A method and system for constructing a personalized virtual human based on personal scenario data

Abstract

This invention belongs to the technical field of personalized virtual human construction methods, and particularly relates to a method and system for constructing personalized virtual humans based on personal contextual data. The method involves collecting user personal contextual data and performing layered labeling, ownership definition, and standardized cleaning to obtain effective contextual data. Based on this effective contextual data, a user-specific personality model is trained, incorporating an intent recognition module and employing guided interaction logic to achieve virtual human modeling, avoiding cognitive biases. The digital personality identity is uniquely bound to the user's DNA, completing the ownership confirmation, inheritance, and restoration mechanism settings for the digital personality.

Description

Technical Field

[0001] This invention belongs to the technical field of personalized virtual human construction methods, and particularly relates to a personalized virtual human construction method and system based on personal scenario data. Background Technology

[0002] While virtual human technology is rapidly developing in the field of artificial intelligence and is gradually being applied to various scenarios such as companionship and interaction, current technologies are mostly based on training generalized AI models using generalized corpora. They lack deep integration of individual contextualized behavior, emotions, and unique interaction data, failing to accurately replicate the user's personalized personality traits. This results in virtual humans lacking exclusive attributes and a realistic, contextualized companionship effect. Furthermore, the industry commonly uses cloud storage for user data, causing users to lose control over their personal data. This data is easily exploited by platforms as commercial assets, and there are risks of illegal access, sale, and leakage. Moreover, digital personalities lack a unique ownership mechanism linked to user identity, and there are no standardized inheritance or recovery processes, making it difficult to meet the core needs for the replication and inheritance of a personalized digital personality.

[0003] Current virtual human technology suffers from numerous technical and ethical shortcomings. Firstly, the personality modeling employs a domestication-style interaction logic, directly outputting decision-making results to users, which can easily lead to cognitive dependence and infringe upon users' intellectual sovereignty. Secondly, it lacks a reality-first anti-addiction design, with no effective interaction time control or guidance mechanism, easily causing users to become addicted to the virtual world. Thirdly, hardware storage security is low, lacking localized encrypted storage, self-destruction upon disassembly, and geofencing protection, resulting in a persistently high risk of data going off-site. Fourthly, the division of terminal interaction permissions is ambiguous, allowing terminals to directly capture and store users' raw data, posing a serious risk of data leakage. Furthermore, there is no graded control or legal adaptation mechanism for cross-border data, making core data vulnerable to illegal access by overseas third parties. In addition, the lack of a robust risk control system for digital personality-related operations makes it prone to misoperation and coercive operations. The overall technical system cannot simultaneously meet the development needs of personalization, security, and ethics, becoming a significant obstacle to the practical application of personalized virtual human technology. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a method for constructing a personalized virtual human based on individual contextual data, the method comprising: Collect users' personal contextual data and perform hierarchical labeling, ownership definition and standardized cleaning to obtain effective contextual data; Based on the aforementioned effective scenario-based data, a user-specific personality model is trained, equipped with an intent recognition module, and guided interaction logic is used to realize virtual human modeling, thus avoiding mind domestication; By uniquely binding the digital personality identity to the user's DNA, the ownership, inheritance, and recovery mechanisms for the digital personality are established. The original contextualized data and the exclusive personality model are stored locally in encrypted hardware, and hardware self-destruction and geofence protection mechanisms are set up to ensure that the data never leaves the domain. A three-level confirmation plus cooling-off period mechanism is set up for core operations related to digital personality, and a two-level confirmation mechanism is set up for general operations to prevent accidental and coerced operations. Monitor the interaction time between users and virtual humans, and adopt a graded anti-addiction strategy of consultation-based guidance + hardware-forced exit to realize the interaction rule of prioritizing reality; User data is classified as core / non-core. Core data is prohibited from leaving the country, while non-core data is de-identified and exported in compliance with the laws of both parties. An overseas access interception mechanism is also set up. The virtual human is packaged into standardized components to adapt to various terminals. Only the terminal is assigned interactive execution permissions, and the terminal is prohibited from obtaining, storing and forwarding raw data.

[0005] Preferably, the scenario-based data collection involves collecting user behavior data, emotional expression data, and scenario-specific interaction data through local hardware audio and video acquisition units and sensor units. The three types of data are uniquely identified, tagged, and stored independently.

[0006] Preferably, when multiple people collaboratively collect contextual data from the same user, a unique collection identifier is assigned to each collector, and the collector, collection time, and collection scenario of the data from each perspective are recorded. This clarifies that the data ownership belongs to the user, and the collector only has the right to view the data authorized by the user. Moreover, the right to view the data can be modified by the user through a risk control mechanism.

[0007] Preferably, the data cleaning involves using a data cleaning algorithm to deduplicatize, reduce noise, and standardize the collected raw contextual data, repairing noisy data, and converting unstructured audio, video, and voice data into structured data to obtain the effective contextual data.

[0008] Preferably, the DNA binding and confirmation involves extracting a user's DNA sample, writing a unique digital personality identifier sequence into a DNA molecule using epigenetic bit DNA storage technology, forming a unique digital personality-DNA binding relationship, and storing it in a DNA verification unit on local encrypted hardware; setting up a digital personality inheritance and recovery process based on the binding relationship, and using the DNA-bound identity identifier as the technical basis for legal confirmation of digital personality rights.

[0009] Preferably, in daily use of the virtual avatar, a two-factor authentication method using at least one fingerprint and a retina sensor is employed. Users can register up to 10 fingerprints for random selection, with the specific finger used for verification randomly assigned by the system to enhance security. Upon successful verification, the user gains full operational privileges over the virtual avatar, while other users only gain limited interaction / viewing permissions authorized by the user.

[0010] Preferably, during the transfer of digital personality, the heir needs to provide their own DNA sample and the DNA sample of the deceased. After the DNA verification unit matches and verifies the sample, the transfer is completed by combining a three-level confirmation + cooling-off period mechanism. When local hardware data is damaged, the unique identifier of the digital personality is read through the user's DNA sample to achieve accurate recovery of the personality model.

[0011] Preferably, in the guided personality modeling, the intent recognition module is a boasting recognition submodule, which extracts behavioral intent features from user statements using natural language processing technology. Statements without actual behavioral support are judged as having no real intent and are not included in the training of personality model decision features. The guided interaction logic is that when a user seeks decision advice, the virtual human does not directly output the decision result, but guides the user to think independently by raising targeted questions.

[0012] Preferably, the local encrypted storage uses the military-grade SM4 encryption algorithm to encrypt the data. The hardware has a built-in pressure sensor and a data erasure chip. When subjected to external force during disassembly, the data erasure chip is triggered to irreversibly erase all stored data within 1 second. The geographical range of data storage is set through a geofencing module. When the hardware is taken away from the set range, it automatically enters a sleep state. Data backup only supports local backup initiated by the user through a three-level confirmation mechanism, and cloud backup is prohibited.

[0013] Furthermore, embodiments of the present invention also provide a personalized virtual human construction system based on individual contextual data, characterized in that it includes: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the aforementioned method for constructing a personalized virtual human based on personal contextual data by executing the machine-executable instructions.

[0014] In another aspect, embodiments of the present invention also provide a computer program product, the computer program product including machine-executable instructions, the machine-executable instructions being stored in a computer-readable storage medium, the processor of a computer device reading the machine-executable instructions from the computer-readable storage medium, the processor executing the machine-executable instructions, causing the computer device to execute the above-described method for constructing a personalized virtual human based on personal contextual data.

[0015] Based on the above, by layering, defining ownership, and standardizing the cleaning of personal scenario-based data, combined with guided personality modeling logic, virtual humans can accurately replicate users' behavioral habits, emotional expressions, and unique personality traits, completely solving the problems of traditional general virtual humans lacking exclusive attributes and insufficient authenticity in scenario-based companionship. The inclusion of a boasting recognition module enables effective differentiation of speech intent, and the guided interaction design avoids AI's manipulation of users' thinking, ensuring users' intellectual sovereignty and decision-making autonomy. Furthermore, the unique binding mechanism between DNA and digital personality fills the technological gap in digital personality rights confirmation, legal inheritance, and data recovery, providing a reliable technical basis for the legal protection of digital personality.

[0016] This invention employs a triple protection mechanism of local encrypted storage, hardware self-destruction upon disassembly, and geofencing to ensure user data never leaves the domain, thoroughly safeguarding user data sovereignty from the ground up and eliminating the risk of data being illegally obtained or sold by third parties. A three-level confirmation + cooling-off period risk control mechanism effectively avoids accidental and coerced operations in core processes. A tiered anti-addiction strategy implements the ethical principle of prioritizing reality, making virtual human interaction more humane. A tiered cross-border data management system and legal firewall ensure the compliance of data export. A terminal-based one-way control interaction mechanism enables plug-and-play virtual humans across terminals while blocking the path for terminals to capture and leak original data. The overall technical system deeply integrates technical protection and ethical rules, achieving secure and controllable construction of a personalized virtual human throughout the entire process. It provides a feasible technical solution for personal digital personality inheritance and family-based scenario-based companionship, possessing both outstanding technological innovation and practical application value. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the execution flow of the method for constructing a personalized virtual human based on individual contextual data provided in an embodiment of the present invention.

[0018] Figure 2 This is a schematic diagram of exemplary hardware and software components of a personalized virtual human construction system based on individual scenario data provided in an embodiment of the present invention. Detailed Implementation

[0019] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a method for constructing a personalized virtual human based on personal contextual data, as provided in one embodiment of the present invention. The following is a detailed description of this method for constructing a personalized virtual human based on personal contextual data.

[0020] The core technical solution of this invention is a method and system for constructing a personalized virtual human based on individual scenario data. The method covers the entire process of "data collection - personality modeling - secure storage - rights confirmation and inheritance - risk control - anti-addiction - cross-border management and control - terminal interaction". The system is divided into a hardware layer and a software layer, and 60 rule-layer ethical rules are written into the bottom layer to achieve dual protection of "technical wall and ethical soul line".

[0021] The key innovation of this invention lies in: For the first time, it deeply integrates personal contextual data with personality modeling, collects data in layers according to life, emotion and scenario, clarifies ownership when collecting data from multiple people, and solves the problem of general virtual humans having no exclusive characteristics; It proposes guided personality modeling rather than domestication modeling, and sets up a boasting detection module to protect users' intellectual sovereignty and avoid mind domestication; Achieving a unique binding between DNA and digital personality, and combining epigenetic bit DNA storage technology, to solve the issues of ownership, inheritance, and legal authentication of digital personality; It employs triple protection: local hardware storage, self-destruct upon disassembly, and geofencing, to ensure that data never leaves the domain and safeguard user data sovereignty. The design incorporates a three-tiered confirmation and cooling-off period risk control mechanism to prevent misoperation and coerced operation, thus providing a "brake" for user decision-making. Constructing a consultative anti-addiction mechanism that prioritizes reality, and implementing anti-addiction measures through prompts and hardware-based forced exit, aligns with the ethical principle that "the virtual world serves reality." Establish a tiered data control and legal firewall system to prohibit third parties from accessing core data without authorization and ensure cross-border data security. It enables one-way control interaction between terminals, allowing virtual humans to be plugged and played across terminals. Terminals only have execution permissions and no right to capture original data, thus technically preventing terminal data leakage.

[0022] Example 1: Layered Collection and Cleaning of Contextualized Data This embodiment addresses the problems of existing virtual humans lacking personal contextual features and the chaotic ownership of data collected from multiple users, by achieving accurate collection of contextualized data and clear ownership.

[0023] Data is collected in layers: The audio and video acquisition unit and sensor unit of the local hardware collect user life behavior data (such as daily routine, language habits, and action characteristics), emotional expression data (such as language, facial expressions, and body movements when emotional fluctuations), and exclusive scene interaction data (such as interaction characteristics of exclusive scenes such as family gatherings, parent-child companionship, and work communication). The three types of data are labeled and stored independently. Multi-user data collection ownership definition: When multiple people collaboratively collect contextual data of the same user (such as family members jointly recording the user's parent-child companionship data), the system assigns a unique collection identifier to each collector, records the collector, collection time, and collection scenario of data from each perspective, and clarifies that the ownership of the data belongs to the user, and the collector only has the right to view it (the collector's viewing right can be modified by the user through the risk control mechanism). Data cleaning: Data cleaning algorithms are used to deduplicatize, reduce noise, and standardize the collected raw data, remove invalid and redundant data, repair noisy data, and transform unstructured data (such as audio, video, and voice) into structured data, providing a high-quality data foundation for subsequent personality modeling.

[0024] The beneficial effects of this embodiment are: it enables precise and layered collection of personal scenario-based data, ensuring the authenticity and unique characteristics of the data, while clarifying the data ownership when multiple people are involved, thus avoiding data ownership disputes.

[0025] Example 2: Unique Ownership and Inheritance of DNA-Bound Digital Personality This embodiment addresses the problem of digital personalities lacking unique identifiers and being unable to be legally inherited. It combines epigenetic molecular bit DNA storage technology to achieve unique ownership and inheritance of digital personalities, corresponding to Rule 5 (DNA binding).

[0026] DNA Identity Generation: Extract the user's DNA sample and write the user's unique digital personality identifier sequence (including user identity information and personality model feature information) into the DNA molecule through epigenetic molecular bit DNA storage technology, forming a unique binding relationship between digital personality and DNA, and storing the binding data in the DNA verification unit of local encrypted hardware; Routine verification and permission allocation: When using the virtual human daily, at least one fingerprint is registered by the user, and the user can register up to 10 fingerprints for random use. The specific finger for fingerprint verification is randomly assigned by the system to improve security and retinal two-factor authentication (avoiding wear and tear caused by frequent use of DNA samples). After successful verification, the user can obtain full operation permissions of the virtual human, while other users can only obtain limited permissions authorized by the user (such as viewing and simple interaction). Inheritance and Recovery Mechanism: When a user needs to pass on their digital personality to an heir, the heir must provide their own DNA sample and the user's DNA sample. After matching and verification by the DNA verification unit, combined with a three-level confirmation and cooling-off period mechanism, the transfer of the digital personality is completed. When local hardware data is damaged, the unique identifier of the digital personality is read through the user's DNA sample to achieve accurate recovery of the personality model. Legal confirmation of rights: DNA-bound digital personality identifiers can be quickly read using portable nanopore sequencers. The reading results can serve as the legal basis for confirming the rights of digital personalities, providing technical support for the legal protection of digital personalities.

[0027] The beneficial effects of this embodiment are: it realizes the unique binding between digital personality and user identity, solves the problems of inheritance, restoration and legal confirmation of digital personality, and fills the technical gap in digital personality identity authentication.

[0028] Example 3: Guided Personality Modeling to Prevent Mind Taming This embodiment addresses the problem of user mind dependence caused by existing AI training-style modeling by implementing guided personality modeling, which protects the user's intellectual sovereignty, corresponding to Rule 14 (Closed Freedom) and Rule 58 (Intellectual Sovereignty), and corresponds to Meilai Patent 001.

[0029] Personality Model Training: Based on cleaned, contextualized data, a user's personalized personality model is trained using a large language model and a computer vision model. The model learns the user's language habits, emotional expression, and behavioral decision-making characteristics to achieve a high degree of similarity between the virtual person and the user. Deployment of the bragging detection module: The bragging detection submodule is integrated into the personality model. It uses natural language processing technology to extract behavioral intent features from user statements. When statements are not supported by actual behavior (such as no specific action plan or behavioral result), they are judged as "joking statements" and are not included in the decision feature training of the personality model, thus achieving "judging by actions, not intentions". Guided interaction design: When users seek decision-making advice from the virtual human, the virtual human does not directly output the decision result, but guides the user's thinking by asking targeted questions (such as when a user asks "Should I resign?", the virtual human does not answer "yes / no", but asks "What are your reasons for resigning? What are your future career plans?"), avoiding the user's dependence on AI thinking and protecting the user's intellectual sovereignty.

[0030] The beneficial effects of this embodiment are: it achieves the accurate construction of a user-specific personality model, and at the same time, through the boasting recognition module and guided interaction design, it avoids the AI ​​from domesticating the user's thinking and protects the user's intellectual sovereignty and freedom of speech.

[0031] Example 4: Hardware-localized secure storage and self-destruct protection This embodiment addresses the issues of low hardware security levels in existing cloud data storage technologies by achieving secure local data storage and protecting user data sovereignty, corresponding to Rule 48 (Data Sovereignty) and Meilai Patent 005.

[0032] Local encrypted storage: The user's original contextual data and trained personality model are all stored in local encrypted hardware. The hardware uses military-grade SM4 encryption algorithm to encrypt the data. The hardware and software layers communicate through a dedicated encryption protocol to prevent unauthorized access to the data from external networks. Disassembly self-destruct mechanism: A pressure sensor and a data erasure chip are built into the local hardware. When the hardware is subjected to disassembly force (such as prying or cutting), the pressure sensor immediately triggers the data erasure chip, erasing all data stored in the hardware within 1 second. Once the data is erased, it cannot be recovered. Geofencing restriction: By using the geofencing module of the hardware, the geographical range of data storage can be set. When the hardware is taken out of the set range, the hardware will automatically enter a sleep state, prohibiting any data reading or operation, so that the data never leaves the domain. Data Backup: Users can enable the local backup function through a three-level confirmation mechanism. Backup data is stored on the local backup hardware specified by the user, cloud backup is disabled, and all data is under the user's control.

[0033] The beneficial effects of this embodiment are: it realizes hardware-local storage of user data, and through technologies such as self-destruction and geofencing, it ensures that the data never leaves the domain and is not illegally obtained by third parties, thus completely protecting the user's data sovereignty.

[0034] Example 5: Operational risk control mechanism with three-level confirmation and cooling-off period This embodiment addresses the problem of existing technologies lacking risk control mechanisms and being prone to misoperation / coercive operation by setting up risk control safeguards for critical operations, corresponding to Meilai Patent 004.

[0035] Key Operation Definition: The system defines data authorization, personality model modification, digital personality inheritance, hardware unbinding, and commercial function activation as core key operations, and virtual human permission modification and data collector viewing rights adjustment as general key operations. Three-level confirmation mechanism: For core and critical operations, a three-level confirmation process is set up, consisting of fingerprint verification → retinal verification → secondary manual confirmation. All verification steps must be completed by the user and none can be omitted. For general critical operations, a two-level confirmation process is set up, consisting of fingerprint verification → secondary manual confirmation. Cooling-off period mechanism: After the three-level confirmation of core critical operations is completed, a cooling-off period is triggered. The duration of the cooling-off period is user-defined, and the system sets the minimum cooling-off period to 24 hours. During the cooling-off period, the operation will not be executed, and the user can cancel the operation at any time. After the cooling-off period ends, the user needs to verify their fingerprint again before the operation can be officially executed. Coercive operation recognition: The system monitors the user's operational status (such as heart rate and limb movements) through sensors. When the system detects that the user is in an abnormal state, it automatically pauses the operation process and triggers a safety alert.

[0036] The beneficial effects of this embodiment are: it provides multiple risk control safeguards for users' critical operations, effectively prevents misoperation and coerced operation, improves the security of system use, and provides a "brake" for users' impulsive decisions.

[0037] Example 6: Reality-Prioritized Consultative Anti-Addiction Virtual Human Interaction This embodiment addresses the problem that existing virtual human interaction can easily lead to user addiction by implementing a reality-first anti-addiction interaction, which corresponds to rule 3 (reality priority) and corresponds to Meilai Patent 007.

[0038] Interaction duration monitoring: The system monitors the continuous interaction duration and the cumulative daily interaction duration between the user and the virtual human in real time. Users can set the duration threshold according to their own needs (such as 1 hour of continuous interaction or 3 hours of cumulative daily interaction). Consultative guidance reminder: When the user interaction time reaches the threshold, the system outputs consultative guidance messages through a virtual human (such as "We've been chatting for a long time, how about we go look at the scenery outside the window and take a break before we continue chatting?"), instead of forcibly interrupting the interaction, thus balancing the user's interactive experience with real needs; Hardware forced exit: When the user interaction time exceeds the limit threshold (set by the user or guardian), the system triggers the hardware-level anti-addiction forced exit module, directly interrupting the virtual human interaction and putting the hardware into a sleep state. It can only be restarted after the user verifies the fingerprint, and the sleep time is not less than 1 hour. Real-world scene linkage: The system can link with the user's real-world scene devices (such as smart lights and smart speakers). When it detects that the user has been interacting for a long time, the linked devices will issue real-world scene reminders (such as the smart lights turning brighter or the smart speaker playing a reminder voice from a family member) to guide the user back to reality.

[0039] The beneficial effects of this embodiment are: it realizes a graded anti-addiction mechanism of "consultative guidance and hardware-forced exit", which is in line with the ethical principle that "the virtual world is not the destination, and gathering is for a better farewell", ensuring that users do not become addicted to the virtual world and prioritize real life.

[0040] Example 7: Cross-border Tiered Control for Data Sovereignty Protection This embodiment addresses the problem of uncontrolled cross-border data and easy access by overseas third parties in existing technologies by implementing cross-border hierarchical data control, corresponding to Rule 48 (Data Sovereignty) and Meilai Patent 008.

[0041] Data classification: User contextual data is divided into core data (such as DNA binding information, core characteristics of personality models, and core data of emotional expression) and non-core data (such as general life behavior data and interaction data without emotional characteristics). Core data is prohibited from leaving the country after being tagged, while non-core data can be left the country with the user's authorization. Non-core data anonymization: Non-core data that needs to be exported is anonymized in a tiered manner, removing sensitive information such as user identification, geographic information, and emotional characteristics, retaining only general characteristics, and ensuring that the anonymized data cannot be associated with the user. Legal firewall construction: The system has a built-in database of data protection laws from various countries around the world. When a user needs to transfer non-core data abroad, the system automatically matches the laws of the user's home country and the data receiving country. Data transfer can only be completed when the laws of both places allow the data to leave the country. End-to-end encryption is used during data transfer. Overseas data retrieval interception: When an overseas third party attempts to retrieve user data, the system immediately triggers an interception mechanism, rejects the retrieval request, sends a security alert to the user, and saves the retrieval record to provide evidence for the user's legal rights protection.

[0042] The beneficial effects of this embodiment are: it realizes cross-border hierarchical management and control of user data, and prevents foreign third parties from forcibly accessing user data by using technologies such as prohibiting core data from leaving the country, de-identifying non-core data, and building legal firewalls, thus protecting users' data sovereignty.

[0043] Example 8: Virtual Human Cross-Device Interaction with One-Way Terminal Control This embodiment addresses the problems of poor cross-terminal adaptability of existing virtual humans and the ability of terminals to capture raw data. It enables virtual humans to be plug-and-play across terminals while ensuring terminal interaction security, corresponding to Rule 52 (one-way control) and Rule 58 (ideological sovereignty) of the rule layer, and corresponding to Meilai Patents 009 and 010.

[0044] Standardized adaptation of virtual humans: The custom virtual humans we build are packaged into standardized interactive components to adapt to various robot terminals (such as home companion robots, desktop virtual human terminals, and smart wearable devices), so as to realize plug-and-play virtual humans; Terminal permission allocation: After the terminal establishes a communication connection with the local hardware, the system only allocates virtual human interaction execution permissions to the terminal. The terminal can only receive encrypted interaction instructions sent by the local hardware (such as the virtual human's language, facial expressions, and action instructions), and has no right to obtain, store, or forward the user's original contextual data and personality model core data. One-time encrypted command communication: The communication between the virtual human and the terminal uses one-time encrypted commands. The local hardware generates a unique encryption key for each interaction. After the command is executed, the key is destroyed immediately. The terminal has no command caching permission, ensuring that the terminal cannot retain any interaction data. Terminal data capture and interception: The system monitors the terminal's operation behavior in real time. When it detects that the terminal is trying to capture or store raw data, it immediately interrupts the communication between the terminal and the local hardware, triggers a security alert, and adds the terminal to the blacklist to prohibit subsequent connections.

[0045] Based on the same inventive concept, please refer to Figure 2 This document shows a schematic block diagram of a personal scenario-based virtual human construction system 100, provided in an embodiment of this application, for executing the above-described personal scenario-based virtual human construction method. The personal scenario-based virtual human construction system 100 may include a communication unit 110, a machine-readable storage medium 120, and a processor 130.

[0046] In this embodiment, both the machine-readable storage medium 120 and the processor 130 are located within the dedicated virtual human creation system 100 based on personal contextual data and are configured separately. However, it should be understood that the machine-readable storage medium 120 may also be independent of the dedicated virtual human creation system 100 based on personal contextual data and may be accessed by the processor 130 via a bus interface. Alternatively, the machine-readable storage medium 120 may also be integrated into the processor 130 and may communicate and interact with external systems through the communication unit 110.

[0047] The processor 130 is the control center of the personalized virtual human creation system 100 based on personal contextual data. It connects to various parts of the system via various interfaces and lines. By running or executing software programs and / or modules stored in the machine-readable storage medium 120, and by calling data stored in the machine-readable storage medium 120, it performs various functions and processes data of the personalized virtual human creation system 100, thereby providing overall monitoring of the system. Optionally, the processor 130 may include one or more processing cores; for example, the processor 130 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor. The machine-readable storage medium 120 is used to store machine-executable instructions for executing the scheme of this application, and the processor 130 is used to execute the machine-executable instructions stored in the machine-readable storage medium 120 to implement the method for constructing a personal virtual human based on personal scenario data provided in the aforementioned method embodiments.

[0048] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

[0049] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for constructing a personalized virtual human based on individual contextual data, characterized in that: Includes the following steps: Collect users' personal contextual data and perform hierarchical labeling, ownership definition and standardized cleaning to obtain effective contextual data; Based on the aforementioned effective scenario-based data, a user-specific personality model is trained, equipped with an intent recognition module, and guided interaction logic is used to realize virtual human modeling, thus avoiding mind domestication; By uniquely binding the digital personality identity to the user's DNA, the ownership, inheritance, and recovery mechanisms for the digital personality are established. The original contextualized data and the exclusive personality model are stored locally in encrypted hardware, and hardware self-destruction and geofence protection mechanisms are set up to ensure that the data never leaves the domain. A three-level confirmation plus cooling-off period mechanism is set up for core operations related to digital personality, and a two-level confirmation mechanism is set up for general operations to prevent accidental and coerced operations. Monitor the interaction time between users and virtual humans, and adopt a graded anti-addiction strategy of consultation-based guidance + hardware-forced exit to realize the interaction rule of prioritizing reality; User data is classified as core / non-core. Core data is prohibited from leaving the country, while non-core data is de-identified and exported in compliance with the laws of both parties. An overseas access interception mechanism is also set up. The virtual human is packaged into standardized components to adapt to various terminals. Only the terminal is assigned interactive execution permissions, and the terminal is prohibited from obtaining, storing and forwarding raw data. 2.The method of claim 1, wherein the method further comprises: receiving a request for a virtual human from a user; and providing the virtual human to the user. The contextualized data collection involves collecting user behavior data, emotional expression data, and exclusive scene interaction data through local hardware audio and video acquisition units and sensor units. The three types of data are uniquely identified, tagged, and stored independently.

3. The method of claim 2, wherein the method further comprises: receiving a request for a virtual human; and providing the virtual human to the user. When multiple people collaboratively collect contextual data from the same user, a unique collection identifier is assigned to each collector, and the collector, collection time, and collection scenario of the data from each perspective are recorded. It is clear that the ownership of the data belongs to the user, and the collector only has the right to view the data with the authorization of the user. Moreover, the right to view the data can be modified by the user through the risk control mechanism.

4. The method of claim 2, wherein the method further comprises: The data cleaning process involves using data cleaning algorithms to deduplicatize, reduce noise, and standardize the collected raw contextual data, repairing noisy data, and converting unstructured audio, video, and voice data into structured data to obtain the effective contextual data.

5. The method of claim 1, wherein the method further comprises: The DNA binding and rights confirmation process involves extracting a user's DNA sample, writing a unique digital personality identifier sequence into a DNA molecule using epigenetic bit DNA storage technology, forming a unique digital personality-DNA binding relationship, and storing it in a DNA verification unit on local encrypted hardware. Based on this binding relationship, the process for inheriting and restoring the digital personality is set up, and the DNA-bound identity identifier serves as the technical basis for the legal confirmation of digital personality rights.

6. The method of claim 5, wherein the method further comprises: When using the virtual human, a two-factor authentication method of at least one fingerprint and retina is adopted. Users can register up to 10 fingerprints for random use. The specific finger for fingerprint verification is randomly assigned by the system. After successful verification, the user obtains full operation permissions for the virtual human, while other users can only obtain limited interaction / view permissions authorized by the user.

7. The method for constructing a personalized virtual human based on individual contextual data according to claim 5, characterized in that: When a digital personality is passed on, the heir must provide their own DNA sample and the DNA sample of the deceased. After the DNA verification unit matches and verifies the sample, the succession is completed using a three-level confirmation plus cooling-off period mechanism. If the local hardware data is damaged, the unique identifier of the digital personality can be read through the user's DNA sample to achieve accurate recovery of the personality model.

8. The method for constructing a personalized virtual human based on individual contextual data according to claim 1, characterized in that: In the guided personality modeling, the intent recognition module is a boasting recognition submodule. It extracts behavioral intent features from user statements using natural language processing technology. Statements without actual behavioral support are judged as non-genuine intents and are not included in the training of personality model decision features. The guided interaction logic is that when a user seeks decision-making advice, the virtual human does not directly output the decision result, but guides the user to think independently by asking targeted questions.

9. The method for constructing a personalized virtual human based on individual contextual data according to claim 1, characterized in that: The local encrypted storage uses the military-grade SM4 encryption algorithm to encrypt the data. The hardware has a built-in pressure sensor and a data erasure chip. When subjected to external force during disassembly, the data erasure chip is triggered to irreversibly erase all stored data within 1 second. The geographical range of data storage can be set through a geofencing module. When the hardware is taken away from the set range, it automatically enters a sleep state. Data backup only supports local backups initiated by users through a three-level confirmation mechanism; cloud backups are prohibited.

10. A personalized virtual human creation system based on individual contextual data, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the personalized virtual human construction method based on personal contextual data as described in any one of claims 1 to 9 by executing the machine-executable instructions.

Images