Digital human personalized emotion interaction method

Abstract

The invention relates to a digital human personalized emotion interaction method, which belongs to the technical field of emotion calculation, and comprises the following steps: S1, carrying out emotion quantification on interaction input of a user to obtain a pleasure-wake-dominate PAD emotion value in a continuous dimension space; s2, performing emotion state evaluation on the user interaction input emotion to obtain an interaction input emotion evaluation function; s3, calculating a digital human emotion transition probability matrix by using a Bayesian network; s4, calculating an emotional response calculation value of the digital human character corrected by the big five-factor model in the current emotional state; s5, the corrected emotion response passes through an emotion influence factor matrix, and a response with complex emotion features is output; and S6, performing PAD emotional space coordinate calibration on the emotional intensity of the digital person on the six basic emotions to obtain an emotional response of the digital person.

Description

Technical Field

[0001] The present invention belongs to the technical field of emotional computing and relates to a method for personalized emotional interaction of digital humans. Background Art

[0002] With the continuous development of technologies such as virtual reality, artificial intelligence, human-computer interaction, and AI big models, the vision of the metaverse has become clearer. People have begun to imagine a virtual world parallel to the real world. In this future world, even if you stay at home, you can experience various activities such as travel, social interaction, entertainment, shopping, and learning in the metaverse through a "digital avatar". Therefore, digital humans have gradually become a hot topic in the field of human-computer interaction. At this stage, mainstream digital human technology mainly focuses on modeling and rendering the appearance of characters. However, from the perspective of interaction, it is still difficult for current digital humans to actively communicate and interact like real people, especially in terms of emotions and personality. The reason for this phenomenon is that human appearance is relatively easy to be objectively captured and modeled, while emotions and psychology are difficult to accurately measure, calculate, and simulate digitally. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a digital human personalized emotional interaction method.

[0004] In order to achieve the above object, the present invention provides the following technical solutions:

[0005] A digital human personalized emotional interaction method, comprising the following steps:

[0006] S1: Quantify the user's interactive input and obtain the pleasure-arousal-dominance (PAD) emotion value in the continuous dimensional space;

[0007] S2: Evaluate the emotional state of the user's interactive input emotion to obtain an interactive input emotion evaluation function;

[0008] S3: Use Bayesian network to calculate the digital human emotion transfer probability matrix;

[0009] S4: Calculate the emotional response value of the digital person's personality after being corrected by the Big Five Factor Model in the current emotional state;

[0010] S5: The corrected emotional response is processed through the emotional influence factor matrix to output a response with complex emotional characteristics;

[0011] S6: Calibrate the PAD emotional space coordinates of the digital human's emotional intensity in six basic emotions to obtain the digital human's emotional response.

[0012] Further, step S2 specifically includes the following steps:

[0013] Inputting sentiment values ​​into user interactions The evaluation is converted into the emotional state vector E under six basic emotional states l , interactive input emotion To basic emotion E l The distance between them is:

[0014]

[0015] Among them, C l is the covariance matrix set of basic emotional states in the PAD space; l = 1, 2, 3, 4, 5, 6 represent the six basic emotions of happiness, surprise, disgust, anger, fear, and sadness respectively;

[0016] The distance from the interactive input emotion to the coordinate value of each emotional state is inversely proportional to the intensity value of the emotional state effect. l Normalize to get the basic emotional state l to interact with the input emotion The intensity value i l , define the interactive input sentiment evaluation function for:

[0017]

[0018] Among them, h l Indicates that the location of the emotional state is considered to be in C l Interactive input sentiment obtained under constraints To basic emotion E l The distance between.

[0019] Further, step S3 specifically includes the following steps:

[0020] S31: Select the user's personality, historical emotional state, gender, age and education level as nodes of the Bayesian network to build a Bayesian model describing emotional transfer:

[0021] p(E j |E i ,P,G,A,C)

[0022] Where p(x) represents E i Transfer to E j The transition probability of the state; P represents personality type; G represents gender; A represents age group; C represents education level;

[0023] S32: define the state space of the node;

[0024] S33: Constructing a conditional probability table, including the following steps:

[0025] S331: Calculate the prior probability p(E i ), p(P), p(G), p(A), p(C);

[0026] S332: For each input feature, calculate the conditional probability p(E j |E i ,P,G,A,C);

[0027] S333: Organize all calculated conditional probabilities into a digital human emotion transfer probability matrix V, where each row in the matrix V corresponds to the conditional probability of a child node under a specific parent node combination.

[0028] Furthermore, the state space of each node includes:

[0029] Historical emotional states: joy, sadness, anger, surprise, fear, disgust;

[0030] Personality: openness, conscientiousness, extraversion, agreeableness, neuroticism;

[0031] Age groups: children (0-12 years old), adolescents (13-18 years old), adults (19-65 years old), and the elderly (over 65 years old);

[0032] Educational level: primary school and below, junior high school, high school, university, graduate school and above.

[0033] Further, step S4 specifically includes the following steps:

[0034] The five factor model (FFM) is used as a personality model, including openness, responsibility, extraversion, altruism and neuroticism. In the five factor model, an individual is represented by a five-dimensional vector P = [p E ,p A ,p N ,p O ,p C ], where p E ,p A ,p N ,p O ,p C Respectively represent their values ​​in the five personality domains, with values ​​ranging from 0 to 1;

[0035] Map each personality in the Big Five Factor Model to a specific PAD value, expressed as:

[0036]

[0037] The above formula is used to correct the PAD three-dimensional emotional state vector of a person. The formula is as follows:

[0038]

[0039] In the formula, The average score of each dimension obtained from the Big Five Factor Table (0.70, 0.73, 0.61, 0.60, 0.60) is used as the basis to define the neutral level of personality; is the character correction vector; e f is the initial emotional component; e m It is the emotional state vector obtained after personality correction.

[0040] Further, step S5 specifically includes the following steps:

[0041] The formula for the mutual influence between emotions is as follows:

[0042] φ(E f )=E f *λ

[0043] The influence factor matrix λ between emotions is:

[0044]

[0045] Among them, λ ij =[-1,1] represents the influence factor of emotion i on emotion j. The mutual influence between emotions includes three situations:

[0046] Incentive: When λ ij When > 0, it means that emotion i has a motivating effect on emotion j;

[0047] Suppression: When λ ij When <0, it means that emotion i has an inhibitory effect on emotion j;

[0048] No effect: When λ ij =0, it means that emotion i has no effect on emotion j;

[0049] According to the above formula, the emotional response vector is corrected to make the digital human have the characteristics of complex human emotions. The correction calculation formula is:

[0050] E c =φ(E f )

[0051] Where: E f is to convert the e in the PAD dimension emotion space f Emotional state vector converted to discrete emotion space; E c It is the emotional state vector obtained after being corrected by the emotional influence factor matrix.

[0052] Further, in step S6, the formula for converting the emotional states under the six basic emotions into the PAD space and calibrating the positions is:

[0053]

[0054] Among them, p l ,a l ,d l Represents the emotional values ​​of the six basic emotions in the PAD space, i l It is the intensity value of the digital human's emotional response in the kth round on the six basic emotions.

[0055] The beneficial effect of the present invention is that the present invention can significantly improve the personalized ability of digital people in emotional interaction by introducing a personalized emotional response model based on a Bayesian network, and cleverly integrate the unique personality characteristics of digital people and the emotional characteristics of users, which can not only solve the problem of the single emotional expression of traditional digital people, but also enable digital people to understand and respond to user needs more accurately. At the same time, this innovative model also allows digital people to show both familiar and independent emotional responses, bringing users a rich and diverse interactive experience, enhancing the user's immersion and participation, and deepening the emotional resonance and understanding between the two parties. As a result, the present invention can realize the personalization, enrichment and diversification of digital people's emotional expression, and significantly improve the application value of digital people in the field of human-computer interaction.

[0056] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0058] Figure 1 This is a diagram of the emotional interaction process of the digital human;

[0059] Figure 2 This is a framework diagram of the emotional response model;

[0060] Figure 3 This is the Bayesian network structure diagram. DETAILED DESCRIPTION

[0061] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0062] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0063] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.

[0064] The present invention provides a method for personalized emotional interaction of digital human. The ability of digital human to understand the user's emotions is an important step in emotional interaction, and only on this basis can emotional response and emotional expression be performed. and They represent the emotional input value transmitted to the digital human by the user in the kth round of interaction and the emotional response value replied to the user by the digital human in the kth round. Emotional computing requires the state evaluation of the user input emotional value. The evaluation result of the user input emotional value in the kth round is used representations and use them for subsequent emotional responses.

[0065] In order to more intuitively explain the personalized emotional interaction model of digital humans, the problem is formalized as follows:

[0066] Assume that the input sentiment value of the user in round k is The emotional response of the digital human in the kth round is: First, to model the user's emotional characteristics, it is necessary to determine the concept nodes of the Bayesian network and the conditional probabilities between the nodes to form the digital human emotion transfer probability matrix V; secondly, by modeling the unique emotions of the digital human, a complex emotional response with the unique personality characteristics of the digital human is generated. This model can be used to establish a connection between the emotions input by the user into the model and the future emotions of the digital human output by the model; finally, the personalized emotional response output by the digital human in the kth round of the model is obtained. In the above-mentioned interaction process between the user and the digital human, the digital human cycles through the above-mentioned emotional response process for many times and outputs a response emotion corresponding to the user's input emotion until the interaction ends. In summary, the problem to be solved by the present invention can be clearly defined as:

[0067] Question input:

[0068] (1) The user inputs the sentiment value in the kth round of interaction

[0069] (2) Digital human emotion transfer probability matrix V.

[0070] Problem output:

[0071] The emotional response value of the digital human replying to the user in the kth round

[0072] Considerations:

[0073] In emotional interaction, the emotional response value of the digital human is obtained by simulating human emotions, analysis and calculation. In order to make the emotional expression of the digital human closer to the user, the emotional characteristics of the user and the personality characteristics of the digital human can be integrated, so that the digital human can generate emotional expressions that are similar to the user but independent of the user, thus becoming the user's digital avatar in terms of emotional interaction.

[0074] In the process of digital human emotional interaction, the emotional interaction between users and digital humans can be described as follows: first, user interaction behavior input, the input form is various, including voice, text, etc.; second, the user interaction behavior is quantified and evaluated as an input emotion value vector, so as to realize the recognition of the emotional state of the participant; then, the user's emotional state is tracked. At this time, it is necessary to consider the complex factors that affect the emotional changes of the digital human, and incorporate these factors into the emotional calculation to obtain the emotional response of the digital human; finally, the emotional response is introduced into the natural language generation process to obtain the final emotional expression of the digital human. The process of digital human emotional interaction is as follows: Figure 1 shown.

[0075] In order to analyze the user's emotional state and track the emotional state in the subsequent interaction process, the user's interactive input is emotionally quantified to obtain its PAD (pleasure-arousal-dominance) emotional value in the continuous dimensional space. Based on the principle of mapping the continuous space to the discrete space to effectively express the emotional state of the digital human, the interactive emotional vector is then The evaluation is converted into the emotional state vector E under six basic emotional states (happy, surprised, disgusted, angry, afraid, sad) l The corresponding relationship between the six basic emotions and the PAD space is shown in Table 1.

[0076] Table 1

[0077]

[0078] Interactive input of emotions To basic emotion E l The distance between them is:

[0079]

[0080] Among them, C l is the covariance matrix set of basic emotional states in the PAD space; l = 1, 2, 3, 4, 5, 6 represent the six basic emotions of happiness, surprise, disgust, anger, fear, and sadness respectively.

[0081] The distance from the interactive input emotion to the coordinate value of each emotional state is inversely proportional to the intensity value of the emotional state effect. l Normalize to get the basic emotional state l to interact with the input emotion The intensity value i l To ensure that the formula is meaningful, h l Not 0, defines the interactive input sentiment evaluation function for:

[0082]

[0083] Among them, h l Indicates that the location of the emotional state is considered to be in C l Interactive input sentiment obtained under constraints To basic emotion E l The distance between.

[0084] Considering that in the process of digital human emotional interaction, the digital human does not yet have emotional characteristics that match the user, it is difficult to show unique and individual emotional expression in the interaction, which makes it difficult for the digital human to establish trust and resonance with the user. Therefore, the present invention proposes a personalized emotional response model based on a Bayesian network. The digital human in this model not only has unique emotional characteristics, but also has the emotional characteristics of the user. This dual emotional setting enables the digital human to reflect certain commonalities of the user in terms of emotional expression and response, while also showing its uniqueness. As a result, when the digital human interacts with the user, it can not only make the user feel a familiar emotional resonance, but also show emotional responses independent of the user, thereby enhancing the richness and diversity of the interactive experience. The emotional response model framework is as follows Figure 2 shown.

[0085] The digital human personalized emotional response model based on Bayesian network includes the following three modules:

[0086] 1. User Personalization Modeling Module

[0087] The present invention utilizes the Bayesian network to perform personalized modeling on users, so that the digital human can have emotional traits similar to those of the users.

[0088] 1) Bayesian Network Definition

[0089] The Bayesian network is a probabilistic graphical model that represents the interdependence or independence between random variables. The network structure is represented as a directed acyclic graph, which is used to qualitatively represent the dependency between variables. The parameters are the conditional probability distribution between nodes, which are used to quantitatively describe the degree of dependency between variables. Given the probability P(A) of event A and the probability P(A / B) of event A when event B occurs, the probability P(B / A) of event A when event B occurs can be calculated. The Bayesian formula is as follows:

[0090]

[0091] 2) Determine the network structure

[0092] In the process of constructing the Bayesian network structure, it is necessary to select appropriate variables. In the process of emotional interaction between people, the generation of current emotions is directly related to factors such as people's personality, historical emotional state, and identity information. The present invention selects the user's personality, historical emotional state, gender, age, and education level as nodes of the network. The Bayesian network structure is as follows: Figure 3 As shown. The present invention assumes that the influence of personality, historical emotional state, gender, age and education level on the current state is independent of each other. Based on formula (4), a Bayesian model describing emotional transfer is proposed:

[0093] p(E j |Ei ,P,G,A,C) (5)

[0094] Where p(x) represents E i Transfer to E j The transition probability of the state; P represents personality type; G represents gender; A represents age group; C represents education level.

[0095] 3) Define the state space of the node

[0096] The value status of each node element is divided as follows:

[0097] Historical emotional states: joy, sadness, anger, surprise, fear, disgust;

[0098] Personality: openness, conscientiousness, extraversion, agreeableness, neuroticism.

[0099] Age groups: children (0-12 years old), adolescents (13-18 years old), adults (19-65 years old), and the elderly (over 65 years old);

[0100] Educational level: primary school and below, junior high school, high school, university, graduate school and above.

[0101] 4) Determine the conditional probability table (CPT)

[0102] Each node in a Bayesian network requires a conditional probability table, which lists the probability values ​​of each node under different states of its parent node. The specific steps to construct a conditional probability table are as follows:

[0103] a) Calculate the prior probability p(E i ), p(P), p(G), p(A), p(C): Calculate the prior probability based on the frequency of each category in the data set.

[0104] b) Calculate the conditional probability p(E j |E i ,P,G,A,C): Assuming all features are discrete, for each input feature, the conditional probability under different sentiment states is calculated.

[0105] c) Construct the digital human emotion transfer probability matrix V: Organize all the calculated conditional probabilities into a matrix V, in which each row corresponds to the conditional probability of the child node under a specific parent node combination.

[0106] So far, an emotion transfer state model that introduces user personalization factors has been constructed. The overall process of the model is: historical emotion state, personality, gender, age, and education level are used as inputs and acted on the Bayesian network to obtain the digital human emotion state transfer matrix V.

[0107] 2. Digital Human Unique Emotion Modeling Module

[0108] 1) The influence of personality on emotional state

[0109] Under the same stimulus, people often have significant individual differences in their emotional expressions, a phenomenon known in psychology as "emotional tendency". Psychologists believe that the generation of this emotional tendency is mainly affected by personality. Therefore, in order to make digital people have personalized emotions similar to humans, it is necessary to introduce personality factors.

[0110] The present invention adopts the five factor model (FFM) as the personality model. The big five factor model divides human personality into five categories, namely openness, responsibility, extroversion, altruism and neuroticism. The description of each trait is shown in Table 2.

[0111] Table 2

[0112]

[0113] Openness refers to a person's acceptance of new experiences, creativity and changes. People with high openness are usually imaginative, curious and creative, and like to explore new things. Responsibility refers to a person's self-discipline, prudence and tendency to pursue goals. People with high responsibility are usually organized, reliable, and have strong goal orientation and perseverance. Extraversion refers to a person's sociability and vitality. People with high extraversion are usually energetic, talkative, like to interact with others, and get energy from social activities. Agreeableness refers to a person's friendliness, sympathy and willingness to cooperate. People with high agreeableness are usually considerate, friendly, willing to help others, and tend to avoid conflicts. Neuroticism refers to a person's emotional stability and ability to cope with stress. People with high neuroticism are prone to experience negative emotions, such as anxiety, depression and anger, and are more sensitive to stress. In the Big Five model, individuals are represented as a five-dimensional vector P = [p E ,p A ,p N ,p O ,p C ], where p E ,p A ,p N ,p O ,p C They represent the values ​​in the five personality domains respectively, and the value range is 0 to 1.

[0114] According to the mapping relationship between the Big Five personality and PAD, each personality is mapped to a specific PAD value. The commonly used mapping equation is shown in formula (6).

[0115]

[0116] The above formula describes the differences in pleasure, arousal and dominance among people with different personalities. According to formula (6), the individual's PAD three-dimensional emotional state vector can be corrected to make the emotional response more personalized. The correction calculation formula is:

[0117]

[0118] In formula (8): The average score of each dimension obtained by the Big Five Personality Scale (0.70, 0.73, 0.61, 0.60, 0.60) is used as the benchmark to define the neutral level of personality; is the personality correction vector; ef is the initial emotion component; e m It is the emotional state vector obtained after personality correction.

[0119] 2) Mutual influence between emotions

[0120] Human emotions are complex and they can experience multiple emotions at the same time. Especially when stimulated by external factors, emotional states fluctuate. Various emotions interact and influence each other, some inhibit each other, and some stimulate each other. In order to better simulate the complex emotions of human beings, the present invention further considers the mutual influence between emotions.

[0121] According to existing literature research, the formula for the mutual influence between emotions is as follows:

[0122] φ(E f )=E f *λ (9)

[0123] The influence factor matrix λ between emotions is shown in formula (10):

[0124]

[0125] Among them, λ ij =[-1,1] represents the influence factor of emotion i on emotion j. There are three main situations of mutual influence between emotions:

[0126] a) Excitation: When λ ij When >0, it means that emotion i has a motivating effect on emotion j.

[0127] b) Suppression: When λ ij When <0, it means that emotion i has an inhibitory effect on emotion j.

[0128] c) No effect: When λ ij When =0, it means that emotion i has no influence on emotion j.

[0129] According to the above formula, the emotional response vector can be corrected to make the digital human have the characteristics of human complex emotions. The correction calculation formula is:

[0130] E c =φ(E f ) (11)

[0131] Where: E f is to convert the e in the PAD dimension emotion space f Emotional state vector converted to discrete emotion space; E c It is the emotional state vector obtained after being corrected by the emotional influence factor matrix.

[0132] 3. Digital Human Emotional Status Update Module

[0133] The present invention mainly considers the influence of the k-th round of user input emotion, user emotional characteristics and the unique emotional characteristics of the digital human on the emotional response of the digital human, thereby obtaining the emotional values ​​of the digital human under the six basic emotional states. Then, these emotional values ​​are mapped to the PAD (pleasure-arousal-dominance) space, their coordinate positions are calibrated, the emotional state of the digital human is updated, and the corresponding emotional response is given. The formula for converting the emotional state under the six basic emotions into the PAD space and calibrating the position is:

[0134]

[0135] Among them, p l ,a l ,d l Represents the emotional values ​​of the six basic emotions in the PAD space, i l is the intensity value of the digital human's k-th round emotional response in the six basic emotions.

[0136] The Bayesian network is used to conduct personalized modeling of the user's emotional data during the emotional interaction process, and the emotional response of the digital human is obtained by combining the user's personalized emotional characteristics and the unique emotional characteristics of the digital human. The model construction is shown in Table 3.

[0137] Table 3

[0138]

[0139] In the above embodiments, the description's reference to "this embodiment" indicates that a particular feature, structure, or characteristic described in conjunction with the embodiment is included in at least some embodiments, but not necessarily all embodiments. Multiple occurrences of "this embodiment" do not necessarily all refer to the same embodiment.

[0140] In the above-described embodiments, although the invention has been described in conjunction with specific embodiments of the invention, many substitutions, modifications, and variations of these embodiments will be apparent to those of ordinary skill in the art based on the foregoing description. For example, other storage structures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed. Embodiments of the invention are intended to encompass all such substitutions, modifications, and variations that fall within the broad scope of the appended claims.

[0141] This embodiment further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, any one of the methods in this embodiment is implemented.

[0142] This embodiment also provides an electronic terminal, including: a processor and a memory;

[0143] The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the terminal executes any one of the methods in this embodiment.

[0144] The computer-readable storage medium in this embodiment can be understood by ordinary technicians in this field: all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

[0145] The electronic terminal provided in this embodiment includes a processor, a memory, a transceiver and a communication interface. The memory and the communication interface are connected to the processor and the transceiver and complete communication with each other. The memory is used to store computer programs, the communication interface is used to communicate, and the processor and the transceiver are used to run computer programs so that the electronic terminal executes each step of the above method.

[0146] In this embodiment, the memory may include a random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0147] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0148] The present invention can be used in many general or special computing system environments or configurations, such as personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like.

[0149] The present invention may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0150] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A digital human personalized emotional interaction method, characterized by: The following steps are involved: S1: Quantify the user's interactive input and obtain the pleasure-arousal-dominance PAD emotion value in the continuous dimensional space; S2: Evaluate the emotional state of the user's interactive input emotion to obtain an interactive input emotion evaluation function; S3: Use Bayesian network to calculate the digital human emotion transfer probability matrix; S4: Calculate the emotional response value of the digital person's personality after being corrected by the Big Five Factor Model in the current emotional state; S5: The corrected emotional response is processed through the emotional influence factor matrix to output a response with complex emotional characteristics; S6: Calibrate the PAD emotional space coordinates of the digital human's emotional intensity in six basic emotions to obtain the digital human's emotional response.

2. The digital human personalized emotional interaction method according to claim 1, characterized in that: Step S2 specifically includes the following steps: Inputting sentiment values ​​into user interactions The evaluation is converted into the emotional state vector E under six basic emotional states l , interactive input emotion To basic emotion E l The distance between them is: Among them, C l is the covariance matrix set of basic emotional states in the PAD space; l = 1, 2, 3, 4, 5, 6 represent the six basic emotions of happiness, surprise, disgust, anger, fear, and sadness respectively; The distance from the interactive input emotion to the coordinate value of each emotional state is inversely proportional to the intensity value of the emotional state effect. l Normalize to get the basic emotional state l to interact with the input emotion The intensity value il defines the interactive input sentiment evaluation function for: Among them, h l Indicates that the location of the emotional state is considered in the area C l Interactive input sentiment obtained under constraints To basic emotion E l The distance between.

3. The digital human personalized emotional interaction method according to claim 1, characterized in that: Step S3 specifically includes the following steps: S31: Select the user's personality, historical emotional state, gender, age and education level as nodes of the Bayesian network to build a Bayesian model describing emotional transfer: p(E j |E i ,P,G,A,C) Where p(x) represents E i Transfer to E j The transition probability of the state; P represents personality type; G represents gender; A represents age group; C represents education level; S32: define the state space of the node; S33: Constructing a conditional probability table, including the following steps: S331: Calculate the prior probability p(E i ), p(P), p(G), p(A), p(C); S332: For each input feature, calculate the conditional probability p(E j |E i ,P,G,A,C); S333: Organize all calculated conditional probabilities into a digital human emotion transfer probability matrix V, where each row in the matrix V corresponds to the conditional probability of a child node under a specific parent node combination.

4. The digital human personalized emotional interaction method according to claim 1, characterized in that: The state space of each node includes: Historical emotional states: joy, sadness, anger, surprise, fear, disgust; Personality: openness, conscientiousness, extraversion, agreeableness, neuroticism; Age groups: children, adolescence, adulthood, and old age; Educational level: primary school and below, junior high school, high school, university, graduate school and above.

5. The digital human personalized emotional interaction method according to claim 1, characterized in that: Step S4 The specific steps include: The Big Five factor model is used as the personality model, including openness, responsibility, extraversion, altruism and neuroticism. In the Big Five factor model, an individual is represented by a five-dimensional vector P = [p E ,p A ,p N ,p O ,p C ], where p E ,p A ,p N ,p O ,p C Respectively represent their values ​​in the five personality domains, with values ​​ranging from 0 to 1; Map each personality in the Big Five Factor Model to a specific PAD value, expressed as: The above formula is used to correct the PAD three-dimensional emotional state vector of a person. The formula is as follows: In the formula, The average score of each dimension obtained from the Big Five Factor Table (0.70, 0.73, 0.61, 0.60, 0.60) is used as the basis to define the neutral level of personality; is the personality correction vector; ef is the initial emotion component; e m It is the emotional state vector obtained after personality correction.

6. The digital human personalized emotional interaction method according to claim 1, characterized in that: Step S5 specifically includes the following steps: The formula for the mutual influence between emotions is as follows: φ(E f )=E f *l The influence factor matrix λ between emotions is: Among them, λ ij =[-1,1] represents the influence factor of emotion i on emotion j. The mutual influence between emotions includes three situations: Incentive: When λ ij When > 0, it means that emotion i has a motivating effect on emotion j; Suppression: When λ ij When <0, it means that emotion i has an inhibitory effect on emotion j; No effect: When λ ij =0, it means that emotion i has no effect on emotion j; According to the above formula, the emotional response vector is corrected to make the digital human have the characteristics of complex human emotions. The correction calculation formula is: AND c =φ(E f ) Where: E f is to convert the e in the PAD dimension emotion space f Emotional state vector converted to discrete emotion space; E c It is the emotional state vector obtained after being corrected by the emotional influence factor matrix.

7. The digital human personalized emotional interaction method according to claim 1, characterized in that: In step S6, the formula for converting the emotional states under the six basic emotions into the PAD space and calibrating the positions is: Among them, pl, al, dl represent the emotional values ​​of the six basic emotions in the PAD space, i l It is the intensity value of the digital human's emotional response in the kth round on the six basic emotions.

Images