Property data sharing method and system based on quantum key AI writing large model

Abstract

The application discloses a property data sharing method and system based on a quantum key AI writing large model, comprising the following steps: S1, collecting property data, generating a dynamic key by quantum key distribution, and performing first encryption on sensitive fields; S2, performing second encryption by combining the dynamic key with the AES-256 algorithm, and storing in the Dream database; S3, constructing a user role key permission level to realize hierarchical permission control; S4, receiving user complaints and service requests, calling an AI writing large model to generate a structured report and performing double encryption; S5, attaching an access boundary token to the encrypted report, generating a shared data digest and uploading to the Dream transmission database; S6, transmitting the shared data digest to the property terminal, calling an AI clustering analysis to generate a credit evaluation; S7, performing quantum signature verification on the evaluation result, and encrypting and distributing to the supervision platform and the property terminal. The application realizes data full-process encryption, intelligent generation and safe sharing, and improves safety, efficiency and collaborative ability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information security and data sharing, and particularly relates to a property data sharing method and system based on a quantum key AI writing large model. BACKGROUND

[0002] With the development of digital economy and the improvement of residents' quality of life, the traditional property management mode has gradually failed to meet the high standard requirements of modern communities for service quality, response speed and data security. In recent years, driven by policy, many property companies have accelerated the construction of digital platforms, trying to improve management efficiency and user satisfaction through data sharing and system collaboration. However, in the actual promotion process, there are great limitations in traditional data processing architecture, security system and intelligent application capability, which have become key obstacles to the intelligent transformation of property services.

[0003] In the prior art, the collection, storage and sharing of property data mostly rely on traditional information management platforms. These platforms usually use plaintext transmission or low-strength encryption algorithms (such as MD5, SHA-1, etc.) for data protection, although they can achieve information recording and partial permission control at the basic level, but they have almost no anti-risk ability when facing new attack means such as quantum computing. The development of quantum computing has broken through the security assumption of classical encryption mechanism, and can crack the core logic of symmetric or asymmetric encryption algorithms in a very short time, so that sensitive fields in traditional property data systems, such as owner identity information, device running status, complaint records, etc., are at risk of being tampered with, leaked, or maliciously tampered with. In addition, most of the current property management systems lack end-to-end secure transmission mechanisms, resulting in data being exposed to unauthorized access channels in the collection, upload, processing, and sharing of multiple links.

[0004] Moreover, the existing property system relies on a large number of manual operations in data processing, especially in high-frequency tasks such as complaint handling, service response, credit evaluation, and information summary, a large number of report documents and service records still need to be manually sorted and uploaded, which not only is inefficient, but also is prone to data omission and human errors, seriously affecting the response speed and accuracy of services. Due to the lack of structured data standards and intelligent processing capabilities, the existing system is difficult to achieve unified information collection and automatic analysis, and cannot form an efficient property knowledge graph and credit analysis model, and it is also difficult to actively identify and respond to residents' demands.

[0005] In terms of data sharing and system collaboration, existing property platforms are mostly deployed as single units or regional platforms. When connecting with government platforms (such as WanShi Tong) and industry supervision platforms (such as the real estate market supervision platform), the following problems generally exist: first, the data interface security is poor, lacking a dynamic encryption token mechanism, which makes it easy to be stolen or tampered with during data transmission across platforms; second, the interface protocol design is not unified, the data format is not standardized, and there is a lack of unified permission authentication logic, making it difficult to achieve hierarchical sharing and fine-grained authorization; third, there is a lack of standardized identity authentication and access control mechanisms between systems, which creates security vulnerabilities for unauthorized users to access highly sensitive data, and can easily lead to security incidents such as the leakage of homeowner privacy and the misuse of enterprise data. Especially in multi-terminal collaborative scenarios, such as the interaction between property enterprise terminals, resident terminals, and government supervision terminals, the interaction requirements are becoming increasingly complex, and existing systems cannot support integrated data flow links, resulting in serious information silos.

[0006] In addition, although quantum key distribution technology and artificial intelligence writing large models have gradually developed and matured in the fields of information security and automatic text generation, there is no research that combines the two and applies them to property service data sharing scenarios. QKD technology can provide a theoretically unlistenable key transmission mechanism, suitable for establishing a high-strength encryption system, while AI writing large models have large-scale natural language generation capabilities and can replace manual implementation of automatic report generation, data semantic structuring, and other tasks. Unfortunately, existing property data platforms have not introduced such cutting-edge technologies into their core architecture, resulting in low levels of system security and intelligent processing, which cannot support future high-load, high-sensitivity, and multi-dimensional data interaction scenarios for property service needs.

[0007] Therefore, how to provide a property data sharing method and system based on quantum key AI writing large models is a problem that needs to be solved by those skilled in the art. SUMMARY

[0008] One object of the present application is to provide a property data sharing method and system based on quantum key AI writing large models. The present application fully integrates quantum key distribution encryption mechanisms and artificial intelligence writing large model technology, constructs an end-to-end encrypted transmission and intelligent document generation process for multiple roles and multiple terminals, and describes in detail the entire process of property data from collection, encryption, permission control, automatic report generation, data summary packaging, to credit evaluation and distribution, achieving data security sharing throughout the process, all roles, and all links, with the advantages of strong security, high intelligence, fast response, and strong cross-platform collaboration capabilities.

[0009] The property data sharing method based on quantum key AI writing large models according to the embodiments of the present application includes the following steps:

[0010] S1, collect cell property data through the terminal, generate dynamic key by quantum key distribution technology, and perform first encryption processing on sensitive fields;

[0011] S2, combine the dynamic key and the AES-256 algorithm to perform secondary encryption operation on the first encrypted field, and store the double-encrypted result in the Dream database;

[0012] S3, construct a key permission level based on user role, establish a user permission access path according to the provincial, municipal and street levels, realize access control through key binding mechanism, and ensure that authorized data content is obtained by users at all levels;

[0013] S4, after receiving the user uploaded complaint information and service request, calling the AI writing large model to automatically generate structured report text, and implementing double encryption strategy on the text content;

[0014] S5, performing quantum session key encryption processing on the structured report text, and attaching access boundary token, generating shared data digest, uploading to Dream database and entering data distribution preparation stage;

[0015] S6, package the shared data digest and access boundary token and transmit them to the property terminal, and trigger the AI writing large model to perform clustering analysis on historical data to generate real-time credit evaluation result;

[0016] S7, integrity verification of the credit evaluation result through quantum signature, and encryption distribution to the designated supervision platform and property terminal through token verification mechanism, realizing the whole process controllable data sharing link.

[0017] Optionally, the cell property data includes device code data D e , owner identity data D u , service interaction data D s .

[0018] Optionally, the S1 specifically includes:

[0019] S11, collect cell property data through the terminal device and perform preprocessing;

[0020] S12, call quantum key distribution mechanism to obtain real-time session key K q , and bind to the current property data collection task identifier T i , forming a bound key pair (K q , T i );

[0021] S13, field-level encryption is performed on the collected sensitive fields, and the key K q is used for the owner identity field D u and the service interaction field Ds , form a first encryption result:

[0022]

[0023] wherein, is a symmetric encryption function using the quantum session key K q , || is a field splicing operation, and C1 is the first encryption result.

[0024] Optionally, the preprocessing includes data cleaning and data conversion.

[0025] Optionally, the S2 specifically includes:

[0026] S21, based on the first encryption result C1, extracting the bound quantum session key K q , setting the symmetric encryption key as K a , calling the AES-256 encryption function to perform a second encryption operation on C1 to obtain a double encryption result of the sensitive field:

[0027]

[0028] wherein, is a symmetric encryption function using the key K a , and C2 is the double encryption result.

[0029] S22, writing the double encryption result into an encrypted data table in the Dream database and establishing a data mapping index I m :

[0030] I m = Index(T i , C2);

[0031] wherein, T i is a current property data collection task identifier, and Index(T i , C2) represents an index mapping relationship based on the task identifier T i and the double encryption result C2, which is used for subsequent data retrieval and permission verification.

[0032] Optionally, the S3 specifically includes:

[0033] S31, setting a user role set R = {r1, r2, r3}, wherein r1 represents a provincial role, r2 represents a municipal role, and r3 represents a street role, and constructing a permission level function for each role:

[0034]

[0035] wherein, W i is a role r iWeight of the managed area, θ i Sensitivity factor of the associated data field for the role, λ i Corresponding user role permission coefficient;

[0036] S32, according to the permission level coefficient λ i Generate permission key And the user identity hash value H u And access token T a Binding, constructing access control credential set Where Bind(·) represents the key binding operation function;

[0037] S33, when accessing the encrypted data table of the dream database, execute the credential verification function:

[0038]

[0039] Where V is the credential verification function, Verify(·) is the access verification function, I m Data mapping index, C2 is the double encryption result, if V=1, access is allowed, if V=0, access is denied.

[0040] Optionally, the S4 specifically comprises:

[0041] S41, when the terminal receives the service request data packet, extracts the associated user identifier U q , service content field F q And timestamp T q Form a request vector group [U q ||F q ||T q ], wherein || is the field splicing operation;

[0042] S42, input the request vector group into the AI writing large model, combine the semantic label vector V s of the double encryption result in the dream database, generate a structured report text:

[0043]

[0044] Where R f is the structured report text, Gen AI (·) is the generation function corresponding to the AI writing large model, φ i The weight coefficient of the i-th layer neural unit is W i The weight matrix is b i The bias term is σ, which is a nonlinear activation function, δ(V s ) is a content enhancement function of the semantic label on the report text, and n is the number of layers of the AI writing model.

[0045] S43. Invoke the double encryption process on the generated structured report text and execute the joint encryption expression:

[0046]

[0047] Among them, C r To encrypt the report text, To use quantum session key K q encryption function, To use key K a The executed symmetric encryption function, K q With K a These are the session key and the symmetric key, respectively.

[0048] Optionally, S5 specifically includes:

[0049] S51, Based on the encrypted result of the report text C r Call the quantum session key encryption function Generate encrypted identifier code Where ψ r This represents encrypted structured content that has been bound to a quantum key;

[0050] S52, Construct the access boundary token τ b =H(U q ||T q ||λ i ), where H(·) is the SHA-512 hash function, U q For user identification, T q For timestamps, λ i This represents the permission coefficient for the corresponding user role; || is the field concatenation operation.

[0051] S53, Combined Shared Data Summary Ω d And upload it to the pending distribution buffer table in the DM database:

[0052] Ω d =ψ r ||τ b ||χ f ;

[0053]

[0054] in, For AES-based symmetric key K a The message authentication code function is used for data integrity verification.

[0055] Optionally, S6 specifically includes:

[0056] S61, Shared data digest Ωd With access to the boundary token tau b Packed combination into a transmission entity package Pi p = Omega d || tau b , transmitted to the property terminal through an encrypted channel, wherein || is a field splicing operation;

[0057] S62, call the AI writing large model, extract the historical encrypted data set formed based on the index mapping relationship Index(T i , C2) in the dream database Where C 2,i is the ith double-encrypted data record, and m is the total number of historical data records;

[0058] Decrypt each data C 2,i and extract the semantic label vector V s,i , and form a semantic feature vector matrix M s after decryption:

[0059] M s = [delta(V s,1 ), delta(V s,2 ),..., delta(V s,m )];

[0060] Where delta(V s,j ) represents the content enhancement function of the semantic label of the jth record on the report text;

[0061] S63, input the semantic feature vector matrix into the clustering function embedded in the AI writing large model to generate real-time credit score results:

[0062]

[0063] Where beta k is the clustering weight coefficient of the kth semantic record, sigma is the ReLU activation function, W k is the weight matrix of the kth record, b k is the clustering bias vector, eta is the score intervention factor, tanh(·) is the hyperbolic tangent function, U is the clustering directionality matrix, is the mean representation of the entire semantic vector matrix, Norm(·) is the result normalization function, and Gamma c is the real-time generated credit score result.

[0064] The property data sharing system based on the quantum key AI writing large model according to the embodiments of the application comprises:

[0065] The data encryption module is used to collect property data and perform first encryption on sensitive fields using a quantum key;

[0066] A storage encryption module is configured to perform secondary encryption on sensitive fields in combination with a dynamic key and an AES-256 encryption algorithm and store the same into the Dream database;

[0067] A permission control module is configured to build a key permission level based on a user role, establish a user permission access path according to a provincial, municipal and street level, and realize hierarchical access control;

[0068] A report generation module is configured to receive a user request, call an AI writing large model to automatically generate a structured report text, and implement a double encryption strategy on the text content;

[0069] An abstract generation module is configured to perform quantum session key encryption processing on the structured report text, attach an access boundary token, generate a shared data abstract and upload the same into the Dream database;

[0070] A clustering analysis module is configured to transmit the shared data abstract and call an AI writing large model to generate a real-time credit evaluation result by clustering historical data;

[0071] A distribution verification module is configured to perform quantum signature verification on the credit evaluation result and encrypt the same for distribution to a supervision platform and a property terminal.

[0072] The present application has the following advantages:

[0073] Firstly, the present application introduces quantum key distribution technology to realize a dynamic encryption protection mechanism for property service data in the whole process of collection, transmission and storage, avoiding the security risks of traditional plaintext or low-intensity encryption methods when facing quantum computing attacks. By building a multi-role key permission control path and binding an access token mechanism, not only is the granular encryption processing of sensitive fields realized, but also different levels of users are ensured to access data according to the authorized range, effectively reducing the risk of data leakage and illegal operation, and significantly improving the security strength and credibility in the data sharing process.

[0074] Secondly, the present application combines an artificial intelligence writing large model to realize automatic structured text generation for typical business scenarios such as property complaint handling, service response and credit evaluation, replacing the original mode of relying on a large amount of manual writing. By inputting user request data and encrypted historical data, the system can efficiently generate a report text and form a credit evaluation result in combination with semantic labels and clustering algorithms, not only improving the response efficiency and processing accuracy of property companies in the service process, but also significantly reducing the delay and errors caused by manual operation, meeting the intelligent and efficient business operation requirements.

[0075] Finally, the application establishes a distribution mechanism with shared data digest as the core, combines quantum signature and token verification, and realizes cross-terminal encrypted distribution of structured reports and credit scoring results. While ensuring data integrity and verifiability, it realizes efficient data collaboration among regulatory platforms, property terminals and residents, effectively solving the problems of non-uniform interfaces, inaccurate permission control and untraceable data in existing systems. Overall, the application has achieved a comprehensive breakthrough in data security, intelligent processing and platform collaboration, providing an innovative technical path for the development of the property management industry towards high security, high intelligence and high collaboration. BRIEF DESCRIPTION OF DRAWINGS

[0076] The accompanying drawings are included to provide a further understanding of the application, and constitute a part of the specification, illustrate the application, and are used to explain the application together with the embodiments of the application, and do not constitute a limitation on the application. In the drawings:

[0077] Figure 1 The flowchart of the property data sharing method based on the quantum key AI writing large model proposed by the application. DETAILED DESCRIPTION

[0078] The application will now be described in further detail with reference to the drawings. These drawings are simplified schematic diagrams, and only illustrate the basic structure of the application in a schematic manner, and therefore only show the components related to the application.

[0079] Reference Figure 1 The property data sharing method based on the quantum key AI writing large model includes the following steps:

[0080] S1, collect community property data through the terminal, generate a dynamic key using quantum key distribution technology, and perform first encryption processing on sensitive fields;

[0081] S2, combine the dynamic key and the AES-256 algorithm to perform a second encryption operation on the first encrypted field, and store the double-encrypted result in the Dream database;

[0082] S3, construct a key permission level based on user roles, establish a user permission access path according to provincial, municipal and street levels, realize access control through a key binding mechanism, and ensure that authorized data content is obtained by users at all levels;

[0083] S4, after receiving the user uploaded complaint information and service request, an AI writing large model is called to automatically generate a structured report text, and a double encryption strategy is implemented on the text content;

[0084] S5, quantum session key encryption processing is performed on the structured report text, and an access boundary token is attached, a shared data digest is generated, uploaded to the Dream database, and enters the data distribution preparation stage;

[0085] S6, the shared data digest and the access boundary token are packaged and transmitted to the property terminal, and an AI writing large model is triggered to cluster and analyze historical data, to generate real-time credit evaluation results;

[0086] S7, the credit evaluation results are integrity-verified through quantum signature, and are encrypted and distributed to the designated supervision platform and the property terminal through the token verification mechanism, to realize a controllable data sharing link throughout the process.

[0087] The present application realizes technical innovation in data acquisition, encrypted storage, permission access, intelligent generation, data distribution, intelligent analysis and security verification, etc. in each key link, and comprehensively improves the security, intelligence and collaboration of property service data, breaks through the limitations of traditional systems in security protection, artificial efficiency and system integration, and has significant application promotion value and industrial transformation potential.

[0088] In the embodiment, the community property data includes equipment coding data D e , owner identity data D u , and service interaction data D s .

[0089] The present application refines the community property data into equipment coding data, owner identity data and service interaction data, not only enhances the structured degree of the data model, but also provides clear data dimension support for subsequent encryption processing, permission division and semantic analysis.

[0090] In the embodiment, the S1 specifically includes:

[0091] S11, the community property data is collected by a terminal device and preprocessed;

[0092] S12, a quantum key distribution mechanism is called to obtain a real-time session key K q , and is bound to a current property data collection task identifier T i , to form a bound key pair (K q , T i );

[0093] S13, the collected sensitive fields are field-level encrypted, the key K q is used on the owner identity field D u and the service interaction field D s , to form a first encryption result:

[0094]

[0095] wherein, is an encryption function using the quantum session key K q , || is a field concatenation operation, and C1 is a first encryption result.

[0096] The present application ensures security protection in the data generation stage by collecting cell property data by the terminal and performing first encryption on sensitive fields, and the dynamic key generated by quantum key distribution technology has unpredictability and non-replicability, which can effectively prevent data from being monitored, tampered or forged during collection and initial transmission, and improve the security and reliability of the data source.

[0097] In the embodiment, the preprocessing includes data cleaning and data conversion.

[0098] The present application introduces data cleaning and data conversion processes before data encryption processing, which helps to eliminate data redundancy, repair abnormal values and unify data format, improves data quality and consistency from the source, provides high-quality input for subsequent encryption, storage and AI generation process, and significantly reduces model bias and processing errors caused by messy data.

[0099] In the embodiment, the S2 specifically includes:

[0100] S21, based on the first encryption result C1, the bound quantum session key K q is extracted, the symmetric encryption key is set as K a , the AES-256 encryption function is called to perform a second encryption operation on C1, and a double encryption result of the sensitive field is obtained:

[0101]

[0102] wherein, is a symmetric encryption function executed using the key K a , and C2 is the double encryption result.

[0103] S22, the double encryption result is written into the encrypted data table of the dream database, and a data mapping index I m is established:

[0104] I m = Index(T i , C2);

[0105] wherein, T i is the current property data collection task identifier, and Index(T i , C2) indicates that the task identifier T iAn index mapping relationship is established with the double encryption result C2, and is used for subsequent data retrieval and permission verification.

[0106] The application combines dynamic keys and AES-256 algorithm to perform secondary encryption operation on sensitive fields, realizes a double encryption mechanism of "quantum key + symmetric encryption", and significantly enhances the anti-cracking ability of data in the storage link. Writing the encrypted data into a domestic Dream Database is conducive to building a self-controllable data infrastructure environment, and at the same time meets the national security standard of security creation.

[0107] In the embodiment, the S3 specifically includes:

[0108] S31, set a user role set R={r1, r2, r3}, wherein r1 represents a provincial role, r2 represents a municipal role, and r3 represents a street role, and a permission level function is constructed for each role:

[0109]

[0110] Wherein, W i is the weight of the region managed by the role r i , theta i is a sensitive factor of the role associated data field, and lambda i is a user role permission coefficient corresponding to the user role;

[0111] S32, according to the permission level coefficient lambda i , a permission key K is generated u , and is bound with a user identity hash value H a and an access token T m to construct an access control credential set , wherein Bind(·) represents a key binding operation function;

[0112] S33, when accessing the Dream Database encrypted data table, a credential verification function is executed:

[0113]

[0114] Wherein, V is a credential verification function, Verify(·) is an access verification function, I m is a data mapping index, C2 is a double encryption result, if V=1, access is allowed, and if V=0, access is refused.

[0115] The application constructs a key permission level based on user roles and implements access control through a key binding mechanism, ensures that data access is only performed within the authorized range, truly realizes the "minimum available permission" principle, prevents unauthorized access and illegal operation, effectively isolates the data boundaries between different role users, and improves the security and accuracy of data scheduling.

[0116] In this embodiment, S4 specifically includes:

[0117] S41, when the terminal receives the service request data packet, extracts the associated user identifier U q , service content field F q and timestamp T q , form a request vector group [U q ||F q ||T q ], wherein || is the field splicing operation;

[0118] S42, input the request vector group into the AI writing large model, combine the semantic label vector V s of the double encryption result in the dream database, and generate a structured report text:

[0119]

[0120] Wherein, R f is the structured report text, Gen AI (·) is the generation function corresponding to the AI writing large model, φ i is the weight coefficient of the i-th layer neural unit, W i is the weight matrix, b i is the bias term, σ is the nonlinear activation function, δ(V s ) is the content enhancement function of the semantic label to the report text, and n is the number of layers of the AI writing model;

[0121] S43, call the double encryption process for the generated structured report text, and execute the joint encryption expression:

[0122]

[0123] Wherein, C r is the encryption result of the report text, is an encryption function using quantum session key K q , is a symmetric encryption function executed using key K a , K q and K a are session key and symmetric key respectively.

[0124] The application inputs the user uploaded complaint and service request data into the AI writing large model, automatically generates a structured report, and executes a double encryption strategy, which not only improves the report generation efficiency, but also guarantees the confidentiality and non-tamperability of the text content, realizes the data processing path of coexistence of high efficiency and high safety, and reduces the error risk caused by manual participation.

[0125] In this embodiment, S5 specifically includes:

[0126] S51, based on the report text encryption result C r , call quantum session key encryption function Generate encrypted identification code Where ψ r Indicates that the encrypted structured content has bound quantum keys;

[0127] S52, build access boundary token τ b = H(U q ||T q ||λ i ), where H(·) is the SHA-512 hash function, U q is the user identification, T q is the timestamp, λ i is the corresponding user role permission coefficient, and || is the field splicing operation;

[0128] S53, combine shared data digest Ω d , and upload to the buffer table of the Dangmeng database to be distributed:

[0129] Ω d = ψ r ||τ b ||χ f ;

[0130]

[0131] Where, Is the message authentication code function based on the AES symmetric key K a , used for data integrity verification.

[0132] The present application encrypts the structured report by quantum session key, adds access boundary token, generates shared data digest and enters data distribution preparation stage, which can ensure that the data to be distributed is completely marked and authenticated in terms of content security, access permission, operation legality, etc., and provides mechanism support for subsequent trusted transmission and audit traceability.

[0133] In the embodiment, the S6 specifically comprises:

[0134] S61, pack and combine the shared data digest Ω d And the access boundary token τ b As a transmission entity package Π p = Ω d ||τ b , transmit to the property terminal through an encrypted channel, wherein || is the field splicing operation;

[0135] S62, call the AI writing large model, extract the index mapping relationship Index (T i , C2) formed by the historical encrypted data set Where C 2,i is the i-th double-encrypted data record, and m is the total number of historical data records.

[0136] For each data C 2,i , decryption and semantic label vector V s,i are extracted, and the semantic feature vector matrix M s is formed after decryption:

[0137] M s = [delta (V s,1 ), delta (V s,2 ),..., delta (V s,m )];

[0138] Where delta (V s,j ) represents the content enhancement function of the semantic label of the j-th record on the report text.

[0139] S63, input the semantic feature vector matrix into the clustering function embedded in the AI writing large model to generate real-time credit score results:

[0140]

[0141] Where beta k is the clustering weight coefficient of the k-th semantic record, sigma is the ReLU activation function, W k is the weight matrix of the k-th record, b k is the clustering bias vector, eta is the score intervention factor, tanh (.) is the hyperbolic tangent function, U is the clustering directionality matrix, is the mean representation of the entire semantic vector matrix, Norm (.) is the result normalization function, and Gamma c is the real-time generated credit score result.

[0142] The present application packages the shared data summary and transmits it to the property terminal, and triggers the AI writing large model to perform clustering analysis on historical data, which can accurately model based on historical service records, dynamically generate personalized credit evaluation results, improve the intelligent level and risk control ability of property supervision, and realize the value transformation from data use to data insight.

[0143] The property data sharing system based on quantum key AI writing large model comprises:

[0144] A data encryption module is used to collect property data and encrypt sensitive fields using quantum keys for the first time.

[0145] A storage encryption module is configured to perform secondary encryption on sensitive fields in combination with a dynamic key and an AES-256 encryption algorithm and store the same in the Dream database;

[0146] A permission control module is configured to construct a key permission level based on a user role, establish a user permission access path according to a provincial, municipal and street level, and implement hierarchical access control;

[0147] A report generation module is configured to receive a user request, call an AI writing large model to automatically generate a structured report text, and implement a double encryption strategy on the text content;

[0148] An abstract generation module is configured to perform quantum session key encryption processing on the structured report text, attach an access boundary token, generate a shared data abstract, and upload the same to the Dream database;

[0149] A clustering analysis module is configured to transmit the shared data abstract and call an AI writing large model to generate real-time credit evaluation results by clustering historical data;

[0150] A distribution verification module is configured to perform quantum signature verification on the credit evaluation results and encrypt and distribute the same to a supervision platform and a property terminal.

[0151] Embodiment 1

[0152] In order to verify the feasibility of the present application in implementation, the present application is applied to a certain large-scale intelligent community property management platform. The community is a typical high-density residential area with more than 4500 households of permanent residents, a property service area of more than 500,000 square meters, covering elevator equipment, fire facilities, underground garage, intelligent access control and other types of equipment terminals, with an average of more than 2800 property interaction data per day. The platform business covers key scenarios such as owner complaints, facility repair, credit evaluation, payment records, and inspection records. The original property management system has outstanding problems in safety, efficiency and collaboration.

[0153] Firstly, the original data protection mechanism of the platform uses static keys and weak encryption algorithms to encrypt only the basic information of the owner, contact information, complaint content and other high-sensitive data, resulting in 3 internal permission overreach leading to sensitive information leakage events in 2024. Secondly, the complaint and service processing process completely relies on manual operation, and the report generation needs to be manually written, uploaded and approved, with an average processing period of 37 hours, often causing repeated submission of resident complaints and delayed service response. Thirdly, when the platform is connected with the superior supervision system, due to the lack of unified permission boundary definition and interface encryption mechanism, the format error rate of the transmitted data reaches 7.6%, which cannot guarantee the accuracy and timeliness of the key indicators.

[0154] After the deployment of the present application, a dynamic encryption mechanism based on quantum key distribution is integrated in the platform. All sensitive fields related to the identity of the owner, the identification of the equipment, and the service interaction are first encrypted by dynamic keys at the collection link, and then written into the domestic Damos database after double encryption by combining the AES-256 algorithm, replacing the original static encryption scheme. The platform simultaneously performs cleaning and standardized conversion on the collected data, unifies the field format, and improves the accuracy of subsequent AI model processing. In handling complaint business, the system calls an AI writing large model, automatically extracts key information from the text uploaded by the owner and generates a structured report, the average processing time of the model is controlled within 1 second, and the accuracy rate reaches 98.4%. For example, in a complaint about public lighting damage in mid-January 2025, the system identified keywords such as "fault type", "location", "reporting time" from the text, automatically generated a maintenance information form and submitted it to the maintenance unit, and finally the processing cycle was compressed to complete within 6 hours, significantly improving user satisfaction.

[0155] At the same time, the platform establishes a key permission level system according to user roles (platform administrator, property manager, and area supervisor), realizes hierarchical permission control by binding key fingerprints and access tokens. In the data sharing process, the system performs quantum key encryption on the structured report and adds an access boundary token to generate a data digest uploaded to the buffer database; then the data digest is packaged and transmitted to the supervision end through a dynamic interface, and the AI model is triggered to perform clustering analysis and scoring on historical complaints and maintenance records, outputting enterprise credit evaluation results for supervisors to review.

[0156] Since the deployment of the present application, the platform has completed a total of 82,379 structured data transmissions, generating more than 800 structured reports per day, shortening the complaint response cycle from an average of 37 hours to 6.4 hours, and reducing the service processing timeout rate from 21.7% to 4.5%. The credit evaluation accuracy rate has improved from no intelligent identification to 96.8%, and the supervision platform feedback is consistent that the evaluation results are more valuable for reference. In addition, according to system records, during January-February, the platform has not occurred any data unauthorized access and transmission errors, and the data integrity and security has reached 100%. The following are the key operation data collected by the property platform in actual operation, reflecting the implementation effect of the present application:

[0157] Table 1 Statistics of application effect of smart community property platform

[0158]

[0159]

[0160] To sum up, by deploying the application, high-strength security protection of property service data is realized, the overall processing efficiency and user experience of the platform are greatly improved, practical problems such as weak data security, low-efficiency manual processing and cross-system sharing obstacles existing in the traditional system are solved, and the practicability and generalizability of the method in complex scenarios are fully verified.

[0161] The above merely describes preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can make equivalent replacements or changes within the technical range disclosed by the present application according to the technical solution and inventive concept of the present application, which should be covered within the protection scope of the present application.

Claims

1. A property data sharing method based on a quantum key distribution AI writing model, characterized in that: Includes the following steps: S1. Collect community property data through the terminal, generate dynamic keys using quantum key distribution technology, and perform initial encryption processing on sensitive fields; S1 specifically includes: S11. Collect community property data through terminal equipment and preprocess it; the community property data includes device code data D. e Owner identity data D u Service interaction data D s The preprocessing includes data cleaning and data transformation. S12. Invoke the quantum key distribution mechanism to obtain the real-time session key K. q And bind it to the current property data collection task identifier T i To form a binding key pair (K q ,T i ); S13. Encrypt the collected sensitive fields at the field level using key K. q Applying to the owner identity field D u Service interaction field D s This forms the initial encryption result C1; S2. Combining the symmetric encryption key and the AES-256 algorithm, perform a second encryption operation on the first encrypted field, and store the double encryption result in the Dameng database; S2 specifically includes: S21. Based on the initial encryption result C1, extract the bound quantum session key K. q Set the symmetric encryption key to K. a The AES-256 encryption function is called to perform a second encryption operation on C1, resulting in the double-encrypted result C2 of the sensitive field. S22. Write the double encryption result C2 into the encrypted data table of the Dameng database, and combine it with the current property data collection task identifier T. i Establish data mapping index I m This is used for subsequent data retrieval and permission verification; S3. Construct a key permission level based on user roles, establish user permission access paths according to provincial, municipal and street levels, and implement access control through a key binding mechanism to ensure that users at all levels can obtain authorized data content; S3 specifically includes: S31. Define a user role set R = {r1, r2, r3}, where r1 represents a provincial-level role, r2 represents a city-level role, and r3 represents a street-level role. For each role, construct a permission level function: Among them, W i For the character r i The weight of the managed area, θ i λ is a sensitive factor for the role-related data field. i This corresponds to the user's role and permission coefficient. S32, Based on the permission level coefficient λ i Generate permission key and the user identity hash value H u and access token T a Binding, constructing access control credential set Where Bind(·) represents the key binding operation function; S33. When accessing the encrypted data table in the Dameng database, execute the credential verification function: Where V is the credential verification function, Verify(·) is the access verification function, and I m C2 is the data mapping index, and V=1 indicates that access is allowed, while V=0 indicates that access is denied. S4. Upon receiving user-uploaded complaint information and service requests, the AI ​​writing model is invoked to automatically generate structured report text, and a dual encryption strategy is implemented on the text content. S4 specifically includes: S41. When the terminal receives a service request data packet, extract the associated user identifier U from it. q Service content field F q With timestamp T q , forming a request vector group [U q ||F q ||T q ], where || is the field concatenation operation; S42. Input the request vector group into the AI ​​writing model, and combine it with the semantic tag vector V of the double-encrypted result in the Dameng database. s Generate structured report text: Among them, R f For structured report text, Gen AI (·) represents the generator function corresponding to the large AI writing model, φ i W represents the weight coefficient of the i-th layer neural unit. i Let b be the weight matrix. i Here, σ is the bias term, δ(V) is the nonlinear activation function, and δ(V) is the bias term s ) is the semantic tagging function for enhancing the content of the report text, and n is the number of layers in the AI ​​writing model; S43. Invoke the double encryption process on the generated structured report text and execute the joint encryption expression: Among them, C r To encrypt the report text, To use quantum session key K q encryption function, To use key K a The executed symmetric encryption function, K q With K a These are the session key and the symmetric key, respectively. S5. Perform quantum session key encryption on the structured report text, attach an access boundary token, generate a shared data digest, upload it to the DM database, and enter the data distribution preparation stage; S5 specifically includes: S51, Based on the encrypted result of the report text C r Call the quantum session key encryption function Generate encrypted identifier code Where ψ r This represents encrypted structured content that has been bound to a quantum key; S52, Construct the access boundary token τ b =H(U q ||T q ||λ i ), where H(·) is the SHA-512 hash function, U q For user identification, T q For timestamps, λ i This represents the permission coefficient for the corresponding user role; || is the field concatenation operation. S53, Combined Shared Data Summary Ω d And upload it to the pending distribution buffer table in the DM database: Oh d =ψ r ||t b ||x f ; in, For AES-based symmetric key K a The message authentication code function is used for data integrity verification. S6. Package and transmit the shared data summary and access boundary token to the property terminal, and trigger the AI ​​writing big model to perform cluster analysis on historical data to generate real-time credit assessment results; S6 specifically includes: S61, Shared data digest Ω d With access boundary token τ b Packaging and combining into a transmission entity packet Π p =Ω d ||τ b The data is transmitted to the property terminal via an encrypted channel, where || represents a field concatenation operation. S62. Call the AI ​​writing model to extract Index(T) from the DaMeng database based on the index mapping relationship. i The historical encrypted data set H = {C2) formed by C2) 2,1 C 2,2 ,...,C 2,m }, where C 2,i Let be the i-th data record that has undergone double encryption, and m be the total number of historical data records; For each data C 2,i Perform decryption and extract semantic label vector V s,i After decryption, a semantic feature vector matrix M is formed. s : M s =[δ(V s,1 ),δ(V s,2 ),...,δ(V s,m )]; Wherein, δ(V s,j ) represents the semantic tag of the j-th record that enhances the content of the report text; S63. Input the semantic feature vector matrix into the clustering function embedded in the AI ​​writing model to generate real-time credit scoring results: Where, β k Let W be the clustering weight coefficient for the k-th semantic record, σ be the ReLU activation function, and W be the weight coefficient for the k-th semantic record. k Let b be the weight matrix for the k-th record. k Let η be the clustering bias vector, η be the rating intervention factor, tanh(·) be the hyperbolic tangent function, and U be the clustering directionality matrix. Let Γ be the mean representation of the entire semantic vector matrix, Norm(·) be the result normalization function, and Γ be the mean of the set of semantic vector matrices. c For real-time generated credit score results; S7. The credit assessment results are verified for integrity using quantum signatures, and encrypted distribution to designated regulatory platforms and property terminals is achieved through a token verification mechanism, realizing a fully controllable data sharing link.

2. A property data sharing system based on a quantum key AI writing model, implementing the property data sharing method based on a quantum key AI writing model as described in claim 1, characterized in that: include: The data encryption module is used to collect property data and use quantum keys to initially encrypt sensitive fields, including: Community property data is collected through terminal devices and preprocessed; the community property data includes device code data D. e Owner identity data D u Service interaction data D s The preprocessing includes data cleaning and data transformation. Obtain the real-time session key K by calling the quantum key distribution mechanism. q And bind it to the current property data collection task identifier T i To form a binding key pair (K q ,T i ); Sensitive fields collected are encrypted at the field level using key K. q Applying to the owner identity field D u Service interaction field D s This forms the initial encryption result C1; The storage encryption module, used to perform secondary encryption on sensitive fields by combining a symmetric encryption key with the AES-256 encryption algorithm and then store them in the DM database, includes: Based on the initial encryption result C1, the bound quantum session key K is extracted. q Set the symmetric encryption key to K. a The AES-256 encryption function is called to perform a second encryption operation on C1, resulting in the double-encrypted result C2 of the sensitive field. Write the double-encrypted result C2 into the encrypted data table in the Dameng database, and combine it with the current property data collection task identifier T. i Establish data mapping index I m This is used for subsequent data retrieval and permission verification; The access control module is used to build key permission levels based on user roles, establish user access paths according to provincial, municipal, and street levels, and implement hierarchical access control, including: Define a user role set R = {r1, r2, r3}, where r1 represents a provincial-level role, r2 represents a city-level role, and r3 represents a street-level role. For each role, construct a permission level function: Among them, W i For the character r i The weight of the managed area, θ i λ is a sensitive factor for the role-related data field. i This corresponds to the user's role and permission coefficient. Based on the permission level coefficient λ i Generate permission key and the user identity hash value H u and access token T a Binding, constructing access control credential set Where Bind(·) represents the key binding operation function; When accessing an encrypted data table in the DM database, execute the credential verification function: Where V is the credential verification function, Verify(·) is the access verification function, and I m C2 is the data mapping index, and V=1 indicates that access is allowed, while V=0 indicates that access is denied. The report generation module receives user requests, invokes an AI writing model to automatically generate structured report text, and implements a dual encryption strategy for the text content, including: When the terminal receives a service request data packet, it extracts the associated user identifier U. q Service content field F q With timestamp T q , forming a request vector group [U q ||F q ||T q ], where || is the field concatenation operation; The request vector group is input into the AI ​​writing model, combined with the semantic tag vector V of the double-encrypted result in the Dameng database. s Generate structured report text: Among them, R f For structured report text, Gen AI (·) represents the generator function corresponding to the large AI writing model, φ i W represents the weight coefficient of the i-th layer neural unit. i Let b be the weight matrix. i Here, σ is the bias term, δ(V) is the nonlinear activation function, and δ(V) is the bias term s ) is the semantic tagging function for enhancing the content of the report text, and n is the number of layers in the AI ​​writing model; The generated structured report text is subjected to a double encryption process, and the joint encryption expression is executed: Among them, C r To encrypt the report text, To use quantum session key K q encryption function, To use key K a The executed symmetric encryption function, K q With K a These are the session key and the symmetric key, respectively. The summary generation module performs quantum session key encryption on structured report text, appends access boundary tokens, generates a shared data summary, and uploads it to the Chuanmeng database, including: Based on the encrypted result of the report text C r Call the quantum session key encryption function Generate encrypted identifier code Where ψ r This represents encrypted structured content that has been bound to a quantum key; Construct access boundary token τ b =H(U q ||T q ||λ i ), where H(·) is the SHA-512 hash function, U q For user identification, T q For timestamps, λ i This represents the permission coefficient for the corresponding user role; || is the field concatenation operation. Combined Shared Data Summary Ω d And upload it to the pending distribution buffer table in the DM database: Oh d =ψ r ||t b ||x f ; in, For AES-based symmetric key K a The message authentication code function is used for data integrity verification. The clustering analysis module is used to transmit shared data summaries and call upon the AI ​​writing model to cluster historical data to generate real-time credit assessment results, including: Shared data digest Ω d With access boundary token τ b Packaging and combining into a transmission entity packet Π p =Ω d ||τ b The data is transmitted to the property terminal via an encrypted channel, where || represents a field concatenation operation. The AI ​​writing model is invoked to extract Index(T) from the DM database based on the index mapping relationship. i The historical encrypted data set formed by C2) Where C 2,i Let be the i-th data record that has undergone double encryption, and m be the total number of historical data records; For each data C 2,i Perform decryption and extract semantic label vector V s,i After decryption, a semantic feature vector matrix M is formed. s : M s =[δ(V s,1 ),δ(V s,2 ),...,δ(V s,m )]; Wherein, δ(V s,j ) represents the semantic tag of the j-th record that enhances the content of the report text; The semantic feature vector matrix is ​​input into the clustering function embedded in the AI ​​writing model to generate real-time credit scoring results. Where, β k Let W be the clustering weight coefficient for the k-th semantic record, σ be the ReLU activation function, and W be the weight coefficient for the k-th semantic record. k Let b be the weight matrix for the k-th record. k Let η be the clustering bias vector, η be the rating intervention factor, tanh(·) be the hyperbolic tangent function, and U be the clustering directionality matrix. Let Γ be the mean representation of the entire semantic vector matrix, Norm(·) be the result normalization function, and Γ be the mean of the set of semantic vector matrices. c For real-time generated credit score results; The distribution verification module is used to verify the credit assessment results using quantum signatures and then encrypt and distribute them to the regulatory platform and property terminals.

Images