A digital life body configuration method, device, equipment and storage medium

Abstract

The application discloses a configuration method, device and equipment of a digital life body and a storage medium, relates to the technical field of the digital life body, and comprises the following steps: in response to a dialogue operation corresponding to the digital life body, a target image containing a target person is acquired; the target person in the target image is identified, and identity information of the target person is determined; a social relationship between the target person and the digital life body is determined according to the identity information; a private knowledge base corresponding to the target person is determined according to the identity information of the target person; a target knowledge base of the digital life body is configured according to the private knowledge base and a preset basic database; and an initial prompt word of the digital life body is configured according to the social relationship. The application improves the personalized interaction capability of the digital life body.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital life, and in particular to a configuration method, device and equipment of digital life and a storage medium. BACKGROUND

[0002] Digital life refers to an artificial intelligence system based on computer media and having characteristics or behavior patterns of natural life. As an emerging human-computer interaction carrier, its core goal is to achieve deep emotional resonance with users through personification behavior.

[0003] However, the prior art has significant defects: current digital life generally adopts a static interaction paradigm, that is, it presents highly homogenized behavior performance when facing different users. This design is particularly prominent in application scenarios with the prototype of the deceased - when a digital life is given the identity characteristics of a specific deceased person, if it cannot dynamically adjust the behavior pattern according to the interaction object, the following problems will occur: first, the interaction is rigid, the system cannot distinguish the social relationship between the user and the prototype (such as relatives / friends / strangers), and a unified interaction strategy is adopted; second, the authenticity collapses, violating the differentiated behavior guidelines of human social interaction "different people, different behaviors", and weakening the credibility of the digital life; third, the emotional transmission fails, and the digital life cannot reproduce the differentiated emotional expression of the prototype character to different objects, resulting in fragmented user experience.

[0004] Therefore, how to improve the personalized interaction ability of digital life has become a technical problem to be solved. SUMMARY

[0005] The technical problem solved by the present application is that the personalized interaction ability of digital life is poor.

[0006] To solve the above technical problems, the present application provides the following technical solutions: a configuration method of digital life, comprising:

[0007] In response to a dialogue operation corresponding to the digital life, a target image containing a target person is obtained;

[0008] The target person in the target image is identified to determine the identity information of the target person;

[0009] The social relationship between the target person and the digital life is determined according to the identity information;

[0010] A private knowledge base corresponding to the target person is determined according to the identity information of the target person;

[0011] A target knowledge base of the digital life is configured according to the private knowledge base and a preset basic database;

[0012] An initial prompt word of the digital life is configured according to the social relationship.

[0013] Preferably, the target person in the target image is identified, and identity information of the target person is determined, comprising:

[0014] Face information of the target person is extracted from the target image;

[0015] Target genetic information corresponding to a preset genetic feature is extracted from the face information;

[0016] The target genetic information is compared with genetic information of each person template in a plurality of person templates in a preset template library to obtain a first similarity score of each person template;

[0017] Identity information corresponding to a person template with the highest first similarity score is determined as identity information of the target person.

[0018] Preferably, the identity information of the target person is determined according to the identity information corresponding to the person template with the highest first similarity score, comprising:

[0019] The person template with the highest first similarity score is determined as a target person template;

[0020] The remaining person templates in the preset template library except the target person template are determined as to-be-excluded person templates;

[0021] A second similarity score obtained by comparing each to-be-excluded person template with the target person template is obtained;

[0022] It is judged whether there is a to-be-excluded person template satisfying condition one and condition two, wherein the condition one comprises that the first similarity score of the to-be-excluded person template is greater than a preset verification threshold, and the condition two comprises that the first similarity score of the to-be-excluded person template is higher than the second similarity score of the to-be-excluded person template;

[0023] If there is no to-be-excluded person template satisfying the condition one and the condition two, identity information corresponding to the target person template is determined as identity information of the target person;

[0024] If there is a to-be-excluded person template satisfying the condition one and the condition two, the to-be-excluded person template satisfying the condition one and the condition two is determined as a to-be-verified person template;

[0025] Identity information of the target person is determined from identity information corresponding to the to-be-verified person template and the target person template.

[0026] Preferably, the private knowledge base corresponding to the target person comprises a dialogue history of a digital life body and the target person; and the identity information of the target person is determined from the identity information corresponding to the to-be-verified person template and the target person template, comprising:

[0027] determine nickname information corresponding to the character prototype of the target character template;

[0028] find a private knowledge base corresponding to the identity information of the target character template, the private knowledge base corresponding to the identity information of the target character template including a dialogue history between the digital life body and the character prototype of the target character template;

[0029] determine a dialogue topic according to the dialogue history between the digital life body and the character prototype of the target character template;

[0030] generate a question dialogue information according to the nickname information and the dialogue topic;

[0031] obtain reply information of the target character to the question dialogue information;

[0032] determine the number of third person and first person in the reply information;

[0033] if the number of third person is more than or equal to the number of first person, determine the identity information corresponding to the to-be-verified character template as the identity information of the target character;

[0034] if the number of third person is less than the number of first person, determine the identity information corresponding to the target character template as the identity information of the target character.

[0035] Preferably, the number of third person and the number of first person in the reply information are determined, comprising:

[0036] perform part-of-speech tagging on the reply information to determine pronoun information in the reply information;

[0037] select personal pronouns from the pronoun information;

[0038] determine the number of third person and the number of first person from the personal pronouns.

[0039] Preferably, the dialogue history includes a plurality of dialogue information, and the dialogue information includes a dialogue time; the dialogue topic is determined according to the dialogue history between the digital life body and the character prototype of the target character template, comprising:

[0040] the dialogue topic is determined according to dialogue information with the longest dialogue time in the dialogue history between the digital life body and the character prototype of the target character template.

[0041] Preferably, before obtaining the target image containing the target character in response to the dialogue operation corresponding to the digital life body, the method further comprises:

[0042] pre-configure a target knowledge base of the digital life body according to a basic database;

[0043] Before generating the question dialogue information according to the nickname information and the dialogue topic, the method further comprises:

[0044] The target knowledge base of the digital life body is configured according to the social relationship between the target person template and the digital life body and the private knowledge base corresponding to the target person.

[0045] After the target person in the target image is identified and the identity information of the target person is determined, before the target knowledge base of the digital life body is configured according to the private knowledge base and the preset basic database, the method further comprises:

[0046] The configuration of the digital life body is cleared.

[0047] In another aspect, a configuration device of a digital life body is also provided, which comprises: an acquisition module configured to acquire a target image containing a target person in response to a dialogue operation corresponding to the digital life body; an identification module configured to identify the target person in the target image and determine identity information of the target person; a first determination module configured to determine a social relationship between the target person and the digital life body according to the identity information; a second determination module configured to determine a private knowledge base corresponding to the target person according to the identity information of the target person; a first configuration module configured to configure a target knowledge base of the digital life body according to the private knowledge base and a preset basic database; and a second configuration module configured to configure an initial prompt word of the digital life body according to the social relationship.

[0048] In another aspect, an electronic device is also provided, which comprises a memory, a processor and a computer program stored in the memory, and the processor executes the computer program to implement the configuration method of the digital life body according to any one of the above embodiments.

[0049] In another aspect, a computer readable storage medium is also provided, which stores a computer program, and the computer program is executed by a processor to implement the configuration method of the digital life body according to any one of the above embodiments.

[0050] The present application has the following beneficial effects: by identifying the target person in the target image and determining the identity information of the target person, the social relationship between the target person and the digital life body is determined according to the identity information, and the initial prompt word of the digital life body is configured according to the social relationship, so as to limit the emotional tendency of the dialogue information of the digital life body through the initial prompt word; the private knowledge base corresponding to the target person is determined according to the identity information of the target person, and the target knowledge base of the digital life body is configured according to the private knowledge base and the preset basic database, so that when different target persons dialogue with the digital life body, the answer of the digital life body is affected by the common memory of the prototype person of the target person and the digital life body, thereby improving the personalized interaction ability of the digital life body. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 A basic flowchart of a configuration method of a digital life body according to an embodiment of the present application is provided. DETAILED DESCRIPTION

[0052] In order to make the above objectives, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all embodiments.

[0053] Embodiment 1, refer to Figure 1 For an embodiment of the present application, a method for configuring a digital life is provided, which is applied to a configuration system of a digital life, and the system is a smart terminal, such as a mobile phone, a computer, a smart watch, an AR / VR device, etc. The method comprises S110-S160:

[0054] S110, in response to a dialogue operation corresponding to the digital life, obtaining a target image containing a target person.

[0055] When the user initiates a dialogue operation with the digital life (such as a virtual character in the smart screen, AR / VR device or smart assistant application), the system is triggered. This operation can be a voice instruction (such as "Hey, little intelligence, talk to my grandfather"), click on a specific contact avatar, or the system detects that the user gazes at the device for more than a preset time, etc. The system then activates its image acquisition module (such as a camera, a scanner) to capture image data. In the image, the system needs to locate the target person (i.e. the main object of the current request for dialogue) through target detection technology (such as deep learning models such as YOLO, SSD, etc.). The system takes the image region containing the positioned target person (i.e. the target image) as the input for subsequent processing.

[0056] The method for configuring a digital life is suitable for initial configuration of a digital life, and the configuration of the digital life will be reset each time it is reused through a dialogue operation.

[0057] S120, identifying the target person in the target image and determining the identity information of the target person.

[0058] The target person is subjected to facial recognition to determine the identity information of the target person.

[0059] S130, determining the social relationship between the target person and the digital life according to the identity information.

[0060] After the identity information of the target person is determined in S120, the system queries a pre-defined identity-relationship mapping table according to the identity information. The table stores the social relationship (for example, "grandson", "son", "husband") corresponding to the preset "identity" (for example, the prototype of the digital life is set to represent "grandmother" in the family) of each known identity relative to the digital life.

[0061] For example, by S120, it is determined that the target person's identity is "Grandpa", and the preset digital life body represents "Grandma". According to the table, the relationship of "Grandpa" to "Grandma" is "spouse", so the social relationship of the digital life body is configured as "spouse".

[0062] The social relationship is a key configuration parameter, which directly affects the tone, word choice, degree of care, and behavior boundary of the digital life body, so that the interactive roles (such as elders, peers, and juniors) and basic attitudes of the digital life body match the real social relationship, greatly improving the naturalness and emotional temperature of the interaction. For example, using a more respectful and caring tone to grandparents; using a more friendly and instructive tone to children.

[0063] S140, according to the identity information of the target person, determine the private knowledge base corresponding to the target person.

[0064] After S120 determines the identity information of the target person, the system uses the identity information as a unique identifier or retrieval key to access a secure, permission-controlled private knowledge base storage area. According to the identifier, find and load the private knowledge base corresponding to it.

[0065] The private knowledge base includes all past conversation histories (time, content) between the target person and the digital life body. In addition, it can also include information such as personal preference settings, anniversaries, health data (limited to authorized use), or if the prototype of the digital life body is a deceased person, the private knowledge base can also include text conversation information, action information, voice call information, etc. when the deceased person was with the target person.

[0066] For example, the identity "Grandpa" is determined, and its private knowledge base is loaded, which contains multiple conversation records about walking the dog, planting flowers, and seeing a doctor with "Grandma" (digital life body).

[0067] S150, according to the private knowledge base and the preset basic database, configure the target knowledge base of the digital life body.

[0068] The basic database contains information about the prototype of the digital life body known to the public, including friends and relatives, that is, the basic database contains the basic information of the prototype of the digital life body, and further can include the public knowledge mastered by the prototype corresponding to the digital life body. For example, the prototype of the digital life body is a surgeon, and the digital life body is proficient in the knowledge that a surgeon should have.

[0069] The digital life prototype has different memories with each relative and friend. For example, the digital life prototype has different shared memories with college classmates and primary school classmates, respectively. Meanwhile, the digital life prototype also has different conversation histories with the user in the conversation. Therefore, the private database of the target person can include two parts. One part is the exclusive memory of the digital life prototype and the target person when they are together. The other part is the history record of the digital life and the target person when they are in conversation.

[0070] For the conversation context during the use of the digital life, the history in the private knowledge base is used preferentially. For general information query, the basic database is used preferentially. Finally, the target knowledge base that is specific to this conversation and faces the specific target person is formed, which is used as the core basis for the content generation and decision-making of the digital life.

[0071] In S160, the initial prompt words of the digital life are configured according to the social relationship.

[0072] After the social relationship is determined in S130, the system calls a preset relationship-prompt word template generation engine (or a query mapping table). Based on the determined social relationship (such as “grandson”, “father”, “spouse”), a set of initial prompt word instructions are selected or dynamically generated. These instructions describe the role, tone, basic behavior rules and target that the digital life should play in the conversation.

[0073] For example, the social relationship is “grandfather”, and the initial prompt word can be: “You are a gentle, talkative, and like to recall past events, and care about the health of your family. Speak to the grandson / granddaughter (dynamically filled according to the gender of the other party) with love, speak slowly, ask more about life details and feelings, and share positive life experiences. The current task is to accompany and chat.” This prompt word will be used as the primary constraint condition for the large language model (LLM) to generate a reply, so that the language style, focus and interaction mode of the digital life are highly consistent with the current social relationship, greatly enhancing the realism and emotional resonance of the conversation.

[0074] The application identifies the target person in the target image, determines the identity information of the target person, determines the social relationship between the target person and the digital life according to the identity information, configures the initial prompt word of the digital life according to the social relationship, limits the emotional tendency of the digital life to the conversation information through the initial prompt word, determines the private knowledge base corresponding to the target person through the identity information of the target person, and configures the target knowledge base of the digital life according to the private knowledge base and the preset basic database, so that the answer of the digital life is affected by the shared memory of the prototype person of the target person and the digital life when the different target persons are in conversation with the digital life, thereby improving the personalized interaction ability of the digital life.

[0075] Preferably, S120 comprises sub-steps S121-S124:

[0076] S121, extracting face information of the target person from the target image.

[0077] A face detection algorithm (e.g. Haar Cascade, MTCNN, Dlib) is used to precisely locate the face region in the target image, and normalize it (e.g. align, scale, grayscale / standardize RGB values) to obtain face image information in a unified format.

[0078] S122, extracting target genetic information corresponding to the preset genetic features from the face information.

[0079] A feature vector corresponding to the preset genetic features is extracted from the normalized face image information. The preset genetic features refer to facial structural feature points or deep features that are highly similar among direct or collateral relatives, and are inherited from parents. In the industry, this is usually achieved by a trained deep neural network (e.g. FaceNet, DeepFace, ArcFace). These networks encode face images into a high-dimensional vector (embedding vector). This embedding vector is essentially the coordinate position of the face in the feature space learned by the model, and contains target genetic information that encodes the preset genetic features for distinguishing individuals.

[0080] S123, comparing the target genetic information with the genetic information of each person template in the preset template library to obtain a first similarity score of each person template.

[0081] The extracted target genetic information (query vector) is compared with the genetic information (pre-stored embedding vector of the person template) of each person template in the preset template library for similarity calculation. Common similarity measures include cosine similarity (Cosine Similarity) or Euclidean distance (Euclidean Distance). The result of the calculation is to generate a first similarity score (a value between 0 and 1 or a standardized value, the higher the value, the more similar to the current target image) for each person template in the preset template library.

[0082] The preset template library includes face feature information of relatives and friends of the prototype person of the digital life form under social relationships. In the case of the digital life form being used to commemorate the deceased, the preset template library comparison can improve the speed of face recognition to quickly respond to user operations.

[0083] S124, determining the identity information of the target person according to the identity information corresponding to the highest person template in the first similarity score.

[0084] In an alternative way, the identity information corresponding to the character template with the highest first similarity score can be directly determined as the identity information of the target character.

[0085] In the case that the prototype of the digital life is a deceased person, if an identification error occurs, it is easy to cause the emotional transmission to fail, so the identification of the identity information of the target character needs to be very accurate. However, the genetic characteristics between relatives are very similar, especially when relatives live together for a long time, or in the case of twins, it is easy to cause an identification error of the identity information of the target character.

[0086] Therefore, preferably, S124 includes sub-steps S210-S270:

[0087] S210, determining the character template with the highest first similarity score as the target character template.

[0088] Find the character template with the highest first similarity score and mark it as the target character template.

[0089] S220, determining the remaining character templates in the preset template library except for the target character template as the to-be-excluded character templates.

[0090] Determine all other character templates in the preset template library except for the target character template as the to-be-excluded character templates.

[0091] S230, obtaining the second similarity score obtained by comparing the similarity between each to-be-excluded character template and the target character template.

[0092] In order to improve the response speed of the system, the second similarity score obtained by comparing the similarity between each to-be-excluded character template and the target character template can be obtained, and the second similarity score is stored in the system as known data.

[0093] S240, judging whether there is a to-be-excluded character template that satisfies condition one and condition two at the same time, wherein condition one includes that the first similarity score of the to-be-excluded character template is greater than a preset verification threshold, and condition two includes that the first similarity score of the to-be-excluded character template is higher than the second similarity score of the to-be-excluded character template; if there is no to-be-excluded character template that satisfies condition one and condition two at the same time, execute S250 and execute S130-S160; if there is a to-be-excluded character template that satisfies condition one and condition two at the same time, execute S260-S270 and execute S130-S160.

[0094] Since the target image of the target character may be in a state of insufficient light, overexposure, or makeup when shooting, it is easy to mix up two people who look alike. Therefore, it is necessary to judge whether the mixing up occurs during identity recognition based on whether condition one and condition two are established at the same time.

[0095] Condition 1: the first similarity score of the to-be-excluded person template is greater than a verification threshold (e.g., 0.85 or 0.9). Condition 1 is used to determine whether the to-be-excluded person template itself is very similar to the target person.

[0096] Condition 2: the first similarity score of the to-be-excluded person template is greater than the second similarity score of the to-be-excluded person template and the target person template. The second similarity score of the to-be-excluded person template and the target person template is used to represent the similarity between the prototype of the to-be-excluded person template and the prototype of the target person template in a natural state (in multiple states). If the target person is the prototype of the target person template, the similarity between the prototype of the to-be-excluded person template and the prototype of the target person should be reduced in an unnatural shooting state, that is, the first similarity score of the to-be-excluded person template is less than or equal to the second similarity score of the to-be-excluded person template and the target person template, that is, Condition 2 is not established. If Condition 2 is established, it indicates that face recognition may be wrong. For example, the eyebrow width of twin brothers, the elder brother and the younger brother, is different, the elder brother's eyebrow is wider, and the younger brother's eyebrow is narrower. The target person is the younger brother, but the younger brother has thickened the eyebrows. At this time, face recognition is easy to confuse the two brothers.

[0097] S250, determining the identity information corresponding to the target person template as the identity information of the target person.

[0098] S260, determining the to-be-excluded person template that satisfies both Condition 1 and Condition 2 as the to-be-verified person template.

[0099] S270, determining the identity information of the target person from the identity information corresponding to the to-be-verified person template and the target person template.

[0100] When the system cannot determine the identity between two similar person templates (such as twins A and B), the final confirmation can be made through dialogue verification. Specifically, S270 includes sub-steps S271-S278:

[0101] S271, determining the nickname information corresponding to the prototype of the target person template.

[0102] The identity information of the target person template (assuming A), the nickname (such as "Xiaoming").

[0103] S272, searching for a private knowledge base corresponding to the identity information of the target person template, which includes the dialogue history between the digital life body and the prototype of the target person template.

[0104] S273, determining the dialogue theme according to the dialogue history between the digital life body and the prototype of the target person template.

[0105] Extract the historical dialogue topic from A's private knowledge base (e.g., last week's conversation about "pet dogs")

[0106] The dialogue history includes multiple dialogue information, and the dialogue information includes dialogue time; the dialogue topic is determined according to the dialogue history of the digital life body and the character prototype of the target character template.

[0107] S273 includes S273a. S273a, according to the dialogue information with the longest dialogue time in the dialogue history of the digital life body and the character prototype of the target character template, determines the dialogue topic.

[0108] That is, find the latest dialogue information.

[0109] S274, generating the question dialogue information according to the nickname information and the dialogue topic.

[0110] Generate question dialogue information combined with nickname + topic, for example: "Does Xiaoming remember the dog training techniques mentioned last week?"

[0111] S275, obtaining the reply information of the target character to the question dialogue information.

[0112] S276, determining the number of third person and first person in the reply information.

[0113] S277, if the number of third person is more than or equal to the number of first person, the identity information corresponding to the to-be-verified character template is determined as the identity information of the target character.

[0114] S278, if the number of third person is less than the number of first person, the identity information corresponding to the target character template is determined as the identity information of the target character.

[0115] The target character replies to the question, for example: "He (Xiaoming) is too busy recently and has no time to train."

[0116] Analyze the grammatical relationship of the pronoun in the reply: first person ("I"): refers to the speaker himself.

[0117] Third person ("he / she"): refers to others (here refers to "Xiaoming" of template A).

[0118] If the number of third person is greater than or equal to the number of first person, the current dialoger is the to-be-verified template B, and "he" in the reply refers to "Xiaoming" of template A, which means that the speaker is not Xiaoming himself (such as B talking about A).

[0119] If the number of third person is less than the number of first person, the current dialoger is the target template A, and "I" in the reply refers to himself (such as "*I remember, I tried that method*"), which is consistent with the identity of A.

[0120] When the biometric recognition fails, a personalized dialogue is generated using a private knowledge base, and the identity is verified again through linguistic analysis. This not only improves the interaction accuracy of digital life, but also gives it the ability to "understand social relationships" and "remember individual experiences", providing a technical foundation for highly anthropomorphic AI interaction.

[0121] Preferably, S276 comprises sub-steps S276a-S276c:

[0122] S276a, performing part-of-speech tagging on the reply information to determine pronoun information in the reply information.

[0123] Identify pronouns (such as "I", "you", "he", "they") in the sentence.

[0124] S276b, filtering out personal pronouns from the pronoun information.

[0125] Exclude pronouns in non-subject positions (e.g. "I" in "my book" is an adjective and does not directly reflect the dialogue subject).

[0126] S276c, determining the number of third-person pronouns and the number of first-person pronouns from the personal pronouns.

[0127] According to the part-of-speech tagging results, classify and count: first person: I, we, we, we; third person: he, she, it, they, they.

[0128] When the digital life is in dialogue, if the user is a temporary visitor, a temporary configuration can be provided.

[0129] Preferably, before S110, the method further comprises:

[0130] S101, pre-configuring the target knowledge base of the digital life according to the basic database.

[0131] Before S274, the method further comprises S273d:

[0132] According to the character prototype of the target character template and the social relationship of the digital life, the digital life is configured to generate temporary prompt words.

[0133] After S120, before S150, the method further comprises:

[0134] S141, clearing the configuration of the digital life.

[0135] In another aspect, the application provides a device for configuring a digital life body, comprising: an obtaining module configured to obtain a target image containing a target person in response to a dialogue operation corresponding to the digital life body; an identifying module configured to identify the target person in the target image and determine identity information of the target person; a first determining module configured to determine a social relationship between the target person and the digital life body according to the identity information; a second determining module configured to determine a private knowledge base corresponding to the target person according to the identity information of the target person; a first configuring module configured to configure a target knowledge base of the digital life body according to the private knowledge base and a preset basic database; and a second configuring module configured to configure an initial prompt word of the digital life body according to the social relationship. The device can also perform the method for configuring a digital life body according to any one of the above embodiments.

[0136] In another aspect, the application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the method for configuring a digital life body according to any one of the above embodiments.

[0137] In another aspect, the application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the method for configuring a digital life body according to any one of the above embodiments.

[0138] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, a system or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (or computer- readable storage media) having computer-usable program code embodied in the medium. The medium can be any available storage media that can be accessed by a computer. By way of example, and not limitation, such computer-usable storage media can include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other storage medium(s) that can be used to carry or store desired computer program code in the form of instructions or data structures and that can be accessed by a computer. Also, the present application can be embodied in a computer program product that can be traded as goods or merchandise, through a computer-based platform or Figure 1 one or more functions specified in the flow or flows and / or blocks Figure 1 one or more functions specified in the flow or flows and / or blocks

[0139] It should be noted that the above-mentioned embodiments are only used to illustrate but not to limit the technical solutions of the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, and they should be covered in the scope of the claims of the present application.

Claims

1. A method for configuring a digital life form, characterized in that, The method comprises the following steps: in response to a dialogue operation corresponding to a digital life body, obtaining a target image containing a target person; identifying the target person in the target image and determining identity information of the target person; determining a social relationship between the target person and the digital life body according to the identity information; determining a private knowledge base corresponding to the target person according to the identity information of the target person; configuring a target knowledge base of the digital life body according to the private knowledge base and a preset basic database; configuring an initial prompt word of the digital life body according to the social relationship; wherein the step of identifying the target person in the target image and determining the identity information of the target person comprises: extracting face information of the target person from the target image; extracting target genetic information corresponding to a preset genetic feature from the face information; performing similarity comparison between the target genetic information and genetic information of each person template in a preset template library to obtain a first similarity score of each person template; determining the identity information of the target person according to the identity information corresponding to the highest person template in the first similarity score; wherein the step of determining the identity information of the target person according to the identity information corresponding to the highest person template in the first similarity score comprises: determining the highest person template in the first similarity score as a target person template; determining the remaining person templates in the preset template library except the target person template as to-be-excluded person templates; obtaining a second similarity score obtained by performing similarity comparison between each to-be-excluded person template and the target person template; determining whether there is a to-be-excluded person template that satisfies condition one and condition two at the same time, wherein the condition one comprises that the first similarity score of the to-be-excluded person template is greater than a preset verification threshold, and the condition two comprises that the first similarity score of the to-be-excluded person template is higher than the second similarity score of the to-be-excluded person template; if there is no to-be-excluded person template that satisfies the condition one and the condition two at the same time, determining the identity information corresponding to the target person template as the identity information of the target person; if there is a to-be-excluded person template that satisfies the condition one and the condition two at the same time, determining the to-be-excluded person template that satisfies the condition one and the condition two at the same time as a to-be-verified person template; determining the identity information of the target person from the identity information corresponding to the to-be-verified person template and the target person template.

2. The method of claim 1, wherein, The private knowledge base corresponding to the target person comprises a dialogue history between the digital life body and the target person; the step of determining the identity information of the target person from the identity information corresponding to the to-be-verified person template and the target person template comprises: determining nickname information corresponding to a person prototype of the target person template; finding a private knowledge base corresponding to the identity information of the target person template, wherein the private knowledge base corresponding to the identity information of the target person template comprises a dialogue history between the digital life body and the person prototype of the target person template; determine a dialogue theme according to the dialogue history of the digital life body and the character prototype of the target character template; generate a question dialogue information according to the nickname information and the dialogue theme; obtain reply information of the target character to the question dialogue information; determine the number of third person and first person in the reply information; if the third person is more than or equal to the first person, determine the identity information corresponding to the to-be-verified character template as the identity information of the target character; if the third person is less than the first person, determine the identity information corresponding to the target character template as the identity information of the target character.

3. The method of claim 2, wherein, The method further comprises: performing part-of-speech tagging on the reply information to determine pronoun information in the reply information; screening personal pronouns from the pronoun information; determining the number of third person and first person from the personal pronouns.

4. The method of claim 3, wherein, The dialogue history comprises a plurality of dialogue information, and the dialogue information comprises a dialogue time; the method further comprises: determining a dialogue theme according to dialogue information with the longest dialogue time in the dialogue history of the digital life body and the character prototype of the target character template.

5. The method of claim 4, wherein, The method further comprises: pre-configuring a target knowledge base of the digital life body according to the basic database; The method further comprises: configuring the digital life body to generate a temporary prompt word according to the social relationship between the character prototype of the target character template and the digital life body; The method further comprises: clearing the configuration of the digital life body.

6. An apparatus for implementing the configuration method of the digital living entity according to any one of claims 1 to 5, characterized by The apparatus comprises: an obtaining module, configured to obtain a target image containing a target character in response to a dialogue operation corresponding to a digital life body; an identifying module, configured to identify the target character in the target image and determine identity information of the target character; a first determining module, configured to determine a social relationship between the target character and the digital life body according to the identity information; a second determining module, configured to determine a private knowledge base corresponding to the target character according to the identity information of the target character; a first configuring module, configured to configure a target knowledge base of the digital life body according to the private knowledge base and a preset basic database; a second configuring module, configured to configure an initial prompt word of the digital life body according to the social relationship.

7. An electronic device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program comprises instructions that, when executed by the processor, cause the electronic device to perform the method of any one of claims 1-6. The processor executes the computer program to implement the configuration method of the digital life body according to any one of claims 1-5.

8. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the configuration method of the digital life body according to any one of claims 1-5.

Images