099] If [CONTENTS_SENTIMENT] is positive (like) content, the first dialogue generation unit 41 generates dialogue data W saying, "That's great! I want to read it soon too!" If [CONTENTS_SENTIMENT] is negative (dislike) content, the first dialogue generation unit 41 generates dialogue data W saying, "I see! Then I'll read it next time too!" The first dialogue generation unit 41 transitions (jumps) to the state S10:RecommendPrep1 (TO:RecommendPrep1). The first dialogue generation unit 41 generates dialogue data W saying, "Oh, I want to recommend that book too!" as S10:RecommendPrep1.

[0100] S7: In order to ask about impressions of Book B, ContentsStory2 causes the first dialogue generation unit 41 to generate dialogue data W asking, "What did you think of the book?" The obtained answer is stored as [CONTENT_STORY].

[0101] S8: BookFavorite asks about book likes and dislikes, just like S5, except that the "TO:RecommendPrep1" process in S5 is not performed.

[0102] S9 to S11 of scenario E2 are states in which a dialogue about recommended books takes place.

[0103] In S9: RecommendPrep2, the first dialogue generation unit 41 generates dialogue data W of "I have other books I want to recommend!" Then, the process transitions to S10: RecommendTitle (TO: RecommendTitle).

[0104] S10: RecommendPrep1 is as described above.

[0105] In S10: RecommendTitle, the first dialogue generation unit 41 generates dialogue data W asking, "Have you ever read the book [R_RECOMMEND_TITLE]?" The obtained answer is stored as [RECOMMEND_TITLE_YONDA], and a determination is made as to whether the content of [RECOMMEND_TITLE_YONDA] is positive or negative.

[0106] If [RECOMMEND_TITLE_YONDA] has positive content, the first dialogue generation unit 41 generates dialogue data W saying, "I see you read it! Actually, this book is my recommendation!" If [RECOMMEND_TITLE_YONDA] has negative content, the first dialogue generation unit 41 generates dialogue data W saying, "I see. Actually, this book is my recommendation!"

[0107] S11: In RecommendReasonAsk, the first dialogue generation unit 41 generates dialogue data W saying, "It's a story about [WHAT_STORY]. Do you want to read it?" At this time, the dialogue data W is generated by replacing [WHAT_STORY] in the template with the [WHAT_STORY] information obtained from the first dialogue database DB3. The obtained answer is stored as [TEXT]. The first dialogue generation unit 41 generates dialogue data W saying, "That's right!"

[0108] S11: In RecommendReason, the first dialogue generation unit 41 generates dialogue data W such as "There is [IMPRESSIVE_WHAT] and it is [IMPRESSIVE_IMPRESSION]!" and "Please give it a read!" At this time, the dialogue data W is generated by replacing [IMPRESSIVE_WHAT] and [IMPRESSIVE_IMPRESSION] in the template with the information of [IMPRESSIVE_WHAT] and [IMPRESSIVE_IMPRESSION] acquired from the first dialogue database DB3.

[0109] S12 to S15 of scenario E2 (FIG. 15) are ending states.

[0110] In S12: Ending, if book B is in the first dialogue database DB3 (IF:hasISBN), the first dialogue generation unit 41 generates dialogue data W saying, "I'll read [YONDA_TITLE] too!" In this case, the dialogue data W is generated by replacing [YONDA_TITLE] in the template with the information of [YONDA_TITLE] obtained from the first dialogue database DB3.

[0111] The first dialogue generation unit 41 generates dialogue data W of "Thank you for talking to me!" The response is stored as [TEXT].

[0112] In S13: Thanks, the first dialogue generation unit 41 generates dialogue data W including "It was fun chatting with you! Please come again!" and "Finally, please answer the questionnaire!", and the process ends (S15: <END>).

[0113] <Example of Dialogue Between User and System> Fig. 16 is a diagram showing an example of a dialogue between a user and a system when using the dialogue data generation system 100. This example is a different scenario from scenarios E1 and E2, and is an example in which the dialogue progresses in the order of quiz dialogue Q, impression dialogue R, and casual conversation dialogue C. In Fig. 16, "S*:" (* indicates the order of the utterance) indicates an utterance by the dialogue data generation system 100, and "U*:" (* indicates the order of the utterance) indicates an utterance by user U.

[0114] In the example of Figure 16, the title "Gurigura" is selected as book B. S1 to S5 and U1 are dialogues related to quiz dialogue Q generated by the first dialogue generation unit 41. In this example, the information shown in Figure 9 is used. In Figure 9, the quiz options included information on "castella, omelette, dorayaki." In the example dialogue of Figure 16, the first dialogue generation unit 41 rearranges the quiz options and outputs "A. omelette, B. dorayaki, C. castella."

[0115] 16, a dialogue regarding impressions is taking place between the user U and the dialogue data generation system 100 along the impression dialogue R generated by the first dialogue generation unit 41. In the dialogues S6 to S8 and U2 to U3 in FIG.

[0116] 16, a casual conversation is taking place between the user U and the dialogue data generation system 100 in accordance with the casual conversation dialogue C generated by the second dialogue generation unit 42. Here, three exchanges of conversations related to casual conversation are conducted based on instructions from the dialogue generation unit 40.

[0117] The dialogue data generation system 100 according to this embodiment has been described above. Based on the input identification information IB of book B, a scenario E for generating dialogue data W related to book B is identified, and based on the identified scenario E, dialogue data W corresponding to at least one of an impression dialogue R, a quiz dialogue Q, and a casual conversation dialogue C is generated. This reduces the likelihood of the system providing a response that is inconsistent with the content of the book, allowing the user to feel that the system understands the content of the user's utterances and the content of book B. Furthermore, because the dialogue data generation system 100 utilizes predetermined databases (DB1 to DB5), it can access accurate information about book B with which user U wishes to have a dialogue, enabling dialogue and recommendations based on, for example, past dialogue information and records.

[0118] The dialogue data generation system 100 allows the user U to answer a quiz in a quiz dialogue Q based on the content of the book B selected by the user U, and to share his or her impressions of the book B in an impression dialogue R. In addition to the quiz dialogue Q and the impression dialogue R, a dialogue can be held in a chat dialogue C starting from the book B. This makes it possible to present the overall flow of the dialogue based on the information in the book B.

[0119] The dialogue data generation system 100 can set quizzes through the quiz dialogue Q. Setting quizzes tends to make even small children more familiar with the system and has an ice-breaking effect. Just as in the impression dialogue R, in which the dialogue data generation system 100 talks about its impressions of book B, the quiz dialogue Q can make the user feel that the system understands the contents of book B. For example, when setting a quiz question, the dialogue data generation system 100 does not tell the user the correct answer even if the answer is incorrect, and instead configures scenario E so that the user U finds the correct answer on their own, which can further encourage reading and solidify the memory.

[0120] The dialogue data generation system 100 allows the user U to chat with the dialogue data generation system 100 using the chat dialogue C, starting from book B. By combining the chat dialogue with the impression dialogue and quiz dialogue, the variety of dialogues can be increased. This allows the user U to enjoy the dialogue with the dialogue data generation system 100 without getting bored. Furthermore, use of the dialogue data generation system 100 can encourage an increase in opportunities to read books. Furthermore, through the responses of the dialogue data generation system 100, the user U can learn that there are various ways of understanding book B, that there are people with various opinions, and can learn about language and terminology through the expressions used by the dialogue data generation system 100.

[0121] The dialogue data generation system 100 can also be configured to freely change the amount of quiz dialogue Q, impression dialogue R, and chat dialogue C. Specifically, the dialogue data generation system 100 may be configured to freely change the amount of dialogue Q, impression dialogue R, and chat dialogue C based on the user U's age, grade, gender, hobbies, and books previously read, as well as the user U's actual responses and information, such as the content, length, and speed of the user U's speech, and the tone and volume of the user U's voice. For example, younger users U tend to find open-ended questions that prompt them to share their impressions and thoughts more difficult than closed questions that offer options such as Yes / No. The dialogue data generation system 100 can also freely change the content of questions based on age and grade, making it possible to encourage more active use of the dialogue data generation system 100 and thereby make reading more familiar to users.

[0122] The dialogue data generation system 100 includes a display 3 １ and Speaker 3 ２ The dialogue data W output from the display 3 can also be changed appropriately to use different terms depending on the age and grade level. １ This means that you can use not only kanji but also hiragana, or output both at the same time.

[0123] The dialogue data generation system 100 can recommend books that suit the user U by taking into consideration the information and responses of the user U described above, and can also guide the user U to discover a new book B.

[0124] The dialogue data generation system 100 includes, for example, a microphone 2 as an input device 2. ３ , output device 3 includes speaker 3 ２ By configuring the output device 3 to include a speaker 3, it is possible to realize a system that is easy to use even for children by enabling dialogue through voice. ２ In addition to the above, there are three displays. １ By providing this function, the system can use both text and voice when speaking, preventing users from missing what is being said through voice alone, and enabling people who cannot read text alone to use the system. Also, when a strange response is given due to an error in voice recognition, the system can make users understand that the response was incorrect at the time of recognition.

[0125] The dialogue data generation system 100 can be provided with an agent such as a robot or an avatar as the output device 3. For example, the dialogue data generation system 100 can be provided with a robot 3 physically attached to the output device 3. ３ By configuring the system to provide a display 3, the user U can sense a personality in the dialogue data generation system 100 and feel a sense of familiarity with it, which can improve motivation for dialogue. １ Children tend to find it easier to talk to a robot than one that is displayed in a simple manner, and this is more likely to attract their interest. A robot that can move is more preferable to a robot that is stationary and does not move. Children, in particular, tend to perceive a moving robot as being smarter than a stationary robot, and are therefore more likely to feel a sense of familiarity with the dialogue data generation system 100.

[0126] The dialogue data generation system 100 includes, for example, a display 3 as an output device 3. １ By configuring the device to include the above, it is possible to present not only text information but also image information such as an avatar. The avatar can perform a variety of actions and expressions that a physical robot cannot express, such as expressing a sad expression while saying that it was scary, making it easier to convey user U's thoughts about book B.

[0127] The dialogue data generation system 100 includes an RFID reader as the input device 2. ２１ , barcode reader 2 ２ The act of reading by a machine generates a reaction, which is enjoyable for children in particular, and encourages the use of the dialogue data generation system 100. １ and barcode reader 2 ２ In the case where both the bar code reader 2 and the bar code reader 3 are provided, for example, ２ The RFID reader 2 reads the identification information IU of the user U. １ By dividing the system into separate sections, such as reading the identification information IB of book B, the operation becomes intuitive and easy to understand.

[0128] Through the various points described above, the user U can trust the dialogue data generation system 100, and can have a more enjoyable dialogue about impressions. Therefore, the dialogue data generation system 100 can lead to a new motivation to read books.

[0129] The above describes the embodiments of the present disclosure. The various processes in the above-described embodiments and modifications may not only be executed in chronological order as described, but may also be executed in parallel or individually depending on the processing capabilities of the devices that execute the processes or as needed. Needless to say, other appropriate modifications are possible without departing from the spirit of the present disclosure.

[0130] The present disclosure may further include a device (terminal) for using the device of the present disclosure or the method of the present disclosure via a network (telecommunications line). The "device (terminal) for use" may be provided with functions (e.g., control function, decoding function, restoration function, input / output function, etc.) necessary to obtain the effects of implementing the device of the present disclosure or the method of the present disclosure.

[0131] [Processor, Program, Recording Medium] The functions performed by the components described herein may be implemented in circuitry or processing circuitry, including general-purpose processors, application-specific processors, integrated circuits, ASICs (Application Specific Integrated Circuits), a CPU (a Central Processing Unit), conventional circuits, and / or combinations thereof, programmed to perform the described functions. A processor includes transistors and other circuits and is considered to be circuitry or processing circuitry. A processor may also be a programmed processor that executes a program stored in a memory.

[0132] In this specification, a circuitry, unit, or means is hardware that is programmed to realize or performs the described functions, which may be any hardware disclosed herein or any hardware known to be programmed to realize or perform the described functions.

[0133] If the hardware is a processor considered to be a type of circuitry, the circuitry, means, or unit is a combination of the hardware and software used to configure the hardware and / or processor.

[0134] The various processes described above can be implemented by loading a program that executes each step of the above method into the recording unit 2020 of the computer 2000 shown in Figure 17, and operating the control unit 2010, input unit 2030, output unit 2040, display unit 2050, etc.

[0135] The program describing the processing contents can be recorded on a computer-readable recording medium, which may be, for example, a magnetic recording device, an optical disk, a magneto-optical recording medium, a semiconductor memory, or any other suitable recording medium.

[0136] The program may be distributed by, for example, selling, transferring, lending, etc. portable recording media such as DVDs and CD-ROMs on which the program is recorded. Furthermore, the program may be stored in a storage device of a server computer, and then transferred from the server computer to other computers via a network, thereby distributing the program.

[0137] A computer that executes such a program may first temporarily store the program recorded on a portable recording medium or transferred from a server computer in its own storage device. Then, when executing a process, the computer reads the program stored on its own recording medium and executes the process in accordance with the read program. Alternatively, the computer may read the program directly from a portable recording medium and execute the process in accordance with the program. Furthermore, the computer may execute the process in accordance with the program each time a program is transferred from a server computer to the computer. Alternatively, the server computer may not transfer the program to the computer, but may instead execute the process through a so-called ASP (Application Service Provider) service, which realizes the processing function by issuing an execution instruction and obtaining the results. Furthermore, the server computer may execute the process at the terminal using a so-called SaaS (Software as a Service) service, which allows users to use part of a server computer along with the program. In this embodiment, the program includes information used for processing by an electronic computer that is equivalent to a program (such as data that is not a direct instruction to a computer but has properties that dictate computer processing).

[0138] Furthermore, in this embodiment, the device is configured by executing a predetermined program on a computer, but at least a part of the processing contents may be realized by hardware.< / end> < / start>

Claims

1. A dialogue data generation device having a scenario identification unit that identifies a scenario for generating dialogue data related to a book based on input book identification information, and a dialogue generation unit that generates, based on the identified scenario, dialogue data corresponding to at least one of an impression dialogue related to impressions of the book's contents, a quiz dialogue related to a quiz on the book's contents, and a casual conversation starting from the book's contents and not directly related to the book's contents.

2. The dialogue data generation device of claim 1, wherein the dialogue generation unit comprises: a first dialogue generation unit that acquires information corresponding to the content of the book from a predetermined database and generates the dialogue data corresponding to the impression dialogue or the quiz dialogue; and a second dialogue generation unit that generates the dialogue data corresponding to the casual conversation dialogue using a predetermined trained language model.

3. The dialogue data generation device according to claim 2, further comprising a level change unit that changes the difficulty level of the content of the dialogue data generated by the dialogue generation unit based on the input user identification information or dialogue data with the user prior to the generation of the dialogue data.

4. The dialogue data generation device according to claim 2, further comprising a configuration change unit that changes the configuration of the scenario based on input user identification information or dialogue data with the user up until the dialogue data is generated.

5. The dialogue data generating device according to claim 1, further comprising a book determining unit that determines books to recommend to the user based on user information about the user or dialogue data with the user up until the dialogue data is generated.

6. A dialogue data generation method in which a scenario identification unit of a dialogue data generation device identifies a scenario for generating dialogue data related to a book based on input book identification information, and a dialogue generation unit of the dialogue data generation device generates, based on the identified scenario, dialogue data corresponding to at least one of an impression dialogue related to impressions of the content of the book, a quiz dialogue related to a quiz on the content of the book, and a casual conversation starting from the content of the book and not directly related to the content of the book.

7. A program for causing a computer to function as the dialogue data generation device according to any one of claims 1 to 5.

8. An interaction data generation system comprising: an interaction data generation device according to claim 1; and a device for using said interaction data generation device via a network.

Images