WO2024024065A1

WO2024024065A1 - Sensitivity transformation method, sensitivity transformation device, and sensitivity transformation program

Info

Publication number: WO2024024065A1
Application number: PCT/JP2022/029208
Authority: WO
Inventors: 陽子 ▲徳▼永; 巌樹戸嶋; 史朗小澤
Original assignee: 日本電信電話株式会社
Priority date: 2022-07-28
Filing date: 2022-07-28
Publication date: 2024-02-01

Abstract

A storage unit (14) stores history data (14a) in which information indicating a past situation of a speaker and an interlocutor is associated with an expression expressing sensitivity in that situation. An extraction unit (15b) extracts an expression expressing sensitivity from a phrase uttered by the speaker. An estimation unit (15c) uses the history data (14a) to estimate information expressing a situation corresponding to the extracted expression expressing sensitivity. An identification unit (15d) uses the history data (14a) to identify an expression expressing the sensitivity of the interlocutor corresponding to the estimated information expressing a situation.

Description

Kansei conversion method, kansei conversion device, and kansei conversion program

The present invention relates to a sensibility conversion method, a sensibility conversion device, and a sensibility conversion program.

Conventionally, there is a known technology for converting linguistic expressions in dialogue to suit the other party. For example, there is a known technique such as multilingual translation, which converts text from one language to another while preserving its meaning (see Patent Document 1). Furthermore, a technique is known for converting a certain text into an expression that can be more empathetic (see Non-Patent Document 1).

Japanese Patent Application Publication No. 2021-039501

However, with the conventional technology, it is difficult to resolve the gap in understanding and recognition between the speaker and the receiver in conversational communication. For example, the "good feeling" said by the speaker and the "good feeling" felt by the receiver may differ in degree. This discrepancy in understanding and recognition is thought to be due to the difference in the sensibilities of the speaker and the receiver. Sensitivity differs from person to person and is formed by various factors such as a person's innate personality, past experiences, and relationships with others. Differences in understanding and recognition are a cause of miscommunication, and can become a major problem in conversations such as work meetings and casual chats.

The present invention has been made in view of the above, and aims to eliminate the gap in understanding and recognition between the speaker and the interlocutor in conversational communication.

In order to solve the above-mentioned problems and achieve the purpose, the sensibility conversion method according to the present invention is a sensibility conversion method performed by a sensibility conversion device, wherein the sensibility conversion device The method includes a storage unit that stores historical data in which information representing a situation is associated with an expression representing a sensitivity in the situation, and an extraction step for extracting an expression representing a sensitivity from a sentence uttered by a speaker, and using the historical data. an estimation step of estimating information representing a situation corresponding to the extracted expression representing the sensibility, and an expression representing the sensibility of the interlocutor corresponding to the information representing the estimated situation using the historical data; A specific step of specifying.

According to the present invention, it is possible to eliminate the gap in understanding and recognition between the speaker and the interlocutor in communication through dialogue.

FIG. 1 is a schematic diagram illustrating a schematic configuration of a sensitivity conversion device. FIG. 2 is a diagram illustrating the data structure of attribute data. FIG. 3 is a diagram for explaining the processing of the sensitivity conversion device. FIG. 4 is a flowchart showing the sensibility conversion processing procedure. FIG. 5 is a diagram illustrating a computer that executes the sensitivity conversion program.

Hereinafter, one embodiment of the present invention will be described in detail with reference to the drawings. Note that the present invention is not limited to this embodiment. In addition, in the description of the drawings, the same parts are denoted by the same reference numerals.

[Overview of sensitivity conversion device]
In verbal communication, the sensibility conversion device of this embodiment is capable of inputting a textual version of the speaker's own sensibilities in a certain situation, so that the recipient has the same or similar sensibilities as the speaker. Convert and output to text.

The sensibility conversion device is used, for example, to express the sensibility that a speaker feels about the situation in a certain situation. In addition, the sensibility conversion device can be used to predict the future situation and express the sensibility felt about the predicted future situation when the situation changes due to some trigger or in chronological order. good.

For example, when you are playing a game with multiple people as a team, you may express your feelings about the current situation of the game, or you may predict future actions of your own team or a rival team that may occur in the future. It can also be used to express what you feel about something. Or, when watching sports with multiple people, you may want to express your feelings about the current situation, or imagine the next attack that your supporters or opponent might take, and talk about possible future situations of the situation. It can also be used to describe what you feel.

Here, the expression expressing sensibility (hereinafter also referred to as sensibility expression) is a verbal expression of the sensibility felt by the speaker based on the input situation or trigger. For example, expressions such as "Oh no, I think I'm going to lose", which are expressed based on the match situation and the opponent's attack, fall under this category. Even if people have the same sensibilities, the way they express their sensibilities in words may differ depending on the person. For example, a bearish person often uses words like ``I'm going to lose'' when we're even slightly behind, while a bullish person often says ``I think I'll be fine'' when we're even slightly behind, and even when we're one step closer to losing, we say ``I think I'm going to lose.'' shall not be said. In this case, if you interact without knowing the other person's personality, even if the weak person says, ``I think I'm going to lose,'' the bullish person will not be able to understand why he feels so inferior, leading to miscommunication. There is a fear.

Therefore, in order to reduce such miscommunication, the sensitivity conversion device can, for example, tell a bullish person, ``What the other person just said, ``I think I'm going to lose,'' means to you, ``I think it's going to be okay.'' teach. This will lead to understanding of others and diversity, such as being able to understand the differences in how the situation is perceived, the range of language expression, and the other person's personality, and it will be possible to reduce miscommunication.

The sensibility conversion device described below is applied to a situation in which the content of the dialogue is shogi, and players a and b of a team of two are thinking about their next move while having a dialogue with each other. Note that, for example, it is assumed that the dialogue within the team cannot be heard by the opponent, such as in remote shogi using a computer or with an AI as the opponent.

[Configuration of sensitivity conversion device]
FIG. 1 is a schematic diagram illustrating a schematic configuration of a sensitivity conversion device. As illustrated in FIG. 1, the sensitivity conversion device 10 of this embodiment is realized by a general-purpose computer such as a personal computer, and includes an input section 11, an output section 12, a communication control section 13, a storage section 14, and a control section 15. .

The input unit 11 is realized using an input device such as a keyboard or a mouse, and inputs various instruction information such as starting processing to the control unit 15 in response to input operations by the user. The output unit 12 is realized by a display device such as a liquid crystal display, a printing device such as a printer, an information communication device, and the like. The communication control unit 13 is realized by a NIC (Network Interface Card) or the like, and controls communication via a network between the control unit 15 and external devices such as a server or a management device that manages various data.

The storage unit 14 is realized by a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. Note that the storage unit 14 may be configured to communicate with the control unit 15 via the communication control unit 13. In the present embodiment, the storage unit 14 stores, for example, history data 14a used in the sensibility conversion process described later, a first model 14b, a second model 14c, etc. that are generated and updated in the sensibility conversion process. Ru.

Here, FIG. 2 is a diagram illustrating the data structure of history data. As shown in FIG. 2, the history data 14a associates information representing past situations of the speaker and the interlocutor with expressions representing sensibilities in the situations. In FIG. 2, information such as the board (winning rate), the board (number of moves), the board (pieces held), etc. is recorded as the past situation of shogi. In addition, emotional expressions representing the sensibilities of players a and b, such as "scary" and "dangerous," corresponding to these situations are also associated.

Furthermore, in the history data 14a, the information representing the situation may be information that changes over time, for example. Alternatively, the information representing the situation may be information that changes depending on the trigger for changing the situation. Further, as the information representing the situation, an evaluation value by AI may be used.

For example, in games and sports, the situation is exemplified by points gained or lost, and the trigger for changing the situation is exemplified by the opponent's attack pattern. The situation and the trigger for changing the situation represent the environment that caused the speaker to feel something in their brain, and the speaker is thought to develop a feeling based on this and verbalize this into speech. It will be done. The speaker's utterances may be input in advance as text as in a text chat, or may be voice utterances that are recognized or transcribed manually.

Here, FIG. 3 is a diagram for explaining the processing of the sensitivity conversion device. The sensibility conversion device 10 of this embodiment is configured such that, in the current situation x of shogi, a player a hugs the player immediately after the opponent makes the next move y such as "36 steps" as an opportunity to change the situation x. The obtained sensibility is converted into the sensibility of player b and is conveyed to player b. In the example shown in FIG. 3, player a's emotional expression "scary" is converted into player b's emotional expression "quite dangerous."

The current situation of shogi x represents the current situation that is objectively observable, and includes the win rate value evaluated by AI on the current score, the position of the pieces, the type and number of pieces held, the game record so far, etc. It is expressed as each value or a vector that combines multiple values.

Returning to the explanation of FIG. 1. The control unit 15 is implemented using a CPU (Central Processing Unit), an NP (Network Processor), an FPGA (Field Programmable Gate Array), etc., and executes a processing program stored in a memory. Thereby, the control unit 15 functions as an acquisition unit 15a, an extraction unit 15b, an estimation unit 15c, a specification unit 15d, and a conversion unit 15e, as illustrated in FIG. Note that these functional units may be implemented in different hardware. For example, the acquisition unit 15a may be implemented in hardware different from other functional units. Further, the control unit 15 may include other functional units.

The acquisition unit 15a acquires an uttered sentence that verbalizes the speaker's sensibilities. For example, the acquisition unit 15a acquires an utterance t _a that verbalizes the feelings felt by the player a.

The extracting unit 15b extracts expressions representing sensibilities (sensitivity expressions) from sentences uttered by the speaker. Specifically, the extraction unit 15b divides the uttered sentence into words by morphological analysis, and extracts emotional expressions from the words. For example, the extraction unit 15b may refer to a pre-built dictionary of words as emotional expressions and extract words corresponding to the dictionary as emotional expressions. Alternatively, the extraction unit 15b may extract adjectives that often express the degree or feeling of something as emotional expressions.

Specifically, the extraction unit 15b outputs the emotional expression _{wa, x, y} of the hand y in the situation x by the player a from the input utterance _ta .

Note that the extraction unit 15b may transfer all the words divided by morphological analysis to the estimation unit 15c, which will be described below, without extracting emotional expressions. In that case, the estimation unit 15c may process only the emotional expression.

The estimation unit 15c uses the history data 14a to estimate information representing the situation corresponding to the extracted emotional expression.

For example, the estimation unit 15c uses the first model 14b trained using the history data 14a to output information representing the corresponding situation when the emotional expression of the speaker is input. The information representing the situation corresponding to the emotional expression is estimated.

In other words, the estimation unit 15c uses the first model 14b learned using the history data 14a indicating what kinds of utterances the speaker has made in what situations in the past, to estimate the input emotional expression. Estimate what kind of situation the speaker is describing with respect to a word or word group.

The identification unit 15d uses the history data 14a to identify the emotional expression of the interlocutor that corresponds to the information representing the estimated situation.

For example, when information representing a situation is input, the identifying unit 15d uses the second model 14c learned using the history data 14a to output the emotional expression of the interlocutor in the situation. Identify the emotional expression of the interlocutor that corresponds to the information representing the estimated situation.

In other words, the identification unit 15d uses the second model 14c learned using the history data 14a indicating what kind of utterances the recipient (interlocutor) has made in what situations in the past. Predict how the recipient's interlocutor would verbalize the situation using words or word groups.

In the sensibility conversion device 10 of the present embodiment, the history data 14a includes a set of board conditions and the player's sensibility expression for the next move for past games between players a and b. In the example shown in FIG. 2, the value representing the situation is an evaluation value such as a winning rate evaluated by the AI on the current board and expressed as a numerical value of 0-1. In this embodiment, the opponent is AI, but the AI that calculates the evaluation value and the opponent AI may be the same or different.

The evaluation value p(y|x) of the next move y in a certain situation x is the same value in the same situation, regardless of the player. However, in reality, depending on the player's personality, values, preferred playing style, school of shogi, cultural background, etc., there are moves that the player likes, moves that he dislikes, and moves that cannot be considered as next candidates. Therefore, the evaluation value P(y|x, a) weighted for each player may be used as the evaluation value calculated by AI. For example, for player a who prefers attack-oriented tactics, if the board is easy to attack, a certain number is added to p(y|x), and if the board is difficult to attack, a decimal less than 1 is added to p(y|x). Examples include methods such as multiplying by .

The evaluation value may be calculated for each board in a past game played by player a, taking into account the evaluation value annotated by the player himself or the final game result. A function M _shogi,a (x, y) that outputs an evaluation value by inputting the board x and the hand y may be defined, and this may be calculated as the evaluation value P(y|x, a). Alternatively, instead of calculating for each player, the evaluation value may be calculated using a function M _shogi,all (x, y) for all players for which game data exists. However, by calculating M _shogi,player (x, y) as an evaluation value from the experience of each player, it becomes possible to express more detailed differences in sensibilities of each player using different linguistic expressions.

In addition, the spaces of players a and b are formed from each player's past game experience, so the data is sparse, and the probability that the same situation x will appear again is low, so the evaluation value is calculated. It may not be possible. Therefore, for example, personas may be defined based on user attributes such as school, level, years of experience, age, etc., and evaluation values M _shogi,a~ (x, y) may be defined for players grouped by persona. This makes it possible to calculate the evaluation value using the history data 14a of past games of people who have similar sensibilities to player a. Furthermore, if there is no matching situation x in the past, a value with a small influence, such as the median value of possible numerical values, may be used as the evaluation value.

Then, in the sensitivity conversion process, the emotional expression of player a regarding move y in situation x is converted into text that allows player b to have the same or similar sensitivity as player a.

Here, the emotional expression by player a is expressed by the following equation (1) using a different mapping function F for each player.

The function F _a takes as input the winning probability P of player a in move y in situation x, and obtains the linguistic expression f _a by player a. In other words, F _a is equivalent to defining the process in which player a inputs information about the current situation in his brain and verbalizes what he feels as a function.

Furthermore, the emotional expression by player b is expressed by the following equation (2).

The function F _b takes as input the winning probability P' of player b in move y in situation x, and obtains the linguistic expression f _b by player b. In other words, F _b is equivalent to defining the process of verbalizing what player b feels in his or her brain as a function.

Here, the inverse function F _a ^-1 of F _a can be regarded as a function that returns the verbalized emotional expression to the original situation that caused player a to have that emotion. Therefore, inputting the emotional expression of player a to this function F _a ^-1 can be expressed as F _a ^-1 (wa _{, x, y} ). This can be said to represent the situation that is the source of player a's emotional expression.

Inputting this situation into a function F _b representing the process in the brain of player b can be expressed as F _b (F _a ^-1 (wa _{, x, y} )). This can be said to be an emotional expression in which when the situation perceived by player a is input into player b's brain, player b verbalizes that emotion. Therefore, the following equation (3) holds for the text converted into the emotional expression of player b.

In this way, the emotional expression uttered by player a regarding situation x and move y can be input and converted into an emotional expression that allows player b to feel the same emotion.

Here, F _a ⁻¹ and F _b are modeled by machine learning using the historical data 14a. In other words, F _a ^-1 is the first model 14b that has been trained to receive the emotional expression of player a as input and output an evaluation value P(y|x, a) representing the situation. In addition, F _b is trained to receive the evaluation value P'(y|x, b) of a situation experienced by player b in the past and output the emotional expression w _{b, x, y} uttered in that situation. This is the second model 14c.

The machine learning method is not particularly limited. In addition, the emotional expression input to F _a ^-1 may be just a Bag-of-Words expression, or it may be a numerical value or vector expressing the meaning of a word, a numerical value representing positive/negative, etc., which can be used as a sensory expression. It may also include variables that explain the meaning and usage of the word.

The conversion unit 15e converts the uttered sentence using an expression expressing the sensibilities of the identified interlocutor. For example, only the part of the sentence uttered by the speaker that has been converted as an emotional expression is rewritten so that the grammar is correct. That is, the conversion unit 15e formats the emotional expression output from the identification unit 15d so as not to damage the original context, and presents it to player b.

At that time, the conversion unit 15e may, for example, replace a part of the original uttered sentence with the converted emotional expression and present it so that the grammar becomes natural. Alternatively, the conversion unit 15e may display the converted emotional expression in parallel to the emotional expression portion of the original utterance. Further, the conversion unit 15e may present the converted emotional expression in real time, or may execute the emotional conversion process in advance and present it as feedback when reviewing later.

[Emotional conversion processing]
Next, the sensibility conversion process by the sensibility conversion device 10 will be explained. FIG. 4 is a flowchart showing the sensibility conversion processing procedure. The flowchart in FIG. 4 is started, for example, at the timing when there is an input instructing the start of processing.

First, the acquisition unit 15a acquires an uttered sentence that verbalizes the speaker's sensibilities. Further, the extracting unit 15b extracts an expression representing sensitivity (sensitivity expression) from the utterance (step S1). Specifically, the extraction unit 15b divides the utterance into words by morphological analysis, and extracts emotional expressions from the words.

Next, the estimation unit 15c uses the history data 14a to estimate information representing the situation corresponding to the extracted emotional expression (step S2). For example, the estimation unit 15c uses the first model 14b trained using the history data 14a to output information representing the corresponding situation when the emotional expression of the speaker is input. The information representing the situation corresponding to the emotional expression is estimated.

Furthermore, the identification unit 15d uses the history data 14a to identify the emotional expression of the interlocutor that corresponds to the information representing the estimated situation (step S3). For example, when information representing a situation is input, the identifying unit 15d uses the second model 14c learned using the history data 14a to output the emotional expression of the interlocutor in the situation. Identify the emotional expression of the interlocutor that corresponds to the information representing the estimated situation.

In other words, the identification unit 15d uses the second model 14c learned using the historical data 14a indicating what kind of utterances the receiver (interlocutor) has made in what situations in the past. Predict how the recipient's interlocutor would verbalize the situation using words or word groups.

Then, the conversion unit 15e converts the uttered sentence using an expression expressing the identified sensibilities of the interlocutor (step S4). For example, only the part of the sentence uttered by the speaker that has been converted as an emotional expression is rewritten so that the grammar is correct. That is, the conversion unit 15e formats the emotional expression output from the identification unit 15d so as not to damage the original context, and presents it to the interlocutor. This completes the series of sensibility conversion processes.

[effect]
As described above, in the sensibility conversion device 10 of the present embodiment, the storage unit 14 stores history data in which information representing past situations of the speaker and the interlocutor is associated with expressions representing sensibilities in the situation. Remember. Further, the extracting unit 15b extracts expressions representing sensibilities from the utterances by the speaker. Furthermore, the estimation unit 15c estimates information representing a situation corresponding to the extracted expression representing the sensitivity, using the history data 14a. Further, the identification unit 15d uses the history data 14a to identify an expression representing the sensibility of the interlocutor that corresponds to the information representing the estimated situation.

Specifically, the estimation unit 15c uses the first model trained using the history data 14a so that when an expression representing the speaker's sensibilities is input, information representing the corresponding situation is output. 14b is used to estimate information representing the situation corresponding to the extracted expression representing sensitivity. Further, the identifying unit 15d uses the second model 14c learned using the historical data 14a so as to output an expression representing the sensibilities of the interlocutor in the situation when information representing the situation is input. Then, an expression representing the sensibility of the interlocutor corresponding to the information representing the estimated situation is identified.

In this way, the sensibility conversion device 10 collects data on the experience of the speaker/interlocutor in a dialogue, the experience of personas with similar sensibilities, and the current situation such as evaluation values using AI in competitive games such as shogi. The sensibility expressed by the speaker is converted into an expression that allows the interlocutor to feel the same way.

This makes it possible for the interlocutor to more accurately capture what the speaker feels about the same situation, and it becomes possible to communicate without causing a gap in understanding or recognition. For example, when formulating a strategy to win a game, by aligning each other's understanding of the situation, it becomes possible to look at each other's eyes and communicate efficiently. In addition, the winning rate in games will increase if the person responsible for accurately capturing the verbal expressions spoken by the players and thinking about the team's strategy faithfully captures the players' sensibilities and reflects them in the strategy. In this way, according to the sensibility conversion device 10, it is possible to eliminate the gap in understanding and recognition between the speaker and the interlocutor in communication through dialogue.

Additionally, the information representing the situation may be information that changes over time. Alternatively, the information representing the situation may be information that changes depending on the trigger for changing the situation. Thereby, the sensibility conversion device 10 can convert the sensibility expression even when the situation changes.

Furthermore, an evaluation value obtained by AI may be used as the information representing the situation. In this way, the situation can be simply expressed by using objective evaluation values such as AI as information representing the situation. Furthermore, if the situation is such that an evaluation value can be obtained by AI, it becomes possible to convert the sensibilities even in different scenes. For example, if AI evaluation values are obtained for both shogi and sports, it becomes possible to convert words said in shogi to words said during a sports match.

Furthermore, the conversion unit 15e converts the uttered sentence using an expression that expresses the sensibilities of the identified interlocutor. This makes it possible for the interlocutor to accurately grasp the emotional expression of the speaker and to eliminate any discrepancies in understanding and recognition between the speaker and the speaker.

[program]
It is also possible to create a program in which the processing executed by the sensibility conversion device 10 according to the embodiment described above is written in a computer-executable language. As one embodiment, the sensibility conversion device 10 can be implemented by installing a sensibility conversion program that executes the above-mentioned sensibility conversion process into a desired computer as packaged software or online software. For example, by causing the information processing device to execute the above-mentioned sensibility conversion program, the information processing device can be made to function as the sensibility conversion device 10. In addition, information processing devices include mobile communication terminals such as smartphones, mobile phones, and PHSs (Personal Handyphone Systems), as well as slate terminals such as PDAs (Personal Digital Assistants). Further, the functions of the sensibility conversion device 10 may be implemented in a cloud server.

FIG. 5 is a diagram showing an example of a computer that executes the sensibility conversion program. Computer 1000 includes, for example, memory 1010, CPU 1020, hard disk drive interface 1030, disk drive interface 1040, serial port interface 1050, video adapter 1060, and network interface 1070. These parts are connected by a bus 1080.

The memory 1010 includes a ROM (Read Only Memory) 1011 and a RAM 1012. The ROM 1011 stores, for example, a boot program such as BIOS (Basic Input Output System). Hard disk drive interface 1030 is connected to hard disk drive 1031. Disk drive interface 1040 is connected to disk drive 1041. A removable storage medium such as a magnetic disk or an optical disk is inserted into the disk drive 1041, for example. For example, a mouse 1051 and a keyboard 1052 are connected to the serial port interface 1050. For example, a display 1061 is connected to the video adapter 1060.

Here, the hard disk drive 1031 stores, for example, an OS 1091, an application program 1092, a program module 1093, and program data 1094. Each piece of information described in the above embodiments is stored in, for example, the hard disk drive 1031 or the memory 1010.

Further, the sensibility conversion program is stored in the hard disk drive 1031, for example, as a program module 1093 in which commands to be executed by the computer 1000 are written. Specifically, a program module 1093 in which each process executed by the sensibility conversion device 10 described in the above embodiment is described is stored in the hard disk drive 1031.

Further, data used for information processing by the sensibility conversion program is stored as program data 1094 in, for example, the hard disk drive 1031. Then, the CPU 1020 reads out the program module 1093 and program data 1094 stored in the hard disk drive 1031 to the RAM 1012 as necessary, and executes each of the above-described procedures.

Note that the program module 1093 and program data 1094 related to the sensitivity conversion program are not limited to being stored in the hard disk drive 1031; for example, they may be stored in a removable storage medium and read by the CPU 1020 via the disk drive 1041 or the like. May be served. Alternatively, the program module 1093 and program data 1094 related to the sensitivity conversion program are stored in another computer connected via a network such as a LAN (Local Area Network) or a WAN (Wide Area Network), and are transmitted via the network interface 1070. The data may also be read out by the CPU 1020.

Although embodiments to which the invention made by the present inventor is applied have been described above, the present invention is not limited by the description and drawings that form part of the disclosure of the present invention by this embodiment. That is, all other embodiments, examples, operational techniques, etc. made by those skilled in the art based on this embodiment are included in the scope of the present invention.

For example, in the above embodiment, a game was used as an example to clarify the issue, but communication with a history that generally involves consensus building can be modeled using a similar framework. Instead of AI, it is possible to use the degree of match with the company's decision-making policy, the selection tendencies of general people, etc.

10 Sensitivity conversion device 11 Input section 12 Output section 13 Communication control section 14 Storage section 14a History data 14b First model 14c Second model 15 Control section

15a Acquisition section

15b Extraction section

15c Estimation section 15d Specification section 15e Conversion section

Claims

A sensibility conversion method executed by a sensibility conversion device, the method comprising:
The sensibility conversion device has a storage unit that stores history data in which information representing past situations of the speaker and the interlocutor is associated with expressions representing sensibilities in the situation;
an extraction step of extracting expressions expressing sensitivity from sentences uttered by the speaker;
an estimation step of estimating information representing a situation corresponding to the extracted expression representing the sensitivity using the historical data;
a specifying step of using the historical data to identify an expression representing the sensibility of the interlocutor that corresponds to information representing the estimated situation;
A sensibility conversion method characterized by comprising:
The estimation step uses the first model trained using the historical data to output information representing a corresponding situation when an expression representing the speaker's sensibility is input. estimate information representing the situation corresponding to the expression representing the sensitivity,
The identification step includes estimation using a second model trained using the historical data so that when information representing the situation is input, an expression representing the sensibility of the interlocutor in the situation is output. identifying an expression representing the sensibility of the interlocutor that corresponds to information representing the situation
The sensibility conversion method according to claim 1, characterized in that:
The sensibility conversion method according to claim 1, wherein the information representing the situation is information that changes over time.
The sensibility conversion method according to claim 1, wherein the information representing the situation is information that changes depending on a trigger for changing the situation.
The sensibility conversion method according to claim 1, wherein the information representing the situation uses an evaluation value obtained by AI.
The sensibility conversion method according to claim 1, further comprising a conversion step of converting the uttered sentence using an expression representing the specified sensibility of the interlocutor.
a storage unit that stores history data that associates information representing past situations of the speaker and the interlocutor with expressions representing sensibilities in the situation;
an extraction unit that extracts an expression expressing sensitivity from a sentence uttered by a speaker;
an estimation unit that uses the historical data to estimate information representing a situation corresponding to the extracted expression representing the sensitivity;
an identification unit that uses the history data to identify an expression representing the sensibility of the interlocutor that corresponds to information representing the estimated situation;
A sensitivity conversion device characterized by having the following.
An affective conversion program for causing a computer to execute the affective converting method according to any one of claims 1 to 6.