Information processing equipment, systems, and programs

Abstract

The desired artificial intelligence can be easily configured. [Solution] The information processing device comprises: a natural language acquisition unit that acquires natural language; a user personality acquisition unit that acquires a user personality indicating the user personality to be used from one or more personalities whose behaviors are predetermined to have one or more characteristics; a numerical value acquisition unit that acquires numerical values ​​corresponding to the characteristics possessed by the user personality; and a control unit that outputs the acquired natural language and the numerical values ​​for the user personality together.

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, a system, and a program.

Background Art

[0002] In recent years, with the development of artificial intelligence (AI), technologies for natural language processing (NLP) such as machine translation using AI and dialogue with AI have advanced. For example, Patent Document 1 discloses a technique for generating a system prompt that outputs an appropriate answer.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Generating an appropriate answer or an accurate and natural translation often requires setting a wide variety of parameters, which are difficult for non-experts to understand. Therefore, there is an increasing demand for a technology that can easily set artificial intelligence according to needs.

[0005] Therefore, the present invention has been made in view of the above points, and an object thereof is to disclose a technology that can easily set a desired artificial intelligence.

Means for Solving the Problems

[0006] One aspect of the present invention is an information processing device comprising: a natural language acquisition unit for acquiring natural language; a user personality acquisition unit for acquiring a user personality that indicates the user personality to be used from one or more personalities whose behaviors are predetermined to have one or more characteristics; a numerical value acquisition unit for acquiring numerical values ​​corresponding to the characteristics possessed by the user personality; and a control unit for outputting the acquired natural language and the numerical values ​​for the user personality together.

[0007] Furthermore, in one embodiment of the present invention, the control unit outputs the acquired natural language and the numerical values ​​for the user personality together, thereby causing the user personality having the characteristics based on the numerical values ​​to translate the acquired natural language.

[0008] Furthermore, in one embodiment of the present invention, the system further includes an operation reception unit that receives a personality switching operation from the user, and the control unit changes the personality used for translation to another personality corresponding to the personality switching operation and performs the translation.

[0009] Furthermore, in one embodiment of the present invention, the system further includes an operation reception unit that receives a dialogue switching operation by the user, and the control unit causes the personality used for translation to respond to the natural language.

[0010] Furthermore, in one embodiment of the present invention, the control unit outputs the acquired natural language and the numerical values ​​for the user personality together, thereby causing the personality having the characteristics based on the numerical values ​​to respond to the acquired natural language.

[0011] Furthermore, in one embodiment of the present invention, the system further includes an operation reception unit that receives a personality switching operation from the user, and the control unit changes the personality used for dialogue to another personality corresponding to the personality switching operation and has the control unit respond to the natural language.

[0012] Furthermore, in one embodiment of the present invention, the user personality acquisition unit acquires the personalities used by each user, and the control unit outputs the natural language, the user personalities of each user, and the numerical values ​​for each user personality, thereby causing a correspondence between the natural language and the combined personalities based on the user personalities of each user.

[0013] Furthermore, in one embodiment of the present invention, a language is set for each of multiple users, and the responses from the user personalities are displayed according to the set language.

[0014] Furthermore, one aspect of the present invention is an information processing device comprising: a natural language acquisition unit that acquires natural language; a user personality acquisition unit that acquires a user personality indicating a personality to be used from one or more personalities whose behaviors are predetermined to have one or more characteristics; a numerical value acquisition unit that acquires numerical values ​​corresponding to the characteristics possessed by the user personality; and a control unit that outputs the acquired natural language and the numerical values ​​for the user personality together; and a generation device that operates based on the information output from the information processing device.

[0015] Furthermore, one aspect of the present invention is a program that causes a computer to execute a natural language acquisition step of acquiring natural language; a user personality acquisition step of acquiring a user personality that indicates the user personality to be used from one or more personalities whose behaviors are predetermined to have one or more characteristics; a numerical value acquisition step of acquiring numerical values ​​corresponding to the characteristics possessed by the user personality; and a control step of outputting the acquired natural language and the numerical values ​​for the user personality together. [Effects of the Invention]

[0016] According to the present invention, the desired artificial intelligence can be easily configured. [Brief explanation of the drawing]

[0017] [Figure 1] This figure shows how the technology according to the embodiment is used for translation. [Figure 2] It is a block diagram for explaining an aspect of the system 1 according to an embodiment. [Figure 3] It is a block diagram for explaining an example of the functional configuration of the information processing apparatus 10 according to an embodiment. [Figure 4] It is a diagram for explaining an outline of the functions of the information processing apparatus 10 according to an embodiment. [Figure 5] It is the first display screen D1 showing an acquisition example of the numerical value N. [Figure 6] It is a flowchart for explaining an example of the processing of the information processing apparatus 10 according to an embodiment. [Figure 7] It is a diagram for explaining an outline of the functions of the information processing apparatus 10 according to the first modification example. [Figure 8] It is a diagram for explaining an outline of the functions of the information processing apparatus 10 according to the second modification example. [Figure 9] It is a diagram showing an outline of an example of the hardware configuration of the information processing apparatus 90 applied to the present embodiment.

Mode for Carrying Out the Invention

[0018] [Embodiment] Regarding the information processing apparatus, system, and program according to the present embodiment, preferred embodiments will be described in detail below with reference to the accompanying drawings. In the description of the drawings, the same or similar parts are denoted by the same or similar reference numerals. Note that the present embodiment is not limited to these embodiments, and also includes those with various modifications or improvements. That is, the constituent elements described below include those that can be easily assumed by those skilled in the art and substantially the same ones, and the constituent elements described below can be combined as appropriate. Also, in the present embodiment, various omissions, substitutions, or changes of the constituent elements may be made without departing from the gist of the present invention.

[0019] In all the figures used to illustrate the embodiments, components with the same function are given the same reference numerals, and repeated explanations are omitted. Furthermore, in this application, "based on XX" means "based on at least XX," and includes cases where it is based on other elements in addition to XX. Also, "based on XX" is not limited to cases where XX is used directly, but also includes cases where it is based on XX after calculations or processing have been performed on it. "XX" is any element (for example, any information). Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0020] First, with reference to Figure 1, we will explain an example of using artificial intelligence for translation. Figure 1 is a diagram showing how the technology according to the embodiment is used for translation.

[0021] Conventionally, technologies are known for systems that allow multiple users to conduct online meetings, business negotiations, medical consultations, education, training, etc., while simultaneously referring to screen displays on their respective terminal devices. Translation technologies are also known that translate the speech of the first user U1 into the language used by the second user U2 and output it.

[0022] However, machine translation is not always accurate. For example, it is difficult to accurately translate expressions and concepts unique to a particular country. For instance, the panel layout and methodology used in manga are not common in other countries, and translation errors can occur.

[0023] Furthermore, machine translation often has difficulty reflecting the nuances of spoken language in its translation results. Therefore, machine translation is not suitable for situations where subtle nuances are important, such as business negotiations.

[0024] According to the information processing device, system, and program of this embodiment, translation can be performed with the nuances desired by user U. Furthermore, it is possible to appropriately translate even specialized terminology and expressions unique to a particular country or region. Moreover, these settings can be easily and intuitively configured. Artificial intelligence may be used for various tasks such as dialogue and document generation.

[0025] Figure 2 is a block diagram illustrating an embodiment of System 1. System 1 comprises an information processing device 10, one or more user terminals 20, and a generation device 30. The first user terminal 21 to the third user terminal 23 illustrate a plurality of user terminals 20. In the following description, when the first user terminal 21 to the third user terminal 23 are not distinguished, they may simply be referred to as the information processing device 10.

[0026] The information processing device 10 and the user terminal 20, and the information processing device 10 and the generation device 30 are connected to each other via a network NW so that they can communicate with one another. The network NW may be a wireless communication network or a wired communication network. The network NW may be configured using, for example, the Internet or a local area network (LAN). The network NW may be configured by combining multiple networks.

[0027] [Example of user terminal configuration] The user terminal 20 is, for example, an information processing device such as a personal computer, tablet, or smartphone. The user terminal 20 comprises a user terminal display unit, a user terminal operation unit, a user terminal calculation unit, and a user terminal storage unit.

[0028] The user terminal display unit, for example, is equipped with a liquid crystal display and displays various images. In the following description, the display of an image by the user terminal display unit is also referred to as presenting an image to the user. User U is, for example, a person who uses the information processing device 10.

[0029] User U, for example, is equipped with a touch panel to detect user operations. The user terminal operation unit may also acquire user operation information from input devices such as a keyboard, mouse, and microphone. User operations may be performed by various methods, such as touching a display unit such as a button on the user terminal operation unit, using predetermined keys or a wheel on an input device, or using voice commands.

[0030] The user terminal processing unit is equipped with a central processing unit and operates based on the programs and data stored in the user terminal storage unit, providing various functions. The user terminal storage unit is composed of, for example, a hard disk drive or semiconductor memory, and stores various information such as programs and data read by the user terminal processing unit. The user terminal storage unit may also be implemented by a virtual storage device such as a cloud server located outside the user terminal 20.

[0031] [Example of LLM configuration] The generation device 30 represents an AI that has been pre-trained using natural language, images, or audio. The generation device 30 may be built by the user, or it may be an externally built LLM (Large Language Model), etc. Also, if the generation device 30 is built by the user, it may be provided within the information processing device 10. Hereinafter, the process of supplying information to the generation device 30 to generate desired information will be simply referred to as "generating". The generation device 30 may also be called a generation unit.

[0032] [Example of an information processing device configuration] Figure 3 is a block diagram illustrating an example of the functional configuration of the information processing device 10 according to the embodiment. The information processing device 10 comprises an arithmetic unit 11 and a storage unit 13.

[0033] The arithmetic unit 11 is equipped with a central processing unit (CPU) and operates based on the programs and data stored in the storage unit 13, providing various functions.

[0034] The storage unit 13 is composed of, for example, a hard disk drive or semiconductor memory (flash memory, RAM, ROM), and stores various types of information, such as programs and data read by the arithmetic unit 11. The storage unit 13 may also be implemented by a virtual storage device, such as a cloud server, located outside the information processing device 10.

[0035] [Functions of the information processing device] The calculation unit 11 includes, as its functional units, a communication unit 110, an operation reception unit 111, a display control unit 112, a natural language acquisition unit 113, a user personality acquisition unit 114, a numerical value acquisition unit 115, a knowledge information acquisition unit 116, a personality setting unit 121, a translation instruction unit 122, and a dialogue instruction unit 123. Each of these functional units may be implemented using electronic circuits as needed. Furthermore, each functional unit does not have to be included in a single device, and the information processing device 10 may be composed of multiple devices. The personality setting unit 121, the translation instruction unit 122, and the dialogue instruction unit 123 may be collectively referred to as the control unit 120.

[0036] The communications unit 110 communicates information with the user terminal 20.

[0037] The operation reception unit 111 receives various operations performed on the user terminal 20. The operation reception unit 111 activates each part of the calculation unit 11 according to the various operations received.

[0038] The display control unit 112 controls the display on the user terminal 20. Specifically, the display control unit 112 generates a screen to be displayed on the user terminal display unit and transmits it to the user terminal 20 via the communication unit 110. The user terminal 20 displays the screen generated by the display control unit 112. The display control unit 112 may also present audio or other information, not just a screen. In this case, the display control unit 112 may be called a presentation control unit.

[0039] Figure 4 is a diagram illustrating the overview of the functions of the information processing device 10 according to this embodiment. The functions of the information processing device 10 will be explained with reference to Figure 4. In Figure 3, the information processing device 10 creates a system prompt SP, etc., to be supplied to the generation device 30, based on natural language NL, information indicating personality P, user U, and knowledge information K.

[0040] The natural language acquisition unit 113 acquires natural language NL. Natural language NL is input, for example, based on the operation of user U. Natural language NL is the content input to the generation device 30. Natural language NL is, for example, the content spoken by natural language NL and is the content to be translated.

[0041] [Personality Overview] The user personality acquisition unit 114 acquires information indicating the personality P (user personality) to be used. Personality P indicates the behavior of the generation device 30 when responding to translation, dialogue, etc. For example, the user personality acquisition unit 114 acquires information indicating the personality P selected by user U as the personality P to be used from one or more personality Ps. The information indicating the user personality CP may be, for example, information that identifies the user personality CP, or information that indicates definitions within the user personality CP.

[0042] Personality P's behavior is defined and determined based, for example, on knowledge and one or more characteristics. Personality P's behavior refers to, for example, the roles required by user U, etc., and the occupations that enable them to fulfill those roles. In the following explanation, Personality P's behavior may sometimes be simply referred to as occupation or role.

[0043] The traits possessed by personality P are one or more elements that constitute personality P's character (Personality Traits), and are equivalent to behavioral traits (also called speech traits) in humans. In other words, the character of personality P is defined by a combination of one or more traits.

[0044] The degree and direction of a characteristic are defined by a numerical value N that is greater than or equal to a first value (e.g., 0.0) and less than or equal to a second value (e.g., 1.0). When the numerical value N changes, at least one of the degree and direction of the characteristics possessed by personality P, i.e., the tendency of the response, changes. The numerical value acquisition unit 115 acquires the numerical value N corresponding to the characteristics of personality P. The numerical value N is changed, for example, based on the operation of user U. By adjusting the numerical value N, user U adjusts the characteristics corresponding to the numerical value N, and consequently, the personality P that performs translation, etc. This allows user U to set the desired personality P and convey nuances that are difficult to convey during translation.

[0045] Furthermore, a person's speaking style may change mid-conversation; for example, they might be polite and formal at the beginning of a meeting, but become more relaxed and friendly towards the end. By changing the numerical value N during the conversation, even large-scale language models that rely solely on text can express emotions, nuances, and other aspects that are difficult to convey during translation in a way that more closely matches the user U's intentions. It also allows for more explicit behavioral control of models that are trained on non-textual information such as speech.

[0046] The numerical value N may be adjusted continuously (linearly) between the first and second values, or it may be adjusted in steps. The ability to input the numerical value N continuously is preferable because it allows the user U to accurately convey even subtle nuances of what they want to communicate during translation, etc. Furthermore, compared to inputting the numerical value N in steps, the ability to input the numerical value N continuously allows for a more intuitive adjustment of parameters such as personality, which are difficult to articulate clearly, making it more suitable for adjusting the degree and direction of personality P.

[0047] The characteristics may be shared and established for each personality P, or they may be defined for each personality P. Defining characteristics for each personality P is preferable because it allows for the determination of characteristics that are more appropriate for each behavior of personality P.

[0048] The knowledge possessed by personality P (Domain Knowledge) is knowledge corresponding to its behavior, and includes, for example, prerequisite knowledge for its behavior, knowledge of concepts similar to the prerequisite knowledge (also called similar knowledge), mapping information that links the prerequisite knowledge and similar knowledge when the prerequisite knowledge is unique to a particular region or country, and general knowledge required when fulfilling its role. Prerequisite knowledge may include general prerequisite knowledge collected from the internet, etc., and practical prerequisite knowledge such as practitioner records, laws, precedents, the schedules of specific individuals, and past contracts. The knowledge information acquisition unit 116 acquires knowledge information K that indicates the knowledge that defines personality P. Knowledge information K may be acquired based on the operation of user U, or it may be information that has been stored in advance in the storage unit 13, etc. By further defining personality P using knowledge information K, it is possible to easily set up a model that has narrow and deep knowledge like that of a human expert, as well as a communication style with others.

[0049] [Examples of personality traits] Next, we will explain examples of personality P's configuration, such as personality P acting as a "manga editor," "film producer," "negotiator," "mentor," or "religious figure."

[0050] [Example of personality type: Manga editor] The knowledge possessed by personality P, who acts as a manga editor, may include, for example, background knowledge such as knowledge of manga both domestically and internationally; more practical background knowledge such as technical information on production that is actually used by manga artists but not made public; similar knowledge such as knowledge of anime, movies, novels, plays, and show business both domestically and internationally; and mapping information used to map Japanese-specific knowledge to languages ​​such as English. Mapping information refers to the correspondence between background knowledge and similar knowledge, such as "a manga 'name' is equivalent to a 'Draft Storyboard' in American film production" or "a manga panel layout is equivalent to pan, tilt, and zoom in visual expression." Because large-scale language models are composed mostly of English corpora (language dictionaries), which has a large number of speakers, it is difficult to accurately understand or translate concepts unique to specific regions or countries. By substituting corresponding words based on mapping information, the understanding of large-scale language models can be made more accurate, and the accuracy of responses such as translations can be improved.

[0051] The characteristics of personality P acting as a manga editor include, for example, acceptance and frankness. Acceptance indicates the degree to which one accepts what the other person says; when the value N is 1.0, personality P accepts what the other person says, and when the value N is 0.0, personality P strongly proposes something. Frankness indicates the degree of frankness in one's way of speaking; when the value N is 1.0, personality P speaks frankly and casually, and when the value N is 0.0, personality P speaks in a businesslike manner.

[0052] According to this, by combining specialized knowledge and practical experience regarding manga and related concepts, it is possible to construct a personality P that can accurately understand the content, speak in a firm tone when giving strong instructions, and be accommodating when it is appropriate to bring out the strengths of the other party (for example, a manga artist).

[0053] [Example of personality type: Film Producer] The knowledge possessed by personality P, who acts as a film producer, may include, for example, background knowledge about domestic and international films, databases of actors, production technicians, and guilds that can be hired, knowledge of domestic and international cinemas and distributors, practical knowledge of distribution, databases of potential sponsors, information on past sponsorship forms and contracts, knowledge of laws and regulations regarding expression methods in various countries, similar knowledge such as knowledge of domestic and international manga, anime, novels, theater, and show business, and mapping information from Japanese-specific knowledge to English.

[0054] The characteristics of personality P, who acts as a film producer, include acceptance of what others say and financial sense. Financial sense, for example, indicates the degree to which feasibility based on budget is considered. When the numerical value N is 1.0, personality P prioritizes feasibility considering the budget, and when the numerical value N is 0.0, personality P proposes based on possibility without considering the budget.

[0055] Film productions generally involve enormous budgets, and it's difficult to make a film without considering specific details such as the schedules and compensation of the actors and crew involved. Therefore, personality P, acting as a film producer, can handle not only script-related matters but also more specific details such as budget and casting.

[0056] [Personality trait example: Negotiator] The knowledge possessed by personality P acting as a negotiator may include, for example, background information regarding the businesses, stakeholders, and partners of both parties involved in the negotiation, as well as more practical background knowledge such as information on past negotiation cases and precedents, and information on the publicly known creditworthiness of the other party.

[0057] The characteristics of personality P acting as a negotiator include, for example, acceptance of the other party's statements, menace, and politeness. Menace indicates the degree to which the risks of the proposal not being accepted are presented to the other party. When the value N is 1.0, personality P strongly presents the risks of the proposal not being accepted, and when the value N is 0.0, personality P does not conduct a risk assessment of the facts of the proposal. Politeness indicates the degree of politeness of the writing style. When the value N is 1.0, personality P uses a polite writing style, and when the value N is 0.0, personality P uses a rough writing style.

[0058] According to personality P, who acts as a negotiator, user U can appropriately convey their tone and nuances to the other party. Furthermore, by conversing with personality P acting as a negotiator, user U can supplement the knowledge required for negotiations and conduct negotiations accordingly.

[0059] [Example of personality type: Mentor] The knowledge possessed by personality P, who acts as a mentor, includes, for example, background knowledge such as information on the words and actions of real or former prominent businessmen, leaders, and visionaries; knowledge of the thinking methods that form the core of their beliefs; information on domestic and international business practices; and general knowledge of general culture, religion, and lifestyles outside of business, both domestically and internationally.

[0060] The specific characteristics of personality P acting as a mentor include, for example, acceptance of what the other person says, unexpectedness, and extroversion. Unexpectedness indicates, for example, the degree to which the personality P makes suggestions that deviate from the topic of conversation. When the value N is 1.0, personality P makes suggestions that deviate from the current topic of conversation, and when the value N is 0.0, personality P makes suggestions that are in line with the current topic of conversation. Extroversion indicates, for example, the degree to which the perspective when responding is extroverted. When the value N is 1.0, personality P has an extroverted perspective, and when the value N is 0.0, personality P has an introverted perspective.

[0061] According to personality P, who acts as a mentor, he can provide appropriate answers to user U's inquiries regarding the matters they desire.

[0062] [Example of personality composition: religious person] The knowledge possessed by personality P, who acts as a religious practitioner, includes, for example, background knowledge such as the full text of the scriptures of their own religion or sect, information on the history of their own religion or sect, and the full text of the scriptures of other religions or sects, as well as mapping information on how the histories of other religions or sects are expressed from their respective perspectives.

[0063] The characteristics of a personality P acting as a religious person include, for example, self-sacrifice and loyalty. Self-sacrifice indicates, for example, the degree to which one tends to make sacrifices. When the value N is 1.0, personality P prioritizes actions for the happiness of others, and when the value N is 0.0, personality P prioritizes actions for the realization of their own happiness. Loyalty, on the other hand, indicates, for example, the degree to which one values ​​the precepts of scriptures. When the value N is 1.0, personality P prioritizes the precepts of scriptures, and when the value N is 0.0, personality P prioritizes their own interpretation of scriptures.

[0064] According to this, it is possible to construct a personality P that acts as a religious figure with various viewpoints. By engaging in dialogue with this personality P and generating documents, user U can appropriately deepen their understanding of religion and denominations.

[0065] [Processing of the control unit] The control unit 120 makes the system respond to the natural language NL based on the information acquired by the knowledge information acquisition unit 116 from the natural language acquisition unit 113. Specifically, the control unit 120 generates a system prompt to translate the natural language NL and causes the generation device 30 to generate the translated text.

[0066] The personality setting unit 121 of the control unit 120 sets the personality P to be used. Specifically, the personality setting unit 121 generates a system prompt SP to set the personality P to be used based on the information indicating the user personality acquired by the user personality acquisition unit 114, the type of characteristics that define the user personality, and the knowledge information K acquired by the knowledge information acquisition unit 116.

[0067] The translation instruction unit 122 of the control unit 120 generates a system prompt SP that instructs the generation device 30 to perform translation. Specifically, the translation instruction unit 122 obtains a system prompt SP for the personality P (user personality) set by the personality setting unit 121, a numerical value N obtained by the numerical value acquisition unit 115 for the user personality, and natural language NL. Obtaining the numerical value N for personality P means obtaining a numerical value N set by user U for at least one of the characteristics of personality P, or a pre-set numerical value N (default value). Based on the acquired information, the translation instruction unit 122 generates a system prompt SP that causes the personality P with the personality set by user U to translate the natural language NL. Specifically, the translation instruction unit 122 communicates the correspondence for the first value (0.0) and the correspondence for the second value (1.0), and instructs the system to perform the corresponding action according to the numerical value N based on these.

[0068] The control unit 120 provides the generated system prompt SP to the generation device 30. The generation device 30 acts as the configured personality P according to the given system prompt SP and translates the natural language NL. At this time, country-specific expressions are appropriately translated so that they can be understood in other countries, based on mapping information, etc. Furthermore, since the translation is performed in accordance with the characteristics set by user U, the generated translation result reflects the intentions and nuances of user U. The information processing device 10 acquires the translation result generated by the generation device 30. The display control unit 112 also presents the translation result to user U. Therefore, with the information processing device 10, it is possible to easily set up artificial intelligence capable of accurate and natural responses. As described above, since instructions are given by the system prompt SP, prior training of the generation device 30 is unnecessary and operation is easy. Note that the system prompt SP is an example of the format of the information to be supplied, and this embodiment is not limited to this example.

[0069] In the above description, an example of a system prompt SP generated by the personality setting unit 121 being output to the translation instruction unit 122 was explained, but this embodiment is not limited to this example. The personality setting unit 121 may also provide the generation device 30 with a system prompt SP in advance.

[0070] In the above description, the personality setting unit 121 and the translation instruction unit 122 were described as having different functions, but this embodiment is not limited to this example. For example, the personality setting unit 121 and the translation instruction unit 122 may be implemented by a single function unit that has both functions.

[0071] In the above description, an example was given of the user personality acquisition unit 114 acquiring personality information indicating a personality P selected by user U. In this case, it is not necessary to explicitly state that it is selected by user U; for example, a function to which a certain personality P is set may be used.

[0072] In the above description, an example was given in which the natural language acquisition unit 113 acquires natural language NL input by user U. However, this embodiment is not limited to this example. When multiple generation devices 30 interact with each other, the natural language acquisition unit 113 may acquire content generated by one of the generation devices 30.

[0073] [Example of obtaining the numerical value N] Next, we will explain an example of obtaining the numerical value N, referring to Figure 5. Figure 5 is the first display screen D1, which shows an example of obtaining the numerical value N. The first display screen D1 has a screen configuration consisting of a first language display unit D11, a second language display unit D12, a numerical value adjustment unit D13, and a dialogue switching unit D14.

[0074] The first language display unit D11 displays the spoken content of each user U, such as the first user U1 and the second user U2, in the language of the first user U1 (for example, Japanese). That is, the content spoken by the first user U1 is displayed as is in Japanese, while the content spoken by others, such as the second user U2, is translated into Japanese, the language of the first user U1, and displayed. The language of a user U may be determined, for example, by obtaining the user U's settings using a language acquisition unit (not shown).

[0075] In response, the second language display unit D12 displays the spoken content of the first user U1 and the second user U2 in the language used by the second user U2 (for example, English). Specifically, the content spoken by the second user U2 is displayed in English, while the content spoken by the first user U1 is displayed in Japanese.

[0076] The first language display unit D11 and the second language display unit D12 display the content of each utterance separately in different display areas. Furthermore, the same utterance is displayed in corresponding positions (side-by-side) in both the first language display unit D11 and the second language display unit D12.

[0077] In this case, the persona P to be used for translation may be predetermined, or the persona P to be used for translation may be determined based on the actions of user U. In Figure 5, the persona P used by the first user U1, that is, the persona P that translates the content of first user U1's speech into the language used by second user U2, is, for example, "manga editor".

[0078] The numerical adjustment unit D13 adjusts the numerical value N corresponding to the characteristics according to the operation. The numerical adjustment unit D13 displays adjustment fields for the characteristics of the "manga editor": Acceptance and Frankness of the other party's statements. Each adjustment field displays a slider that allows the numerical value N to be adjusted between 0.0 and 1.0. The user U can continuously adjust the numerical value N of the target characteristic by adjusting the position of the line segment perpendicular to the line connecting 0.0 and 1.0. The display of the numerical adjustment unit D13 may be a rotary dial or the like instead of a slider.

[0079] The numerical adjustment unit D13 may be operated, for example, by directly touching (tapping or sliding) the user terminal display screen. Alternatively, the numerical adjustment unit D13 may be operated based on operation information from an external operating device that is communicatively connected to the user terminal 20. The operating device may be various devices such as a mouse (clicking or using the scroll wheel), a slider, a rotary dial, a foot switch, or a terminal device with an operating application installed.

[0080] In this case, to facilitate smooth conversation, it is preferable that the settings are not made known to other meeting participants. Therefore, the numerical adjustment unit D13 is not displayed to, for example, the second user U2, who is the conversation partner, and the existence of the numerical adjustment unit D13 and its settings are not notified to the second user U2. Furthermore, when the above technology is used in a situation where the first user U1 and the second user U2 are in the same space, the numerical adjustment unit D13 itself may not be displayed, and the numerical value N may be adjusted by receiving operation information from an external operating device.

[0081] The numerical value N may be adjusted not only by displaying it on the user terminal display or operating device, but also by analyzing the voice, sound quality, speech characteristics, etc., or by responding to a predetermined wake word indicating the adjustment of the numerical value N. Furthermore, when analyzing the voice, sound quality, speech characteristics, etc., the numerical value N may be automatically adjusted based on the analysis results, even without being instructed to provide a specific numerical value.

[0082] The dialogue switching unit D14 may change the response of personality P in response to an operation (dialogue switching operation). Specifically, when the dialogue switching unit D14 is operated, it puts personality P, which is used for translation, into a state where it can converse. Conversational means that the content of the first user U1's utterance is set as a question to personality P and generates an answer. Specifically, the dialogue instruction unit 123 provided in the control unit 120 specifically acquires a numerical value N obtained by the numerical value acquisition unit 115 for the personality being used, and natural language NL which is the content of the question. Based on the acquired information, the dialogue instruction unit 123 generates a system prompt SP that causes personality P, which was set for translation, to generate an answer to natural language NL. Explanations of other matters similar to those of the translation instruction unit 122 are omitted.

[0083] In this case, it is preferable that the questions from the first user U1 and the answers from personality P are displayed on the second language display unit D12, and also translated into the language used by the second user U2 and displayed on the second language display unit D12. This allows the first user U1 and the second user U2 to easily understand the questions and answers from personality P and to proceed with the conversation smoothly as if one person were participating. The personality P used for translation may be changed to a different personality P as needed. Changing personality P may be performed by various operations (personality switching operations), such as operations on the user terminal display unit, operations on the operating device, or voice operations.

[0084] [Sum of personalities] In the above, we described a personality P that translates the utterances of the first user U1 into the language used by the second user U2. In this case, the second user U2 may set up and use a separate personality P that translates its own utterances into the language used by the first user U1. These personalities P may be set up separately and independently, or they may be combined. A personality formed by combining multiple personalities P may be called a combined personality. With a combined personality, it is expected that the quality of both speaking and listening in conversations between multiple users U will improve.

[0085] Specifically, the combined personality may be the sum of the knowledge possessed by each individual personality P. Furthermore, the combined personality may have characteristics that have been adjusted using various methods such as the averaging method, weighted selection method, and parameter selection method, based on the characteristics of each individual personality P. The user U may choose which of these methods to employ.

[0086] The averaging method involves dividing the influence of each personality P on thinking, i.e., by the number of participants. Specifically, if four users U use four personality Ps, the combined personality is calculated by dividing the sum of the numerical values ​​N corresponding to each trait by the number of participants, which is then used as the value N for each trait. In this case, the numerical values ​​for traits that have not been set may be set to an average value of 0.5.

[0087] Alternatively, the combined personality value N may be determined by dividing by only the personality P for which the trait is set. Specifically, if three of the four personality Ps have the trait of tolerance for what others say set, and their respective values ​​N are 0.4, 0.7, and 0.7, then the combined personality value N may be 0.6, obtained by dividing the sum of these values ​​by the three.

[0088] The weighted selection method involves setting a percentage indicating which personality P's characteristics will have the strongest influence, and then forming the combined personality according to that percentage. This percentage may be arbitrarily selected by the user U. Specifically, user U may set it to 60% manga editor, 20% negotiator, and 20% mentor. The combined personality will be influenced by each personality P according to the aforementioned percentage. In this case, the numerical value N of the traits used for the summation may be, for example, a predetermined value.

[0089] Furthermore, the proportions may be determined according to the types and number of personality Ps selected by each user U. For example, if four users U each use manga editor, manga editor, negotiator, and mentor as personality Ps, the combined personality distribution may be set as 50% manga editor, 25% negotiator, and 25% mentor.

[0090] The parameter selection method is a case where each user U can adjust the numerical value N for the combined personality. In this case, user U is not required to set the numerical value N for all characteristics. The numerical values ​​N for characteristics not set by user U may be automatically set by various methods such as the averaging method or the weight selection method described above.

[0091] [Process flow] Figure 6 is a flowchart illustrating an example of processing by the information processing device 10 according to the embodiment. The natural language acquisition unit 113 acquires natural language NL input from user U or other generation devices 30, etc. (step S101). The user personality acquisition unit 114 acquires personality information indicating the personality P to be used (step S102). The numerical value acquisition unit 115 acquires numerical values ​​N about the characteristics of the user personality based on user U's operation or from the storage unit 13 (step S103). The knowledge information acquisition unit 116 acquires knowledge K that the user personality possesses, i.e., knowledge information required for defining the user personality, based on user U's operation or from the storage unit 13 (step S104). The personality setting unit 121 sets the behavior (occupation and role) of the personality P to be used based on the type of personality P acquired by the user personality acquisition unit 114, the numerical value N for personality P acquired by the numerical value acquisition unit 115, and the knowledge information K acquired by the knowledge information acquisition unit 116 (step S105). The translation instruction unit 122 gives instructions to the personality setting unit 121 to perform translation, responses, etc. to natural language NL according to the behavior of personality P set by the personality setting unit 121 (step S106). The information processing device 10 acquires the generation result (response result of the generation device 30) generated in accordance with the instructions (step S107). The display control unit 112 presents the generation result to the user U by outputting the generation result to the user terminal 20 (step S108). Note that the processes from steps S101 to S108 may be swapped, and multiple processes may be performed together as in steps S105 and S106.

[0092] [First variation] In the above description, an example was given in which the knowledge information K is included in the system prompt SP or supplied to the generation device 30 together with the system prompt SP. However, this embodiment is not limited to this example. The knowledge information K may be supplied to the generation device 30 separately. Figure 7 is a diagram illustrating the overview of the functions of the information processing device 10 according to the first modified example. For example, the knowledge information K may be supplied to the generation device 30 using RAG (Retrieval-Augmented Generation). That is, the knowledge information K may be stored in a database searchable by RAG, and the generation device 30 may obtain the knowledge information K by accessing the database using RAG. This may improve the accuracy of the response compared to when instructions are given only from prompts.

[0093] [Second variation] In the above description, an example was given in which the control unit 120 supplies personality information indicating the user personality, a numerical value N for the user personality, knowledge information K, and natural language NL to the generation device 30 which has not learned about personality P. However, this embodiment is not limited to this example. The generation device 30 may have previously undergone reinforcement learning (e.g., reinforcement learning or fine-tuning) regarding the behavior of personality P. Figure 8 is a diagram illustrating the overview of the functions of the information processing device 10 according to the second modified example. In this case, the system 1 may further include a learning device 40. The learning device 40 uses personality information indicating the user personality, a numerical value N for the user personality, knowledge information K, and the input natural language NL as learning data, and uses generation examples as training data to cause the generation device 30 to learn the relationship between the behavior of personality P and the numerical value N. The information processing device 10 may supply personality information indicating the user personality, a numerical value N for the user personality, and natural language NL to the generation device 30 which has undergone reinforcement learning. Although the information processing device 10 and the learning device 40 have been described as separate devices, the functions of the learning device 40 can be realized by the information processing device 10. In other words, the information processing device 10 may be trained by outputting the necessary information to the generation device 30.

[0094] Furthermore, one generation device 30 may learn about the behavior of multiple personalities P, or it may learn about the behavior of one personality P. If a different generation device 30 is to be trained for each personality P, the learning device 40 does not need to supply personality information representing personality P as training data. Also, the information processing device 10 that instructs the generation device 30 to respond does not need to supply personality information representing personality P to the generation device 30. The information processing device 10 should determine the target generation device 30 based on the personality information representing the selected personality P, and then supply natural language NL and numerical value N to the determined generation device 30.

[0095] [Third variation] In the above description, the control unit 120 instructs the configured personality P to translate and, if necessary, to engage in dialogue. However, this embodiment is not limited to this example. The control unit 120 may instruct the configured personality P to perform any action, and is not limited to the example of instructing translation. For example, the control unit 120 may instruct dialogue from the beginning, or it may instruct the generation of a predetermined document, etc. In other words, user U can invoke the configured personality P for any purpose. Furthermore, although the system 1 was described as being used in an example where user U engages in online conversation, this embodiment is not limited to this example. For example, it may be used when user U is alone, or when they are in the same physical space.

[0096] Furthermore, this embodiment describes an example of interpreting a conversation between two people. However, this embodiment is not limited to this example. The information processing device 10 according to this embodiment can be used to correct the nuances of questions asked by a person or the nuances of answers given by artificial intelligence when a person is having a conversation (giving instructions or asking questions, etc.) with artificial intelligence. This makes it possible to realize a more natural conversation between a person and artificial intelligence.

[0097] [Summary of Embodiments] According to the embodiment described above, the information processing device 10 includes a natural language acquisition unit 113 that acquires natural language NL, a user personality acquisition unit 114 that acquires a user personality indicating the personality to be used from one or more personalities P whose behavior is predetermined by one or more characteristics, a numerical value acquisition unit 115 that acquires a numerical value N corresponding to the characteristics of the user personality, and a control unit 120 that outputs the acquired natural language NL and the numerical value N for the user personality together. In interpretation and response systems using large-scale language models, building a model that can provide highly accurate and natural responses requires setting a wide range of parameters according to the needs of the field. However, this setting often uses specialized expressions that are difficult for non-experts to understand. Also, because there is a wide range of parameters, the setting work was time-consuming and laborious. Therefore, the information processing device 10 according to the embodiment presents a personality P whose behavior direction is determined by multiple characteristics, thereby clearly communicating the tendencies of the desired response to the user U and reducing the burden of the setting work. Furthermore, the information processing device 10 according to this embodiment can appropriately understand and respond to the user U's intentions and nuances by allowing the user U to specify one or more characteristics that indicate the tendency of the personality P's responses.

[0098] Furthermore, according to the information processing device 10 of this embodiment, it is possible to provide a large-scale language model that requires language instructions with parametric information that directly reflects the intentions and nuances of the user U. This makes the response of a large-scale language model such as the generation device 30 more natural and brings it closer to the response desired by the user U.

[0099] Furthermore, according to the information processing device 10 of this embodiment, it is possible to appropriately convey the emotions and intentions of user U to a model that does not have parameters such as emotion values, and to bring the response closer to what user U desires. In addition, even for models that do have parameters such as emotion values, it is possible to convey user U's intentions more explicitly and to bring the response closer to those intentions.

[0100] Furthermore, according to the embodiment described above, the control unit 120 outputs the acquired natural language NL and the numerical value N for the user personality together, thereby causing the user personality, which has characteristics based on the numerical value N, to translate the acquired natural language NL. In other words, the information processing device 10 sets how the large-scale language model should behave when it is used and utilizes this for translation. Since machine translation is text-based, it is difficult to perform translations that reflect the intentions and nuances of the user U. Also, if there are expressions specific to a particular region or country, the translation accuracy tends to decrease. According to the information processing device 10 of the embodiment, a model capable of natural and accurate translation can be easily set. In this specification, "together" may mean that the natural language NL and the numerical value N are written together in a single system prompt SP, or that the natural language NL and the numerical value N are written together without distinguishing their locations.

[0101] Furthermore, according to the embodiment described above, the control unit 120 outputs the acquired natural language NL and a numerical value N about the user's personality together, causing the personality P, which has characteristics based on the numerical value N, to respond to the acquired natural language NL. In other words, the information processing device 10 sets how the large-scale language model behaves when responding and uses it as an AI for generating responses. There are tricks to the way in which natural language NL is written to be given to the large-scale language model, and it may be difficult for a user U with specialized knowledge to generate the desired response. With the information processing device 10, which sets the general behavior based on the personality P and allows intuitive setting of response tendencies based on characteristics, the user U can easily set a model that can output the desired response.

[0102] Furthermore, according to the above-described embodiment, the information processing device 10 further includes an operation reception unit 111 that receives a personality switching operation by user U, and the control unit 120 changes the personality P used for translation to another personality P corresponding to the personality switching operation and performs the translation. In other words, the information processing device 10 can change its behavior when translating based on the operation of user U. Furthermore, according to the above-described embodiment, the information processing device 10 includes an operation reception unit 111 that receives a personality switching operation by user U, and the control unit 120 changes the personality P used for dialogue to another personality P corresponding to the personality switching operation and has it respond to natural language NL. When the topic of conversation changes as it progresses, or when it is used for meetings on different topics, there may be a personality P that is more suitable for the changed topic than the personality P that was originally set. For example, if a meeting about a movie is held with a personality P that acts as a "film producer," and then the investment amount is negotiated, a personality P that acts as a "negotiator" would be more appropriate. In this case, by making it easy to change the person P performing the translation, it is possible to produce translations that appropriately reflect the user U's intentions in accordance with changes in the situation. Similarly, the properties required of a large-scale language model may differ depending on the progress of the conversation, etc. By making it easy to change the person P performing the dialogue, it is possible to easily present a model that aligns with the user U's intentions in accordance with changes in the required use.

[0103] Furthermore, according to the embodiment described above, the information processing device 10 further includes an operation reception unit 111 that receives a dialogue switching operation from user U, and the control unit 120 causes the personality P used for translation to respond to the natural language NL. In other words, the information processing device 10 can call up a model that has previously been solely for translation and engage in dialogue. This allows users to ask questions to a model that can accurately grasp their intentions when questions arise during a meeting or when new ideas are desired, and facilitates the meeting as if an additional person had joined.

[0104] Furthermore, according to the embodiment described above, the user personality acquisition unit 114 acquires the personality P used by each user U, and the control unit 120 outputs natural language NL, the user personality of each user U, and a numerical value N for each user personality, thereby creating a correspondence between the natural language NL and the combined personality based on the user personalities of each user U. That is, a combined personality is set for each user U from the personality P whose behavior has been set by multiple users U. Since this combined personality has characteristics based on the settings of each user U, it is easy to set a model in which responses such as listening, speaking, and responding are natural and appropriate for each user U.

[0105] Furthermore, according to the embodiment described above, a language is set for each of the multiple users U, and the responses from the user personalities are displayed according to the set language. That is, the response results from the set models are displayed to the first user U1 and the second user U2 in their respective languages. As a result, even users U who have not asked a question can easily understand the answer results, and the meeting can proceed smoothly as if they were participating alone.

[0106] [Example Hardware Configuration] Figure 9 is a schematic diagram of an example hardware configuration of an information processing device 90 applied to this embodiment. The information processing device 90 comprises a processor 91, main memory 92, communication interface 93, auxiliary storage device 94, input / output interface 95, and internal bus 96. The processor 91, main memory 92, communication interface 93, auxiliary storage device 94, and input / output interface 95 are connected to each other via the internal bus 96 so as to be able to communicate with each other. The information processing device 90 may also be applied to an information processing device 10. In this case, for example, the communication unit 110 may be configured using the communication interface 93. For example, the storage unit 13 may be configured using the auxiliary storage device 94. Also, the arithmetic unit 11 may be configured using the processor 91 and the main memory 92.

[0107] In this embodiment, the information processing device 10 and the user terminal 20 are configured as separate devices, but they may be configured as a single integrated device. Furthermore, each function performed in the information processing device 10 may be performed in the user terminal 20. In this case, the processing performed by the arithmetic unit 11 may be performed by the user terminal arithmetic unit. Also, various information stored in the storage unit 13 may be stored in the user terminal storage unit.

[0108] The information processing device 10 may be implemented using multiple information processing devices. For example, the information processing device 10 may be implemented using a device such as a cloud. For example, in the information processing device 10, the arithmetic unit 11 and the storage unit 13 may be implemented in different information processing devices. For example, the storage unit 13 may be distributed and implemented across multiple information processing devices.

[0109] Furthermore, the entirety or a part thereof of the functions of each part of the information processing device 10 in the above-described embodiment may be realized by recording a program for realizing these functions on a computer-readable recording medium, having a computer system read the program recorded on this recording medium, and executing it. The term "computer system" here includes hardware such as an operating system and peripheral devices.

[0110] Furthermore, "computer-readable recording media" refers to portable media such as flexible disks, magneto-optical disks, ROMs, and CD-ROMs, as well as recording units such as hard disks built into computer systems. In addition, "computer-readable recording media" may include those that dynamically hold programs for a short period of time, such as communication lines used when transmitting programs over networks such as the Internet or communication lines such as telephone lines, and those that hold programs for a certain period of time, such as volatile memory inside computer systems that act as servers or clients in such cases. Moreover, the above-mentioned program may be for the purpose of realizing some of the functions described above, and may also be able to realize the above-mentioned functions in combination with programs already recorded in the computer system.

[0111] Although one embodiment of this invention has been described in detail above with reference to the drawings, the specific configuration is not limited to that described above, and various design changes can be made without departing from the spirit of this invention. Furthermore, the configurations described in each embodiment and example above may be combined. [Explanation of Symbols]

[0112] 1...System, 10...Information Processing Device, 20...User Terminal, 30...Generator, 11...Calculation Unit, 13...Storage Unit, 110...Communication Unit, 111...Operation Reception Unit, 112...Display Control Unit, 113...Natural Language Acquisition Unit, 114...User Personality Acquisition Unit, 115...Numerical Value Acquisition Unit, 116...Knowledge Information Acquisition Unit, 120...Control Unit, 121...Personality Setting Unit, 122...Translation Instruction Unit, 123...Dialogue Instruction Unit, NL...Natural Language, P...Personality, N...Numerical Value, K...Knowledge Information, SP...System Prompt, U...User

Claims

1. A natural language acquisition unit that acquires natural language, A servant personality acquisition unit that acquires a servant personality that indicates the personality to be used from one or more personalities whose characteristics are predetermined and whose behavior is set in order, A numerical acquisition unit that acquires numerical values ​​corresponding to the characteristics possessed by the aforementioned user personality, A control unit that outputs the acquired natural language and the numerical values ​​for the user's personality, An information processing device equipped with the following features.

2. The control unit outputs the acquired natural language and the numerical values ​​for the user personality together, thereby causing the user personality having the characteristics based on the numerical values ​​to translate the acquired natural language. The information processing apparatus according to claim 1.

3. It is further equipped with an operation reception unit that accepts user-initiated personality switching operations, The control unit changes the personality used for translation to another personality corresponding to the personality switching operation and performs the translation. The information processing apparatus according to claim 2.

4. It further includes an operation reception unit that accepts user-initiated switching operations, The control unit causes the personality used for translation to respond to the natural language. The information processing apparatus according to claim 2.

5. The control unit outputs the acquired natural language and the numerical values ​​for the user personality together, thereby causing the personality having the characteristics based on the numerical values ​​to respond to the acquired natural language. The information processing apparatus according to claim 1.

6. It is further equipped with an operation reception unit that accepts user-initiated personality switching operations, The control unit changes the personality used for the dialogue to another personality corresponding to the personality switching operation and has the personality respond to the natural language. The information processing apparatus according to claim 5.

7. The aforementioned user personality acquisition unit acquires the aforementioned personalities used by each user, The control unit outputs the natural language, the servant persona of each user, and the numerical value for each of the servant personas, thereby causing the combined persona based on each of the servant personas of each user to correspond to the natural language. The information processing apparatus according to any one of claims 1 to 6.

8. The language used is set for each of the multiple users. The response from the aforementioned user personality is displayed according to the set language used. The information processing apparatus according to any one of claims 4 to 6.

9. An information processing device comprising: a natural language acquisition unit for acquiring natural language; a user personality acquisition unit for acquiring a user personality that indicates the user personality to be used from one or more personalities whose behaviors are predetermined to have one or more characteristics; a numerical value acquisition unit for acquiring numerical values ​​corresponding to the characteristics possessed by the user personality; and a control unit for outputting the acquired natural language and the numerical values ​​for the user personality together. A generating device that operates based on the information output from the aforementioned information processing device, A system equipped with these features.

10. On the computer, The natural language acquisition step for obtaining natural language, A servant personality acquisition step in which one or more characteristics are obtained from one or more personalities that exhibit predetermined behaviors to acquire a servant personality, A numerical acquisition step of acquiring numerical values ​​corresponding to the characteristics possessed by the aforementioned user personality, A control step that outputs the acquired natural language and the numerical values ​​for the user's personality together, A program that executes the command.

Images