Online house and real estate leasing system and method based on block chain technology

Abstract

The invention discloses an online house and real estate lease system and method based on a block chain technology. The system comprises a block chain basic service module, a user management module, a house resource management module, an intelligent recommendation module, a lease transaction module, a lease cancellation equipment damage identification module, a data security module and a system management module. According to the invention, non-tampering evidence storage of key data in the whole leasing process is realized through the block chain technology, the multi-node consensus and digital signature are combined, the problems of false house resource information and untrusted transaction data are solved, and the credibility between a tenant and a landlord is improved. The intelligent recommendation module fuses user basic information, preference and behavior features, combines structured and unstructured house source features, and realizes accurate house source matching, reduces user screening cost and improves lease matching efficiency through multi-dimensional score fusion and diversity adjustment.

Description

Technical Field

[0001] This invention relates to the field of real estate leasing service technology, specifically to an online housing and real estate leasing system and method based on blockchain technology. Background Technology

[0002] The current online housing and real estate rental market faces the following problems: First, the authenticity of housing information is difficult to verify, with frequent issues such as fake listings and ambiguous property rights information, leading to a lack of trust between tenants and landlords. Second, housing matching efficiency is low; traditional recommendation methods rely heavily on basic screening criteria and cannot combine deep user preferences and behavioral characteristics to achieve accurate matching, resulting in wasted resources and a poor user experience. Third, the assessment of equipment damage during the move-out process lacks objective standards, relying on manual inspection, which easily leads to disputes, and the assessment results are difficult to trace and prove. Fourth, rental transaction data (such as contract texts, payment records, and performance status) is stored on centralized platforms, which are at risk of being tampered with or lost, and subsequent dispute resolution lacks credible evidence. Fifth, consensus mechanisms are not adaptable enough; most existing rental platforms have not introduced or use simple consensus mechanisms, which cannot guarantee the security and efficiency of data collaboration among multiple participants (landlords, tenants, regulatory agencies, and banks).

[0003] In existing technologies, some rental platforms have attempted to introduce blockchain technology to solve data storage problems, but have not achieved deep integration with intelligent recommendation and equipment damage identification. The application of intelligent recommendation algorithms in the real estate sector is mostly limited to structured feature matching, ignoring the value of unstructured data (property images, user behavior sequences). Equipment damage identification during property lease terminations largely relies on manual or simple image comparison, lacking intelligent and automated judgment solutions. Furthermore, existing blockchain consensus mechanisms suffer from low throughput and insufficient fault tolerance in rental scenarios, making it difficult to meet the needs of high-frequency transactions and multi-node collaboration. Therefore, there is an urgent need for an online housing and real estate rental system and method based on blockchain technology to address the aforementioned problems. Summary of the Invention

[0004] The purpose of this invention is to provide an online housing and real estate rental system and method based on blockchain technology to solve the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an online housing and real estate rental system based on blockchain technology, comprising a blockchain basic service module, a user management module, a housing management module, an intelligent recommendation module, a rental transaction module, a return-of-lease equipment damage identification module, a data security module, and a system management module;

[0006] The blockchain infrastructure service module is responsible for realizing the immutable notarization of key data throughout the entire leasing process, multi-node collaboration, and automated contract execution. It includes a consortium blockchain network submodule, a consensus algorithm submodule, a smart contract engine submodule, and a cross-chain gateway submodule.

[0007] The user management module is responsible for user identity authentication, permission allocation and credit management, ensuring the credibility of user identity and the traceability of behavior, including an identity authentication sub-module and a credit management sub-module;

[0008] The housing management module realizes housing information collection, authenticity verification and on-chain storage to ensure that housing information is credible and traceable. It includes a housing information collection sub-module, a housing authenticity verification sub-module and a housing on-chain storage module.

[0009] The intelligent recommendation module, based on user profiles and property features, achieves accurate property matching and improves rental matching efficiency. It includes a user profile construction submodule, a property feature modeling submodule, and a recommendation strategy execution submodule.

[0010] The leasing transaction module enables online operation and reliable evidence storage for the entire leasing process, including a transaction matching submodule, a contract signing submodule, and a payment settlement submodule.

[0011] The equipment damage identification module for lease termination enables automated identification and quantitative assessment of equipment damage during lease termination, reducing lease termination disputes. It includes an image acquisition and preprocessing submodule, an equipment positioning and identification submodule, a damage detection submodule, and a damage assessment and reporting submodule.

[0012] The data security module ensures the security and privacy of system data during collection, transmission, storage, and use, and includes a data encryption submodule, an access control submodule, and a security detection and vulnerability repair submodule.

[0013] The system management module is responsible for the daily maintenance, parameter configuration, and log management of the system to ensure stable system operation. It includes a user account management submodule, a parameter configuration submodule, and a log management submodule.

[0014] Preferably, the consortium blockchain network sub-module includes platform operation nodes, government regulatory nodes (such as real estate registration centers and market supervision departments), financial service nodes (banks and insurance companies), and third-party authoritative nodes (notary offices and law firms). It adopts a permissioned blockchain architecture, where only authorized nodes can participate in consensus and data access. It is built on Hyperledger Fabric 2.5+, supports dynamic node addition and permission adjustment, and adapts to the collaborative needs of multiple participants in leasing scenarios.

[0015] The consensus algorithm submodule adopts an optimized "fault-tolerant high-throughput consensus algorithm" that supports the number of nodes n≥3f+1 (f is the number of Byzantine nodes), meeting the system's fault tolerance requirements;

[0016] Fault-tolerant high-throughput consensus algorithm process:

[0017] Leader election: Nodes trigger an election by randomizing the election timeout. Candidate nodes broadcast a digitally signed "VoteRequest" containing the candidate node ID, current term, and the node's latest log index and term. When a candidate node collects ≥2f+1 signed "VoteResponses", it generates an "Election-Certificate", becomes the new leader, and upgrades the term.

[0018] Log replication: After receiving a request from a client (landlord / tenant), the Leader generates a log entry containing term, index, and request content, and broadcasts "AppendProposal". Followers verify the Leader's signature and log prefix consistency (prevIndex matches the local log), and then return a signed "Prepare". After collecting ≥2f+1 "Prepare" entries, the Leader generates a "Prepare-Certificate" and broadcasts "CommitRequest". Nodes verify the validity of the "Prepare-Certificate" and return "Commit". When any node collects ≥2f+1 "Commit" entries, it generates a "Commit-Certificate", marks the log entry as "committed", and executes the request.

[0019] View Change: If the Leader fails to advance the log, initiate a conflict proposal, or double-sign, the node broadcasts a ViewChange containing its highest Prepared log information and the Prepare-Certificate; the new Leader candidate collects ≥2f+1 ViewChanges, confirms the highest Prepared log, publishes NewView, and starts the log replication process for the new term;

[0020] Optimized design: The BLS aggregate signature technology is adopted to merge the signatures of multiple nodes into a single signature, reducing message complexity (from O(n2) to O(n)), while supporting batch processing of log entries (1-100 requests / entries) to improve system throughput;

[0021] The smart contract engine submodule includes a lease agreement contract, a payment settlement contract, a deposit escrow contract, and a dispute arbitration contract.

[0022] Rental Agreement Contract: Automatically generates standardized electronic contract templates, supporting landlords and tenants to fill in and modify online. After the contract is signed, the hash value is calculated and uploaded to the blockchain, while the original contract is stored off-chain in IPFS (InterPlanetary File System).

[0023] Payment and settlement contract: Based on the rent amount and payment cycle agreed in the contract, the rent transfer is automatically triggered from the tenant's account to the landlord's account, and the payment transaction hash is recorded on the blockchain; it supports automatic reminders for overdue payments, and triggers the insurance compensation process if the overdue period exceeds the agreed time;

[0024] Escrow Agreement: The deposit paid by the tenant is locked in a smart contract account and will be automatically refunded when there is no dispute upon termination of the lease and inspection. If there is a dispute over damage, the corresponding amount will be automatically deducted from the landlord's account based on the damage identification results and arbitration conclusion.

[0025] Dispute Arbitration Contract: Receives the application and evidence hash from the party initiating the dispute (landlord / tenant), automatically notifies the arbitration node (third-party law firm / notary office) to make a ruling, and stores the ruling result and evidence hash on the blockchain for evidence storage;

[0026] The cross-chain gateway submodule connects to external government blockchains (such as local real estate registration blockchains) and financial blockchains (bank payment blockchains) to achieve cross-chain data interaction. For example, it can obtain property ownership verification results from the real estate registration blockchain and synchronize rent payment status from the bank payment blockchain. It adopts the WeCross cross-chain framework to support data format conversion and secure transmission between blockchains, ensuring the consistency and credibility of cross-chain data.

[0027] Preferably, the identity authentication submodule includes individual users and enterprise users;

[0028] For individual users: It connects to the Ministry of Public Security's "Internet and Trusted Identity Authentication" (CTID) platform, and completes three-factor authentication through facial recognition, comparison of the front and back of the ID card, and liveness detection; it also supplements the four-factor authentication of UnionPay bank cards, namely, consistency verification of name, ID card number, bank card number, and mobile phone number. After successful authentication, a unique blockchain digital identity DID is generated.

[0029] Enterprise users (such as institutional landlords): Submit business license, legal representative's ID card, and real estate ownership certificate. After manual review and verification by the government's enterprise credit information disclosure system, an enterprise DID will be generated. Enterprise employees need to configure additional role permissions (RBAC model) to operate, and operation records are linked to the enterprise DID and uploaded to the blockchain.

[0030] The credit management submodule includes credit scoring and handling of credit defaults;

[0031] Credit scoring: Based on users' on-chain behavior (such as timely rent payment, contract performance, and records of lease termination disputes) and external credit data (connected to "Credit China" and the list of dishonest persons subject to enforcement by the court), a credit scoring model (maximum score of 1000 points) is constructed, and the scoring results are updated in real time and associated with the user's DID;

[0032] Handling of Dishonesty: If a user engages in false listings, malicious rent arrears, damages property equipment, or other dishonest behavior, their DID will be marked as "dishonest" and synchronized to all nodes; dishonest users will be unable to initiate new rental requests, and already initiated requests will be automatically blocked.

[0033] Preferably, in the housing information collection submodule, landlords upload basic housing information (address, apartment type, area, rent, payment method, and supporting facilities), multimedia information (housing photos, VR panoramic videos, and surrounding environment videos), and property information (scanned copy of the property ownership certificate and the owner's identity certificate) through the system. The system provides upload guidelines, requiring housing photos to include close-ups (0.5-1.5m distance) of the four corners of the house and major equipment (bed, sofa, air conditioner, refrigerator), and VR videos to fully cover every room of the house. The image resolution should be no less than 1600px, and backlighting and blurriness should be avoided.

[0034] The property authenticity verification submodule connects to the government API of the real estate registration center. By inputting the property's real estate unit number and the owner's ID number / corporate unified social credit code, it retrieves data on the property's location, area, mortgage status, seizure information, and co-ownership status to verify the legality and availability of the property rights. If the city where the property is located does not have access to the government API, it queries the property's ownership information through a partner bank (a large state-owned bank). The landlord must sign an authorization agreement allowing the bank to connect to the real estate registration center through its professional system, and the verification results are synchronized to the system. A ViT-based image recognition model is used to detect whether property images show signs of Photoshop manipulation (such as edge tampering or color inconsistencies) and whether they are reused fake images. Frame consistency checks are performed on the property's VR video to prevent video editing fraud.

[0035] The property listing on-chain module includes a unique property identifier (PropertyID, automatically generated by the system), the property owner's DID, a hash of the property's core attributes (the hash value is calculated from the property's basic information, multimedia information, and property information), the property's status (available for rent / rented / under maintenance), and the authenticity verification result (verification agency ID, verification time, and verification conclusion). The original property images, VR videos, and scanned copies of the property ownership certificate are stored in IPFS, and the system records the corresponding IPFS hash value to facilitate quick retrieval of the original data through the hash value.

[0036] Preferably, the user profile construction submodule includes basic information features, preference features, behavioral features, and user vector fusion;

[0037] Basic information features: Extract user age, occupation, income level, and family situation (single / with children / with elderly), and quantify them into feature vectors (e.g., income level is quantified as "below 3000 yuan = 1, 3000-8000 yuan = 2, above 8000 yuan = 3").

[0038] Preference features: A preference feature vector is constructed by combining user-initiated settings (price range, area requirements, maximum commute time, preferred apartment type (1-bedroom / 2-bedroom / 3-bedroom)), decoration style (minimalist / Chinese / European), and floor preference (low / middle / high floor)) with passive data collected by the system (duration of time spent browsing properties, click frequency, and favorites). For example, if a user browses properties with a "commute time ≤ 30 minutes" multiple times, the system will include "commute time preference = 30 minutes" as a feature.

[0039] Behavioral characteristics: Extract user browsing history (property IDs and browsing time viewed in the past 3 months), collection history (feature tags of collected properties), viewing history (property features and viewing results of scheduled viewings), and historical transaction history (property types, rent ranges, and rental periods of past rentals). Use the Word2Vec algorithm to convert the behavioral sequence into a 128-dimensional behavioral embedding vector historyembedding.

[0040] User vector fusion: The basic information feature vector, preference feature vector, and behavior embedding vector are normalized (e.g., the budget is normalized to "budget / 10000"), and then the final user vector user_vec is generated through weighted fusion (with weights of 0.2, 0.3, and 0.5 respectively).

[0041] The property feature modeling submodule includes structured features, unstructured features, and property vector fusion;

[0042] Structured features: Extract property price, area, unit type, floor, commuting time (public transportation time from the property to the user's workplace, calculated by connecting to the map API), and supporting facilities (subway = 1 / no subway = 0, shopping mall = 1 / no shopping mall = 0, school = 1 / no school = 0, hospital = 1 / no hospital = 0), and quantify them into a structured feature vector house_struct_vec;

[0043] Unstructured features: A CNN convolutional neural network is used to extract features from the property images to generate a 256-dimensional image embedding vector; a Transformer model is used to semantically encode the property description text (such as "north-south facing, move-in ready, near subway") to generate a 128-dimensional text embedding vector; the image embedding vector and the text embedding vector are concatenated to obtain the unstructured feature vector house_unstruct_vec;

[0044] Property Vector Fusion: After normalizing the structured feature vector and the unstructured feature vector, they are fused with weights of 0.6 and 0.4 to generate the final property vector house_vec;

[0045] The recommendation strategy execution submodule includes a filtering layer, a similarity matching layer, a collaborative filtering layer, a ranking layer, and diversity adjustment.

[0046] Filtering layer: Based on the hard conditions actively set by the user (price range, area requirement, apartment type), the entire housing list is filtered to select a candidate housing set; for example, if the user's budget is "3000-5000 yuan / month", area requirement is "60-80m2", and apartment type is "2 bedrooms", then housing that does not meet the conditions will be filtered out.

[0047] Similarity matching layer: Calculate the cosine similarity between the user vector user_vec and the candidate house vector house_vec, using the following formula:

[0048] CosineSimilarity=(user_vec·house_vec) / (||user_vec||×||house_vec||)

[0049] Candidate listings are sorted in descending order of similarity, and the top 100 are selected as the listings to be sorted.

[0050] Collaborative Filtering Layer: Based on the principle of "similar users have similar preferences", find historical users whose vector similarity with the current user is ≥0.8, extract the favorite and transaction listings of these users; calculate the feature similarity between the listings to be sorted and these listings, and generate a collaborative filtering score (weight 0.3);

[0051] The ranking layer combines "content matching score (similarity matching score, weight 0.5)," "collaborative filtering score (weight 0.3)," and "popularity / freshness score (the more recent the listing time and the higher the number of views, the higher the score, weight 0.2)" to generate a comprehensive score; the top 10 listings are selected as the final recommended listings (Top-N recommendation) based on their comprehensive scores in descending order.

[0052] Diversity Adjustment: If the proportion of recommended properties in the same area (e.g., "Chaoyang District") or at the same price range (e.g., "4000-5000 yuan") is ≥60%, diversity optimization is triggered, replacing some properties with those in other areas / prices that have the second highest overall scores, thus ensuring diversity in recommendations.

[0053] Preferably, the transaction matching submodule has a built-in IM tool that supports communication between tenants and landlords via text, voice, images, and video. Communication records are encrypted and stored, and the hash values ​​of key communication content (such as rent adjustments and furniture additions) are recorded on the blockchain. Tenants can select viewing times online, and landlords can confirm or adjust times through the system. The system generates a viewing appointment form, which includes the property ID, tenant DID, landlord DID, and viewing time, and the appointment form hash is recorded on the blockchain. Tenants submit their rental intentions for satisfactory properties, filling in their desired rental period and payment method. The system synchronizes the intentions with the landlord. After the landlord confirms the intentions, a draft rental agreement is generated.

[0054] The contract signing submodule is based on a standardized system template and automatically generates an electronic contract by combining the terms confirmed by the tenant and landlord, including rent, lease term, deposit amount, and maintenance responsibility. The contract includes the DIDs of both parties, the property ID, and the smart contract address for deposit escrow and rent payment. Using a digital certificate issued by an authoritative CA institution, the tenant and landlord complete the online signing. After signing, the contract calculates a hash value and synchronizes it to the smart contract engine of the blockchain basic service module. The hash value is stored on the blockchain as evidence, and the original contract is stored in IPFS. The system automatically synchronizes the contract hash value and filing information to the government regulatory node. The filing information includes the contract number, the DIDs of both parties, and the property ID, completing the online filing. The filing result is fed back to both parties in real time.

[0055] In the payment and settlement submodule, tenants bind their bank cards or digital wallets to the system. According to the payment cycle stipulated in the contract, the system triggers a smart contract to automatically deduct the rent. After payment, the payment transaction hash (including payment amount, payment time, payer's DID, and receiver's DID) is recorded on the blockchain, while the payment voucher is stored off-chain. After the tenant pays the deposit, the system transfers the deposit to the smart contract escrow account. The smart contract records the deposit status (escrowed / pending refund / partially deducted), and status change records are recorded on the blockchain. During the lease period, the system automatically records the collected and paid fees such as water, electricity, and property management fees (connected to the building's smart water meter, electricity meter, or property management system), and the fee details hash is recorded on the blockchain. Upon termination of the lease, the system automatically calculates the total fees and deducts the due fees from the deposit.

[0056] Preferably, in the image acquisition and preprocessing submodule, when a tenant terminates a lease, the landlord / inspection personnel take images of the house's equipment via the system's APP. These images must cover all major equipment, including beds, sofas, air conditioners, refrigerators, lamps, floors, walls, and bathroom facilities. Each piece of equipment should be photographed from 3-5 angles, including front, side, and close-up views of key components. The system provides photo-taking guidance (such as overlayguides), automatically detects image blur, and prompts for reshooting when the blur exceeds a threshold. The acquired images undergo size standardization (uniformly adjusted to 1024×768 pixels), lighting correction (using histogram equalization to eliminate lighting differences), and noise removal (using Gaussian filtering to remove image noise). The RGB images are then converted to BGR format for subsequent model input.

[0057] The device localization and recognition submodule uses a MaskR-CNN model pre-trained on the COCO dataset, optimized through transfer learning to adapt to the device recognition scenario in a house. The model performs object detection on the pre-processed image, outputting the device's bounding box and instance mask to locate the device's position in the image. Supported device categories include, but are not limited to, beds, sofas, air conditioners, refrigerators, washing machines, lamps, floors, walls, toilets, and sinks. Based on the device's features (such as shape, texture, and components), the model outputs device category labels and confidence scores. The confidence score threshold is set to 0.7; if it is below the threshold, it is marked as "awaiting manual confirmation." The system extracts the segmented image of each device separately for subsequent damage detection.

[0058] The damage detection submodule uses ResNet-50 as the base network to build binary classification (normal / damaged) and multi-classification (damage type: scratches, stains, cracks, holes, corrosion, missing parts) models. The models are trained on a labeled dataset (containing over 100,000 images of normal / damaged devices), using the cross-entropy loss function, the Adam optimizer, a learning rate of 1e-3, a batch size of 32, and a training duration of 30 epochs. For damage requiring pixel-level localization (such as cracks and holes), the U-Net model is used. Defect segmentation is performed, and a mask of the damaged area is output. The ratio of the damaged area (number of pixels) to the total area (number of pixels) of the device is calculated to quantify the degree of damage (<1% is minor damage, 1%-5% is moderate damage, and >5% is severe damage). For damage scenarios without labeled data, a heuristic algorithm is used to calculate the damage score: the segmented image of the device is converted into a grayscale image, and the standard deviation (reflecting local contrast, with higher contrast in damaged areas) and mean (reflecting brightness, with lower mean in stain / rust areas) of the grayscale values ​​are calculated. The scoring formula is as follows:

[0059] DamageScore=(std(gray)×0.5)+((128-mean(gray))×0.01)

[0060] Set a damage threshold (e.g., 10.0). A score greater than the threshold is considered a damage, and the higher the score, the more severe the damage.

[0061] The damage assessment and reporting submodule combines damage classification results, damage segmentation area proportions, and heuristic damage scores to generate an equipment damage assessment table, including equipment ID, category, damage type, damage severity, and confidence level (the confidence level is the model output confidence level). The system automatically generates a damage report for returned equipment, including the damage assessment table, the original image and segmented image of the damaged equipment (with marked damage areas), and damage score details. The report hash value is uploaded to the blockchain, and the original report is stored in IPFS. If the damage detection confidence level is <0.8, the system marks the equipment as "awaiting manual review" and provides a manual annotation interface (supporting selecting damage areas and modifying damage types). After the manual review results are updated, the report is regenerated and uploaded to the blockchain.

[0062] Preferably, in the data encryption submodule, off-chain data (property images, contract text, user identity information) is stored using the AES-256 encryption algorithm, and the key is generated jointly by the user's DID and the system master key; on-chain data (hash value, DID, smart contract status) is encrypted using the national cryptographic algorithm SM2 to ensure that the data cannot be leaked; all data transmission in the system uses the SSL / TLS 1.3 protocol, and API interface calls use token authentication + signature verification to prevent data from being stolen or tampered with;

[0063] The access control submodule uses role-based access control (RBAC): it defines roles such as administrators, landlords, tenants, regulatory agencies, and banks, and assigns different operating permissions to each role (e.g., administrators can manage nodes, tenants can only view their own rental records); permission change records are recorded on the blockchain to ensure traceability; users can only access data related to themselves (e.g., tenants can only view their own rental property information), regulatory agencies can only view data required for supervision (e.g., contract filing information, transaction records), and access to other data requires authorization from the data owner, with authorization records recorded on the blockchain;

[0064] The security detection and vulnerability remediation submodule employs an Intrusion Detection System (IDS) and an Intrusion Prevention System (IPS) to monitor system access logs and transaction logs in real time, detect abnormal operations (such as multiple failed login attempts and a large number of abnormal API calls), and trigger alarms when anomalies are detected. Weekly vulnerability scans are performed on the system using the Nessus vulnerability scanning tool to conduct security checks on blockchain nodes, servers, and databases. Quarterly penetration tests are conducted to simulate attack scenarios, identify, and remediate security vulnerabilities. A security emergency response mechanism is established; upon discovering a vulnerability or data breach, an emergency plan is immediately activated (such as suspending related functions, isolating affected nodes, and restoring data backups), and the incident handling process and results are recorded on the blockchain.

[0065] Preferably, in the user account management submodule, the administrator can create accounts with different roles (such as government regulatory node accounts and bank node accounts), and ordinary users create accounts through the registration process. The accounts are bound to DIDs. The system supports account password reset, account freezing / unfreezing, and permission adjustment. Operation records are stored in the system log, and the log hash is uploaded to the blockchain.

[0066] The parameter configuration submodule configures blockchain consensus algorithm parameters (e.g., electiontimeout = 300-500ms, heartbeat = 100ms, batch_size = 50 log entries), smart contract parameters (e.g., deposit ratio = 1 month's rent, overdue days threshold = 7 days), and recommendation algorithm parameters (e.g., similarity threshold = 0.6, diversity adjustment ratio = 60%); it also configures parameters for data acquisition devices (cameras, microphones, smart water meters and electricity meters) (e.g., image resolution, sampling rate), supporting device calibration and status monitoring.

[0067] The log management submodule collects system operation logs (user login, account operations, parameter modifications), transaction logs (rent payment, deposit escrow, contract signing), and node logs (consensus process, block generation, cross-chain interaction). Logs include the operator's DID, operation time, operation content, and operation result. Logs are stored in the time-series database InfluxDB and support queries by time, operation type, and user DID. The hash values ​​of critical logs (such as transaction logs and permission change logs) are uploaded to the blockchain to ensure log immutability. Visual analysis of logs is supported (such as transaction volume trends and node running status), and alarms are triggered when abnormal logs are detected.

[0068] A rental method for an online housing and real estate rental system based on blockchain technology includes the following steps:

[0069] Step 1: System Initialization: Configure consortium blockchain nodes, consensus / recommendation / damage identification parameters, deploy smart contracts, and create initial accounts;

[0070] Step 2: User authentication: The user submits information, completes authentication with UnionPay via CTID, generates DID, and hashes the authentication result onto the blockchain;

[0071] Step 3: Property Listing and Verification: Landlords upload property information and verify it through government API / bank / AI. After verification, key property information is uploaded to the blockchain.

[0072] Step 4: Intelligent Recommendation and Matching: Generate user and property vectors, filter, match similarity, perform collaborative filtering and diversity adjustment to output recommended properties, and communicate with both parties to reach an agreement;

[0073] Step 5: Contract Signing and Payment: Generate an electronic contract and digitally sign it; the contract hash is uploaded to the blockchain. The tenant pays the deposit and rent; the deposit is held in escrow by the smart contract; the rent deduction record is uploaded to the blockchain.

[0074] Step 6: Damage identification and deposit processing upon termination of lease: Collect and preprocess equipment images, locate the equipment, detect damage and generate a report; if there is no dispute, the deposit will be refunded; if there is a dispute, it will be handled through arbitration.

[0075] Step 7: Data storage and update: Key data hashes are stored on-chain, and raw data is stored off-chain to update user credit and property status;

[0076] Step 8: System monitoring: Collect full-process logs, perform real-time security checks, and regularly back up and patch vulnerabilities.

[0077] Compared with the prior art, the beneficial effects of the present invention are:

[0078] This invention utilizes blockchain technology to achieve immutable notarization of key data throughout the entire rental process. Combined with multi-node consensus and digital signatures, it addresses issues of false property information and unreliable transaction data, enhancing trust between tenants and landlords. The intelligent recommendation module integrates user basic information, preferences, and behavioral characteristics, along with structured and unstructured property features. Through multi-dimensional scoring fusion and diversity adjustment, it achieves precise property matching, reducing user screening costs and improving rental matching efficiency. The equipment damage identification module employs multi-model fusion to automate damage assessment, generating objective and traceable damage reports, reducing subjective biases from manual inspections and lowering the incidence of termination disputes. The data security module protects user privacy and data security through multi-layered encryption and fine-grained access control. The blockchain consensus algorithm supports fault tolerance of n≥3f+1, and combined with aggregated signatures and batch processing optimization, it balances security and high throughput, meeting the multi-node collaboration needs of rental scenarios. The system integrates third-party services such as government regulation, financial services, and legal arbitration, achieving data interaction through a cross-chain gateway, constructing a full-process, one-stop real estate rental ecosystem, and enhancing user experience and system credibility. Attached Figure Description

[0079] Figure 1 This is a system schematic diagram of the present invention;

[0080] Figure 2 This is a schematic diagram of the damaged equipment identification module for returned leased equipment of the present invention;

[0081] Figure 3 This is a schematic diagram of the blockchain basic service module of this invention;

[0082] Figure 4 This is a schematic diagram of the intelligent recommendation module of the present invention;

[0083] Figure 5 This is a flowchart of the method of the present invention. Detailed Implementation

[0084] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0085] Please see Figure 1-5 This invention provides an online housing and real estate rental system based on blockchain technology, including a blockchain basic service module, a user management module, a housing management module, an intelligent recommendation module, a rental transaction module, a return equipment damage identification module, a data security module, and a system management module;

[0086] The blockchain infrastructure service module is responsible for realizing the immutable notarization of key data throughout the entire leasing process, multi-node collaboration, and automated contract execution. It includes the consortium blockchain network sub-module, consensus algorithm sub-module, smart contract engine sub-module, and cross-chain gateway sub-module.

[0087] The blockchain infrastructure service module includes the following code:

[0088]

[0089]

[0090]

[0091] The user management module is responsible for user identity authentication, permission allocation and credit management, ensuring the credibility of user identity and the traceability of behavior, including the identity authentication sub-module and the credit management sub-module.

[0092] The housing management module realizes housing information collection, authenticity verification and on-chain evidence storage to ensure that housing information is credible and traceable. It includes a housing information collection sub-module, a housing authenticity verification sub-module and a housing on-chain module.

[0093] The intelligent recommendation module is based on user profiles and property features to achieve accurate property matching and improve rental matching efficiency. It includes a user profile construction submodule, a property feature modeling submodule, and a recommendation strategy execution submodule.

[0094] The intelligent recommendation module includes the following code:

[0095]

[0096]

[0097]

[0098] The leasing transaction module enables online operation and reliable evidence storage for the entire leasing process, including a transaction matching submodule, a contract signing submodule, and a payment settlement submodule.

[0099] The equipment damage identification module for lease termination enables automated identification and quantitative assessment of equipment damage during lease termination, reducing lease termination disputes. It includes an image acquisition and preprocessing submodule, an equipment positioning and identification submodule, a damage detection submodule, and a damage assessment and reporting submodule.

[0100] The damaged equipment identification module for returned equipment includes the following code:

[0101]

[0102]

[0103]

[0104]

[0105] The data security module ensures the security and privacy of system data during collection, transmission, storage, and use, including a data encryption submodule, an access control submodule, and a security detection and vulnerability remediation submodule.

[0106] The system management module is responsible for the daily maintenance, parameter configuration, and log management of the system to ensure stable system operation. It includes user account management submodule, parameter configuration submodule, and log management submodule.

[0107] The consortium blockchain network sub-module includes platform operation nodes, government regulatory nodes (such as real estate registration centers and market supervision departments), financial service nodes (banks and insurance companies), and third-party authoritative nodes (notary offices and law firms). It adopts a permissioned blockchain architecture, where only authorized nodes can participate in consensus and data access. Built on Hyperledger Fabric 2.5+, it supports dynamic node addition and permission adjustment, adapting to the collaborative needs of multiple participants in leasing scenarios.

[0108] The consensus algorithm submodule adopts an optimized "fault-tolerant high-throughput consensus algorithm" that supports the number of nodes n≥3f+1 (f is the number of Byzantine nodes), meeting the system's fault tolerance requirements;

[0109] Fault-tolerant high-throughput consensus algorithm process:

[0110] Leader election: Nodes trigger an election by randomizing the election timeout. Candidate nodes broadcast a digitally signed "VoteRequest" containing the candidate node ID, current term, and the node's latest log index and term. When a candidate node collects ≥2f+1 signed "VoteResponses", it generates an "Election-Certificate", becomes the new leader, and upgrades the term.

[0111] Log replication: After receiving a request from a client (landlord / tenant), the Leader generates a log entry containing term, index, and request content, and broadcasts "AppendProposal". Followers verify the Leader's signature and log prefix consistency (prevIndex matches the local log), and then return a signed "Prepare". After collecting ≥2f+1 "Prepare" entries, the Leader generates a "Prepare-Certificate" and broadcasts "CommitRequest". Nodes verify the validity of the "Prepare-Certificate" and return "Commit". When any node collects ≥2f+1 "Commit" entries, it generates a "Commit-Certificate", marks the log entry as "committed", and executes the request.

[0112] View Change: If the Leader fails to advance the log, initiate a conflict proposal, or double-sign, the node broadcasts a ViewChange containing its highest Prepared log information and the Prepare-Certificate; the new Leader candidate collects ≥2f+1 ViewChanges, confirms the highest Prepared log, publishes NewView, and starts the log replication process for the new term;

[0113] Optimized design: Employing BLS aggregate signature technology, multi-node signatures are merged into a single signature, reducing message complexity (from O(n...). 2 The throughput is reduced to O(n), while supporting batch processing of log entries (1-100 requests / entries), improving system throughput;

[0114] The smart contract engine submodule includes a lease agreement contract, a payment settlement contract, a deposit escrow contract, and a dispute arbitration contract.

[0115] Rental Agreement Contract: Automatically generates standardized electronic contract templates, supporting landlords and tenants to fill in and modify online. After the contract is signed, the hash value is calculated and uploaded to the blockchain, while the original contract is stored off-chain in IPFS (InterPlanetary File System).

[0116] Payment and settlement contract: Based on the rent amount and payment cycle agreed in the contract, the rent transfer is automatically triggered from the tenant's account to the landlord's account, and the payment transaction hash is recorded on the blockchain; it supports automatic reminders for overdue payments, and triggers the insurance compensation process if the overdue period exceeds the agreed time;

[0117] Escrow Agreement: The deposit paid by the tenant is locked in a smart contract account and will be automatically refunded when there is no dispute upon termination of the lease and inspection. If there is a dispute over damage, the corresponding amount will be automatically deducted from the landlord's account based on the damage identification results and arbitration conclusion.

[0118] Dispute Arbitration Contract: Receives the application and evidence hash from the party initiating the dispute (landlord / tenant), automatically notifies the arbitration node (third-party law firm / notary office) to make a ruling, and stores the ruling result and evidence hash on the blockchain for evidence storage;

[0119] The cross-chain gateway submodule connects to external government blockchains (such as local real estate registration blockchains) and financial blockchains (bank payment blockchains) to achieve cross-chain data interaction. For example, it can obtain property ownership verification results from the real estate registration blockchain and synchronize rent payment status from the bank payment blockchain. It adopts the WeCross cross-chain framework to support data format conversion and secure transmission between blockchains, ensuring the consistency and credibility of cross-chain data.

[0120] The identity authentication submodule includes individual users and enterprise users;

[0121] For individual users: It connects to the Ministry of Public Security's "Internet and Trusted Identity Authentication" (CTID) platform, and completes three-factor authentication through facial recognition, comparison of the front and back of the ID card, and liveness detection; it also supplements the four-factor authentication of UnionPay bank cards, namely, consistency verification of name, ID card number, bank card number, and mobile phone number. After successful authentication, a unique blockchain digital identity DID is generated.

[0122] Enterprise users (such as institutional landlords): Submit business license, legal representative's ID card, and real estate ownership certificate. After manual review and verification by the government's enterprise credit information disclosure system, an enterprise DID will be generated. Enterprise employees need to configure additional role permissions (RBAC model) to operate, and operation records are linked to the enterprise DID and uploaded to the blockchain.

[0123] The credit management submodule includes credit scoring and handling of credit defaults;

[0124] Credit scoring: Based on users' on-chain behavior (such as timely rent payment, contract performance, and records of lease termination disputes) and external credit data (connected to "Credit China" and the list of dishonest persons subject to enforcement by the court), a credit scoring model (maximum score of 1000 points) is constructed, and the scoring results are updated in real time and associated with the user's DID;

[0125] Handling of Dishonesty: If a user engages in false listings, malicious rent arrears, damages property equipment, or other dishonest behavior, their DID will be marked as "dishonest" and synchronized to all nodes; dishonest users will be unable to initiate new rental requests, and already initiated requests will be automatically blocked.

[0126] In the property information collection submodule, landlords upload basic property information (address, apartment type, area, rent, payment method, and amenities), multimedia information (property photos, VR panoramic videos, and surrounding environment videos), and property information (scanned copy of the property ownership certificate and the owner's identity document) through the system. The system provides upload guidelines, requiring that property photos include close-ups (taken from a distance of 0.5-1.5m) of the four corners of the house and major appliances (bed, sofa, air conditioner, refrigerator), and that VR videos fully cover every room of the house. The image resolution should be no less than 1600px, and backlighting and blurriness should be avoided.

[0127] The property authenticity verification submodule connects to the government API of the real estate registration center. By inputting the property's real estate unit number and the owner's ID number / corporate unified social credit code, it retrieves data on the property's location, area, mortgage status, seizure information, and co-ownership status to verify the legality and availability of the property rights. If the city where the property is located does not have access to the government API, it queries the property's ownership information through a partner bank (a large state-owned bank). The landlord must sign an authorization agreement allowing the bank to connect to the real estate registration center through its professional system, and the verification results are synchronized to the system. A ViT-based image recognition model is used to detect whether property images show signs of Photoshop manipulation (such as edge tampering or color inconsistencies) and whether they are reused fake images. Frame consistency checks are performed on the property's VR video to prevent video editing fraud.

[0128] The property listing on-chain module includes a unique property identifier (PropertyID, automatically generated by the system), the property owner's DID, a hash of the property's core attributes (the hash value is calculated from the property's basic information, multimedia information, and property information), the property's status (available for rent / rented / under maintenance), and the authenticity verification result (verification agency ID, verification time, and verification conclusion). The original property images, VR videos, and scanned copies of the property ownership certificate are stored in IPFS, and the system records the corresponding IPFS hash value to facilitate quick retrieval of the original data through the hash value.

[0129] The user profile construction submodule includes basic information features, preference features, behavioral features, and user vector fusion;

[0130] Basic information features: Extract user age, occupation, income level, and family situation (single / with children / with elderly), and quantify them into feature vectors (e.g., income level is quantified as "below 3000 yuan = 1, 3000-8000 yuan = 2, above 8000 yuan = 3").

[0131] Preference features: A preference feature vector is constructed by combining user-initiated settings (price range, area requirements, maximum commute time, preferred apartment type (1-bedroom / 2-bedroom / 3-bedroom)), decoration style (minimalist / Chinese / European), and floor preference (low / middle / high floor)) with passive data collected by the system (duration of time spent browsing properties, click frequency, and favorites). For example, if a user browses properties with a "commute time ≤ 30 minutes" multiple times, the system will include "commute time preference = 30 minutes" as a feature.

[0132] Behavioral characteristics: Extract user browsing history (property IDs and browsing time viewed in the past 3 months), collection history (feature tags of collected properties), viewing history (property features and viewing results of scheduled viewings), and historical transaction history (property types, rent ranges, and rental periods of past rentals). Use the Word2Vec algorithm to convert the behavioral sequence into a 128-dimensional behavioral embedding vector historyembedding.

[0133] User vector fusion: The basic information feature vector, preference feature vector, and behavior embedding vector are normalized (e.g., the budget is normalized to "budget / 10000"), and then the final user vector user_vec is generated through weighted fusion (with weights of 0.2, 0.3, and 0.5 respectively).

[0134] The property feature modeling submodule includes structured features, unstructured features, and property vector fusion;

[0135] Structured features: Extract property price, area, unit type, floor, commuting time (public transportation time from the property to the user's workplace, calculated by connecting to the map API), and supporting facilities (subway = 1 / no subway = 0, shopping mall = 1 / no shopping mall = 0, school = 1 / no school = 0, hospital = 1 / no hospital = 0), and quantify them into a structured feature vector house_struct_vec;

[0136] Unstructured features: A CNN convolutional neural network is used to extract features from the property images to generate a 256-dimensional image embedding vector; a Transformer model is used to semantically encode the property description text (such as "north-south facing, move-in ready, near subway") to generate a 128-dimensional text embedding vector; the image embedding vector and the text embedding vector are concatenated to obtain the unstructured feature vector house_unstruct_vec;

[0137] Property Vector Fusion: After normalizing the structured feature vector and the unstructured feature vector, they are fused with weights of 0.6 and 0.4 to generate the final property vector house_vec;

[0138] The recommendation strategy execution submodule includes a filtering layer, a similarity matching layer, a collaborative filtering layer, a ranking layer, and diversity adjustment.

[0139] Filtering layer: Based on user-defined hard criteria (price range, area requirements, apartment type), all listings are filtered to select a candidate set; for example, a user's budget is "3000-5000 yuan / month" and their area requirement is "60-80 square meters". 2 If the apartment type is "2 bedrooms", then properties that do not meet this condition will be filtered out.

[0140] Similarity matching layer: Calculate the cosine similarity between the user vector user_vec and the candidate house vector house_vec, using the following formula:

[0141] CosineSimilarity=(user_vec·house_vec) / (||user_vec||×

[0142] ||house_vec||)

[0143] Candidate listings are sorted in descending order of similarity, and the top 100 are selected as the listings to be sorted.

[0144] Collaborative Filtering Layer: Based on the principle of "similar users have similar preferences", find historical users whose vector similarity with the current user is ≥0.8, extract the favorite and transaction listings of these users; calculate the feature similarity between the listings to be sorted and these listings, and generate a collaborative filtering score (weight 0.3);

[0145] The ranking layer combines "content matching score (similarity matching score, weight 0.5)," "collaborative filtering score (weight 0.3)," and "popularity / freshness score (the more recent the listing time and the higher the number of views, the higher the score, weight 0.2)" to generate a comprehensive score; the top 10 listings are selected as the final recommended listings (Top-N recommendation) based on their comprehensive scores in descending order.

[0146] Diversity Adjustment: If the proportion of recommended properties in the same area (e.g., "Chaoyang District") or at the same price range (e.g., "4000-5000 yuan") is ≥60%, diversity optimization is triggered, replacing some properties with those in other areas / prices that have the second highest overall scores, thus ensuring diversity in recommendations.

[0147] The transaction matching submodule has a built-in IM tool that allows tenants and landlords to communicate via text, voice, images, and video. Communication records are encrypted and stored, and the hash values ​​of key communication content (such as rent adjustments and furniture additions) are recorded on the blockchain. Tenants can select viewing times online, and landlords can confirm or adjust times through the system. The system generates a viewing appointment form, which includes the property ID, tenant DID, landlord DID, and viewing time. The appointment form's hash is recorded on the blockchain. Tenants submit their rental intentions for satisfactory properties, filling in their desired rental period and payment method. The system synchronizes the intentions with the landlord. After the landlord confirms the intention, a draft rental agreement is generated.

[0148] The contract signing submodule is based on a standardized system template and automatically generates an electronic contract by combining the terms confirmed by the tenant and landlord, including rent, lease term, deposit amount, and maintenance responsibility. The contract includes the DIDs of both parties, the property ID, and the smart contract address for deposit escrow and rent payment. Using a digital certificate issued by an authoritative CA institution, the tenant and landlord complete the online signing. After signing, the contract calculates a hash value and synchronizes it to the smart contract engine of the blockchain basic service module. The hash value is stored on the blockchain as evidence, and the original contract is stored in IPFS. The system automatically synchronizes the contract hash value and filing information to the government regulatory node. The filing information includes the contract number, the DIDs of both parties, and the property ID, completing the online filing. The filing result is fed back to both parties in real time.

[0149] In the payment and settlement submodule, tenants bind their bank cards or digital wallets to the system. According to the payment cycle stipulated in the contract, the system triggers a smart contract to automatically deduct the rent. After payment, the payment transaction hash (including payment amount, payment time, payer's DID, and receiver's DID) is recorded on the blockchain, while the payment voucher is stored off-chain. After the tenant pays the deposit, the system transfers the deposit to the smart contract escrow account. The smart contract records the deposit status (escrowed / pending refund / partially deducted), and status change records are recorded on the blockchain. During the lease period, the system automatically records the collected and paid fees such as water, electricity, and property management fees (connected to the building's smart water meter, electricity meter, or property management system), and the fee details hash is recorded on the blockchain. Upon termination of the lease, the system automatically calculates the total fees and deducts the due fees from the deposit.

[0150] In the image acquisition and preprocessing submodule, when a tenant moves out, the landlord / inspection personnel use the system's app to take images of the house's equipment. All major equipment must be covered, including beds, sofas, air conditioners, refrigerators, lights, floors, walls, and bathroom fixtures. Each piece of equipment should be photographed from 3-5 angles, including front, side, and close-up views of key components. The system provides photo-taking guidance (such as overlayguides), automatically detects image blur, and prompts for reshoots when blur exceeds a threshold. The acquired images undergo size standardization (uniformly adjusted to 1024×768 pixels), lighting correction (using histogram equalization to eliminate lighting differences), and noise removal (using Gaussian filtering to remove image noise). The RGB images are then converted to BGR format for subsequent model input.

[0151] The device localization and recognition submodule uses a MaskR-CNN model pre-trained on the COCO dataset, optimized through transfer learning to adapt to the device recognition scenario in a house. The model performs object detection on the pre-processed image, outputting the device's bounding box and instance mask to locate the device's position in the image. Supported device categories include, but are not limited to, beds, sofas, air conditioners, refrigerators, washing machines, lamps, floors, walls, toilets, and sinks. Based on the device's features (such as shape, texture, and components), the model outputs device category labels and confidence scores. The confidence score threshold is set to 0.7; if it is below the threshold, it is marked as "awaiting manual confirmation." The system extracts the segmented image of each device separately for subsequent damage detection.

[0152] The damage detection submodule uses ResNet-50 as the base network to build binary classification (normal / damaged) and multi-classification (damage type: scratches, stains, cracks, holes, corrosion, missing parts) models. The models are trained on a labeled dataset (containing over 100,000 images of normal / damaged devices), using the cross-entropy loss function, the Adam optimizer, a learning rate of 1e-3, a batch size of 32, and a training duration of 30 epochs. For damage requiring pixel-level localization (such as cracks and holes), the U-Net model is used. Defect segmentation is performed, and a mask of the damaged area is output. The ratio of the damaged area (number of pixels) to the total area (number of pixels) of the device is calculated to quantify the degree of damage (<1% is minor damage, 1%-5% is moderate damage, and >5% is severe damage). For damage scenarios without labeled data, a heuristic algorithm is used to calculate the damage score: the segmented image of the device is converted into a grayscale image, and the standard deviation (reflecting local contrast, with higher contrast in damaged areas) and mean (reflecting brightness, with lower mean in stain / rust areas) of the grayscale values ​​are calculated. The scoring formula is as follows:

[0153] DamageScore=(std(gray)×0.5)+((128-mean(gray))×0.01)

[0154] Set a damage threshold (e.g., 10.0). A score greater than the threshold is considered a damage, and the higher the score, the more severe the damage.

[0155] The damage assessment and reporting submodule combines damage classification results, damage segmentation area proportions, and heuristic damage scores to generate an equipment damage assessment table, including equipment ID, category, damage type, damage severity, and confidence level (the confidence level is the model output confidence level). The system automatically generates a damage report for returned equipment, including the damage assessment table, the original image and segmented image of the damaged equipment (with marked damage areas), and damage score details. The report hash value is uploaded to the blockchain, and the original report is stored in IPFS. If the damage detection confidence level is <0.8, the system marks the equipment as "awaiting manual review" and provides a manual annotation interface (supporting selecting damage areas and modifying damage types). After the manual review results are updated, the report is regenerated and uploaded to the blockchain.

[0156] In the data encryption submodule, off-chain data (property images, contract text, user identity information) is stored using the AES-256 encryption algorithm, with the key generated jointly by the user's DID and the system master key; on-chain data (hash value, DID, smart contract status) is encrypted using the national cryptographic algorithm SM2 to ensure data integrity; all system data transmissions use the SSL / TLS 1.3 protocol, and API calls use token authentication + signature verification to prevent data theft or tampering.

[0157] The access control submodule uses role-based access control (RBAC): it defines roles such as administrators, landlords, tenants, regulatory agencies, and banks, and assigns different operating permissions to each role (e.g., administrators can manage nodes, tenants can only view their own rental records); permission change records are recorded on the blockchain to ensure traceability; users can only access data related to themselves (e.g., tenants can only view their own rental property information), regulatory agencies can only view data required for supervision (e.g., contract filing information, transaction records), and access to other data requires authorization from the data owner, with authorization records recorded on the blockchain;

[0158] The security detection and vulnerability remediation submodule employs an Intrusion Detection System (IDS) and an Intrusion Prevention System (IPS) to monitor system access logs and transaction logs in real time, detect abnormal operations (such as multiple failed login attempts and a large number of abnormal API calls), and trigger alarms when anomalies are detected. Weekly vulnerability scans are performed on the system using the Nessus vulnerability scanning tool to conduct security checks on blockchain nodes, servers, and databases. Quarterly penetration tests are conducted to simulate attack scenarios, identify, and remediate security vulnerabilities. A security emergency response mechanism is established; upon discovering a vulnerability or data breach, an emergency plan is immediately activated (such as suspending related functions, isolating affected nodes, and restoring data backups), and the incident handling process and results are recorded on the blockchain.

[0159] In the user account management submodule, administrators can create accounts for different roles (such as government regulatory node accounts and bank node accounts). Ordinary users create accounts through the registration process, and the accounts are bound to DIDs. It supports account password reset, account freezing / unfreezing, and permission adjustment. Operation records are stored in the system log, and the log hash is uploaded to the blockchain.

[0160] The parameter configuration submodule configures blockchain consensus algorithm parameters (e.g., electiontimeout = 300-500ms, heartbeat = 100ms, batch_size = 50 log entries), smart contract parameters (e.g., deposit ratio = 1 month's rent, overdue days threshold = 7 days), and recommendation algorithm parameters (e.g., similarity threshold = 0.6, diversity adjustment ratio = 60%); it also configures parameters for data acquisition devices (cameras, microphones, smart water meters and electricity meters) (e.g., image resolution, sampling rate), supporting device calibration and status monitoring.

[0161] The log management submodule collects system operation logs (user login, account operations, parameter modifications), transaction logs (rent payment, deposit escrow, contract signing), and node logs (consensus process, block generation, cross-chain interaction). Logs include the operator's DID, operation time, operation content, and operation result. Logs are stored in the time-series database InfluxDB and support queries by time, operation type, and user DID. The hash values ​​of critical logs (such as transaction logs and permission change logs) are uploaded to the blockchain to ensure log immutability. Visual analysis of logs is supported (such as transaction volume trends and node running status), and alarms are triggered when abnormal logs are detected.

[0162] A rental method for an online housing and real estate rental system based on blockchain technology includes the following steps:

[0163] Step 1: System Initialization and Parameter Configuration

[0164] The administrator completes system initialization through the parameter configuration submodule of the system management module:

[0165] Blockchain basic service module: Configure the IP addresses and ports of consortium blockchain nodes (platform operation nodes, government regulatory nodes, bank nodes), set consensus algorithm parameters (electiontimeout=300-500ms, heartbeat=100ms, f=1 (tolerates 1 Byzantine node), n=4 (4 nodes)), and deploy smart contracts (lease agreement contract, payment settlement contract, deposit escrow contract, dispute arbitration contract).

[0166] Intelligent recommendation module: Configure user vector fusion weights (basic information 0.2, preferences 0.3, behavior 0.5), recommendation ranking weights (content matching 0.5, collaborative filtering 0.3, popularity / freshness 0.2), and diversity adjustment threshold (60%).

[0167] Damage identification module for returned equipment: Configure damage detection model parameters (confidence threshold 0.7, damage score threshold 10.0, and area percentage threshold for minor / moderate / severe damage).

[0168] The administrator creates initial user accounts (such as government regulatory node accounts and bank node accounts) and assigns roles and permissions; the user profile building sub-module of the intelligent recommendation module initializes the user feature database synchronously, and the housing listing on-chain sub-module of the housing management module initializes the IPFS storage node.

[0169] Step 2: User Registration and Identity Verification

[0170] Users (landlords / tenants) submit registration information through the system registration interface (individual users: name, ID number, mobile phone number; corporate users: company name, business license number, legal representative information).

[0171] The system calls the identity authentication submodule of the user management module to complete the authentication:

[0172] For individual users: Connect to the CTID platform to complete facial recognition, ID card comparison, and liveness detection; connect to UnionPay to complete the four-factor authentication of bank cards; generate user DID after successful authentication.

[0173] For enterprise users: The system reviews the submitted business license and legal representative information, and connects with the government's enterprise credit information disclosure system to verify the enterprise's legality; after the review is approved, an enterprise DID is generated, and employee roles and permissions are configured.

[0174] The system uploads the user's DID and the hash value of the authentication result (authentication authority ID and authentication time) to the blockchain, while the user's original identity information is encrypted and stored in an off-chain database.

[0175] Step 3: Listing and Authenticity Verification

[0176] Landlords upload property information (basic information, multimedia information, and property rights information) through the system's property management module, and the system generates a unique property identifier (PropertyID).

[0177] The system invokes the property listing authenticity verification submodule to complete the verification:

[0178] If the city where the property is located has opened the government affairs API, the system calls the real estate registration center API, enters the ID number / enterprise code associated with the PropertyID and the property owner's DID, and obtains the property verification result.

[0179] If the government API is not activated, the system will trigger the bank verification process. After the landlord signs the authorization agreement, the cooperating bank will query the property information and provide feedback on the verification results.

[0180] The system calls the AI-assisted verification submodule to detect whether the property photos / videos are fake and generates a verification conclusion.

[0181] After verification, the system calls the property listing on-chain module to upload the property ID, property owner DID, property core attribute hash, and verification result to the blockchain, while the original property information is stored in IPFS. If verification fails, the system will prompt the reason for the failure (such as property seizure or image forgery), and the landlord needs to modify and resubmit.

[0182] Step 4: Intelligent Recommendation and Transaction Matching

[0183] After a tenant logs into the system, the system calls the user profile building submodule of the intelligent recommendation module:

[0184] Extract basic information features of tenants and actively set preference features, and quantify them into feature vectors.

[0185] Collect tenants' browsing, favorites, and viewing records to generate behavioral embedding vectors.

[0186] By combining the above vectors, a user vector (user_vec) for the tenant is generated.

[0187] The system calls the housing feature modeling submodule to extract structured and unstructured features from verified housing listings and generate housing vectors (house_vec).

[0188] The recommendation strategy execution submodule generates recommendation results according to the following process:

[0189] Filtering layer: Based on hard criteria of tenant preferences, filter the set of candidate properties.

[0190] Similarity matching layer: Calculate the cosine similarity between user_vec and house_vec, and select the top 100 listings after sorting.

[0191] Collaborative filtering layer: Finds preferred listings for similar users and generates collaborative filtering scores.

[0192] Sorting layer: Integrates multi-dimensional scores, selects the top 10 listings in descending order of comprehensive score, triggers diversity adjustment (if necessary), and pushes them to tenants.

[0193] After browsing recommended listings, tenants can communicate with landlords and schedule viewings through the transaction matching sub-module. Once both parties reach an agreement, the tenant submits their rental intention, and the landlord confirms the intention.

[0194] Step 5: Contract signing and payment settlement

[0195] The system calls the contract signing submodule of the rental transaction module to generate an electronic contract based on the terms confirmed by both parties, including the DIDs of both parties, the property ID, and the smart contract address (escrow of deposit and rent payment).

[0196] Tenants and landlords complete digital signatures through the system. The signed contract calculates a hash value and uploads it to the smart contract engine of the blockchain basic service module. The hash value is recorded on the blockchain, and the original contract is stored in IPFS. The system automatically synchronizes the contract filing information to the government regulatory node.

[0197] Tenants can bind their payment method through the payment settlement sub-module and pay the deposit and first rent installment as stipulated in the contract:

[0198] The deposit is transferred to the smart contract escrow account, and the smart contract records the deposit status as "escrowed" and the status is recorded on the blockchain.

[0199] The first installment of rent is deducted from the landlord's account via a smart contract, the payment transaction hash is recorded on the blockchain, and the payment voucher is stored off-chain.

[0200] Subsequent rent payments: The system will automatically deduct the rent according to the payment cycle agreed in the contract, and the payment record will be uploaded to the blockchain in real time; if the tenant is late in making payments, the system will send an overdue reminder. If the overdue period exceeds the agreed number of days (e.g., 7 days), the insurance compensation process will be triggered. After the insurance company compensates the rent, the compensation record will be uploaded to the blockchain, and the tenant's credit score will be reduced.

[0201] Step 6: Damage identification of returned equipment and deposit processing

[0202] When the lease expires or the contract is terminated early, the landlord / inspection personnel use the image acquisition sub-module of the equipment damage identification module to take pictures of the equipment in the house, and the system completes the preprocessing.

[0203] The device localization and identification submodule calls the MaskR-CNN model to locate and classify devices and extract device segmentation images.

[0204] The damage detection submodule calls the ResNet-50 classification model and the U-Net segmentation model to detect the type and degree of equipment damage and calculate the damage score; equipment with a confidence score <0.8 is marked as "awaiting manual review", and the results are updated after manual review.

[0205] The system generates a damage report for returned equipment, hashes the report on the blockchain, and stores the original report in IPFS. Both parties can view the report; if there is no dispute, the deposit processing procedure begins; if there is a dispute, arbitration is initiated.

[0206] In the absence of disputes: the smart contract will automatically refund the remaining deposit to the tenant's account based on the cost calculation results (after deducting the water and electricity fees / property fees payable), and the deposit status will be updated to "refunded" and recorded on the blockchain.

[0207] In case of a dispute: The initiating party (landlord / tenant) submits a dispute application and evidence hash through the dispute arbitration contract, and the system notifies the arbitration node (third-party law firm); After reviewing the evidence, the arbitration node issues an award, and the award is recorded on the blockchain; Based on the award, the smart contract deducts the corresponding amount (such as damage compensation) to the landlord's account, the remaining deposit is refunded to the tenant, and the deposit status is updated to "processed".

[0208] Step 7: Data Storage and Update

[0209] The system categorizes and stores all data throughout the entire process:

[0210] On-chain data: User DID, property core attribute hash, contract hash, payment transaction hash, deposit status, damage report hash, arbitration result hash, are stored in the consortium blockchain network of the blockchain basic service module.

[0211] Off-chain data: Original user identity information, original photos / videos of the property, contract text, payment vouchers, and original damage report files are stored in IPFS or an encrypted database (MySQL / AES-256 encryption).

[0212] The system regularly updates user credit scores (based on performance and dispute records), property status (available / rented / under maintenance), user profiles and property feature databases in the intelligent recommendation module, and uploads the updated records to the blockchain.

[0213] Step 8: System Monitoring and Maintenance

[0214] The system management module's log management submodule collects system operation logs, transaction logs, and node logs in real time, and hashes the logs on the blockchain; the data security module's security detection submodule monitors the system's security status in real time and triggers alarms when it detects anomalies (such as intrusion attempts or vulnerabilities).

[0215] Administrators perform regular maintenance through the system management module:

[0216] Parameter adjustment: Adjust the consensus algorithm parameters and recommendation algorithm parameters according to the system operation (such as consensus latency and recommendation accuracy).

[0217] Data backup: Off-chain database and IPFS storage data are backed up weekly, and backup files are stored in encrypted form.

[0218] Vulnerability remediation: Based on security testing results, remediate system vulnerabilities; conduct system upgrades quarterly to optimize module functions (such as improving the accuracy of the damage identification model and optimizing the recommendation algorithm).

[0219] The consensus algorithm submodule of the blockchain infrastructure service module monitors the node status in real time. If a node fails, it triggers a view replacement process to ensure the continuous operation of the system.

[0220] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An online housing and real estate rental system based on blockchain technology, characterized in that: It includes a blockchain basic service module, a user management module, a housing management module, an intelligent recommendation module, a rental transaction module, a lease termination equipment damage identification module, a data security module, and a system management module; The blockchain infrastructure service module is responsible for realizing the immutable notarization of key data throughout the entire leasing process, multi-node collaboration, and automated contract execution. It includes a consortium blockchain network submodule, a consensus algorithm submodule, a smart contract engine submodule, and a cross-chain gateway submodule. The user management module is responsible for user identity authentication, permission allocation and credit management, ensuring the credibility of user identity and the traceability of behavior, including an identity authentication sub-module and a credit management sub-module; The housing management module realizes housing information collection, authenticity verification and on-chain storage to ensure that housing information is credible and traceable. It includes a housing information collection sub-module, a housing authenticity verification sub-module and a housing on-chain storage module. The intelligent recommendation module, based on user profiles and property features, achieves accurate property matching and improves rental matching efficiency. It includes a user profile construction submodule, a property feature modeling submodule, and a recommendation strategy execution submodule. The leasing transaction module enables online operation and reliable evidence storage for the entire leasing process, including a transaction matching submodule, a contract signing submodule, and a payment settlement submodule. The equipment damage identification module for lease termination enables automated identification and quantitative assessment of equipment damage during lease termination, reducing lease termination disputes. It includes an image acquisition and preprocessing submodule, an equipment positioning and identification submodule, a damage detection submodule, and a damage assessment and reporting submodule. The data security module ensures the security and privacy of system data during collection, transmission, storage, and use, and includes a data encryption submodule, an access control submodule, and a security detection and vulnerability repair submodule. The system management module is responsible for the daily maintenance, parameter configuration, and log management of the system to ensure stable system operation. It includes a user account management submodule, a parameter configuration submodule, and a log management submodule.

2. The online housing and real estate rental system based on blockchain technology according to claim 1, characterized in that: The consortium blockchain network sub-module includes platform operation nodes, government regulatory nodes, financial service nodes, and third-party authoritative nodes. It adopts a permissioned blockchain architecture, where only authorized nodes can participate in consensus and data access. It is built on Hyperledger Fabric 2.5+, supports dynamic node addition and permission adjustment, and is adapted to the collaborative needs of multiple participants in leasing scenarios. The consensus algorithm submodule adopts an optimized "fault-tolerant high-throughput consensus algorithm" that supports the number of nodes n≥3f+1, meeting the system's fault tolerance requirements; Fault-tolerant high-throughput consensus algorithm process: Leader election: Nodes trigger an election by randomizing the election timeout. Candidate nodes broadcast a digitally signed "VoteRequest" containing the candidate node ID, current term, and the node's latest log index and term. When a candidate node collects ≥2f+1 signed "VoteResponses", it generates an "Election-Certificate", becomes the new leader, and upgrades the term. Log replication: After receiving a client request, the Leader generates a log entry containing term, index, and request content, and broadcasts "AppendProposal". Followers verify the consistency of the Leader's signature and log prefix, and return a signed "Prepare". After collecting ≥2f+1 "Prepare", the Leader generates a "Prepare-Certificate" and broadcasts "CommitRequest". After verifying the validity of the "Prepare-Certificate", nodes return "Commit". When any node collects ≥2f+1 "Commit", it generates a "Commit-Certificate", marks the log entry as "committed", and executes it. View Change: If the Leader fails to advance the log, initiate a conflict proposal, or double-sign, the node broadcasts a ViewChange containing its highest Prepared log information and the Prepare-Certificate; the new Leader candidate collects ≥2f+1 ViewChanges, confirms the highest Prepared log, publishes NewView, and starts the log replication process for the new term; Optimized design: The BLS aggregate signature technology is adopted to merge the signatures of multiple nodes into a single signature, reducing message complexity, while supporting batch processing of log entries to improve system throughput; The smart contract engine submodule includes a lease agreement contract, a payment settlement contract, a deposit escrow contract, and a dispute arbitration contract. Rental Agreement Contract: Automatically generates standardized electronic contract templates, supporting landlords and tenants to fill in and modify online. After the contract is signed, the hash value is calculated and uploaded to the blockchain, while the original contract is stored off-chain in IPFS; Payment settlement contract: Based on the rent amount and payment cycle agreed in the contract, the rent transfer is automatically triggered from the tenant's account to the landlord's account, and the payment transaction hash is recorded on the blockchain; Supports automatic reminders for overdue payments; if the overdue period exceeds the agreed time, the insurance compensation process will be triggered. Escrow Agreement: The deposit paid by the tenant is locked in a smart contract account and will be automatically refunded when there is no dispute upon termination of the lease and inspection. If there is a dispute over damage, the corresponding amount will be automatically deducted from the landlord's account based on the damage identification results and arbitration conclusion. Dispute Arbitration Contract: Receives the application and evidence hash from the party initiating the dispute, automatically notifies the arbitration node to make a ruling, and stores the ruling result and evidence hash on the blockchain for evidence storage; The cross-chain gateway submodule connects to external government and financial chains to enable cross-chain data interaction. For example, it can obtain property ownership verification results from the real estate registration chain and synchronize rent payment status from the bank payment chain. It adopts the WeCross cross-chain framework to support data format conversion and secure transmission between chains, ensuring the consistency and credibility of cross-chain data.

3. The online housing and real estate rental system based on blockchain technology according to claim 1, characterized in that: The identity authentication submodule includes individual users and enterprise users; For individual users: It connects to the Ministry of Public Security's "Internet and Trusted Identity Authentication" platform, and completes three-factor authentication through facial recognition, comparison of the front and back of the ID card, and liveness detection; it also supplements the four-factor authentication of UnionPay bank cards, namely, consistency verification of name, ID number, bank card number, and mobile phone number. After successful authentication, a unique blockchain digital identity DID is generated. Enterprise users: Submit business license, legal representative's ID card, and real estate ownership certificate. After manual review and verification by the government's enterprise credit information disclosure system, an enterprise DID will be generated. Enterprise employees need to be configured with additional role permissions to operate, and operation records will be linked to the enterprise DID and uploaded to the blockchain. The credit management submodule includes credit scoring and handling of credit defaults; Credit scoring: Based on user on-chain behavior and external credit data, a credit scoring model is built, and the scoring results are updated in real time and associated with the user's DID; Handling of Dishonesty: If a user engages in false listings, malicious rent arrears, damages property equipment, or other dishonest behavior, their DID will be marked as "dishonest" and synchronized to all nodes; dishonest users will be unable to initiate new rental requests, and already initiated requests will be automatically blocked.

4. The online housing and real estate rental system based on blockchain technology according to claim 1, characterized in that: In the housing information collection submodule, landlords upload basic housing information, multimedia information, and property rights information through the system. The system provides upload guidelines, requiring housing photos to include the four corners of the house and close-up views of the main equipment, VR videos to fully cover every room of the house, and images with a resolution of no less than 1600px, avoiding backlighting and blurriness. The property authenticity verification submodule connects to the government API of the real estate registration center. By inputting the property's real estate unit number and the owner's ID number / enterprise unified social credit code, it retrieves data on the property's location, area, mortgage status, seizure information, and co-ownership status to verify the legality and availability of the property rights. If the city where the property is located does not have access to the government API, it queries the property's ownership information through a partner bank. The landlord must sign an authorization agreement allowing the bank to connect to the real estate registration center through its professional system, and the verification results are synchronized to the system. A ViT-based image recognition model is used to detect whether property images show signs of Photoshop manipulation or are reused fake images. Frame consistency checks are performed on the property's VR video to prevent video editing fraud. The property listing on-chain module includes a unique property identifier, owner's DID, property core attribute hash, property status, and authenticity verification result. Original property images, VR videos, and scanned copies of property ownership certificates are stored in IPFS, and the system records the corresponding IPFS hash value to facilitate quick retrieval of original data through the hash value.

5. The online housing and real estate rental system based on blockchain technology according to claim 1, characterized in that: The user profile construction submodule includes basic information features, preference features, behavioral features, and user vector fusion; Basic information features: Extract user age, occupation, income level, and family situation, and quantify them into feature vectors; Preference features: A preference feature vector is constructed by combining user-initiated settings, decoration style, floor preference, and passive data collected by the system. Behavioral characteristics: Extract users' browsing history, favorites history, house viewing history, and historical transaction history, and use the Word2Vec algorithm to convert the behavioral sequence into a 128-dimensional behavioral embedding vector historyembedding; User vector fusion: The basic information feature vector, preference feature vector, and behavior embedding vector are normalized and then weighted to generate the final user vector user_vec. The property feature modeling submodule includes structured features, unstructured features, and property vector fusion; Structured features: Extract housing price, area, unit type, floor, commute time, and supporting facilities, and quantify them into a structured feature vector house_struct_vec; Unstructured features: A CNN convolutional neural network is used to extract features from the property images, generating a 256-dimensional image embedding vector; a Transformer model is used to semantically encode the property description text, generating a 128-dimensional text embedding vector; the image embedding vector and the text embedding vector are concatenated to obtain the unstructured feature vector house_unstruct_vec; Property Vector Fusion: After normalizing the structured feature vector and the unstructured feature vector, they are fused with weights of 0.6 and 0.4 to generate the final property vector house_vec; The recommendation strategy execution submodule includes a filtering layer, a similarity matching layer, a collaborative filtering layer, a ranking layer, and diversity adjustment. Filtering layer: Based on user-defined hard conditions, all listings are filtered to select a set of candidate listings; Similarity matching layer: Calculate the cosine similarity between the user vector user_vec and the candidate house vector house_vec, using the following formula: CosineSimilarity=(user_vec·house_vec) / (||user_vec||× ||house_vec||) Candidate listings are sorted in descending order of similarity, and the top 100 are selected as the listings to be sorted. Collaborative Filtering Layer: Based on the principle of "similar users have similar preferences", find historical users whose vector similarity to the current user is ≥0.8, extract the favorite and transaction listings of these users; calculate the feature similarity between the listings to be sorted and these listings, and generate collaborative filtering scores; Sorting layer: Combine "content matching score", "collaborative filtering score" and "popularity / freshness score" to generate a comprehensive score; sort by comprehensive score in descending order, and select the top 10 as the final recommended listings; Diversity Adjustment: If the proportion of properties in the same area or price range among the recommended properties is ≥60%, diversity optimization is triggered, replacing some properties with properties in other areas / prices that have the second highest overall scores, to ensure diversity in recommendations.

6. The online housing and real estate rental system based on blockchain technology according to claim 1, characterized in that: The transaction matching submodule has a built-in IM tool that supports communication between tenants and landlords via text, voice, images, and video. Communication records are encrypted and stored, and the hash values ​​of key communication content are recorded on the blockchain. Tenants can select viewing times online, and landlords can confirm or adjust the times through the system. The system generates a viewing appointment form, which includes the property ID, tenant DID, landlord DID, and viewing time. The appointment form's hash is recorded on the blockchain. Tenants submit their rental intentions for satisfactory properties, filling in their desired rental period and payment method. The system synchronizes the intentions with the landlord. After the landlord confirms the intention, a draft rental agreement is generated. The contract signing submodule is based on a standardized system template and automatically generates an electronic contract by combining the terms confirmed by the tenant and landlord, including rent, lease term, deposit amount, and maintenance responsibility. The contract includes the DIDs of both parties, the property ID, and the smart contract address for deposit escrow and rent payment. Using a digital certificate issued by an authoritative CA institution, the tenant and landlord complete the online signing. After signing, the contract calculates a hash value and synchronizes it to the smart contract engine of the blockchain basic service module. The hash value is stored on the blockchain as evidence, and the original contract is stored in IPFS. The system automatically synchronizes the contract hash value and filing information to the government regulatory node. The filing information includes the contract number, the DIDs of both parties, and the property ID, completing the online filing. The filing result is fed back to both parties in real time. In the payment and settlement submodule, tenants bind their bank cards or digital wallets to the system, and the system triggers a smart contract to automatically deduct rent according to the payment cycle agreed in the contract. After payment is completed, the payment transaction hash is recorded on the blockchain, while the payment voucher is stored off-chain. After the tenant pays the deposit, the system transfers the deposit to a smart contract escrow account. The smart contract records the deposit status, and status changes are recorded on the blockchain. During the rental period, the system automatically records the collected and paid fees such as water, electricity, and property management fees, and the fee details are recorded on the blockchain. When terminating the lease, the system automatically calculates the total fees and deducts the due fees from the deposit.

7. The online housing and real estate rental system based on blockchain technology according to claim 1, characterized in that: In the image acquisition and preprocessing submodule, when a tenant terminates a lease, the landlord / inspection personnel use the system's app to take images of the house's equipment. These images must cover all major equipment, including beds, sofas, air conditioners, refrigerators, lights, floors, walls, and bathroom fixtures. Each piece of equipment should be photographed from 3-5 angles, including front, side, and close-up views of key components. The system provides photo-taking guidance, automatically detects image blur, and prompts for reshooting when blur exceeds a threshold. The acquired images undergo size standardization, lighting correction, and noise removal. RGB images are converted to BGR format for subsequent model input. The device localization and recognition submodule uses a MaskR-CNN model pre-trained on the COCO dataset, optimized through transfer learning to adapt to the device recognition scenario in a house. The model performs object detection on the pre-processed image, outputting the device's bounding box and instance mask to locate the device's position in the image. Supported device categories include, but are not limited to, beds, sofas, air conditioners, refrigerators, washing machines, lamps, floors, walls, toilets, and sinks. Based on the device's features, the model outputs device category labels and confidence scores, with a confidence threshold of 0.

7. Devices below the threshold are marked as "awaiting manual confirmation." The system extracts the segmented image of each device separately for subsequent damage detection. The damage detection submodule uses ResNet-50 as the base network to build binary and multi-class classification models. The model is trained on a labeled dataset, using the cross-entropy loss function, the Adam optimizer, the learning rate 1e-3, the batch size = 32, and the training time is 30 epochs. For damages that require pixel-level localization, the U-Net model is used for defect segmentation, outputting a mask of the damaged area, calculating the ratio of the damaged area to the total area of ​​the device, and quantifying the degree of damage. For damage scenarios without labeled data, a heuristic algorithm is used to calculate the damage score: the device segmentation image is converted into a grayscale image, and the standard deviation and mean of the grayscale values ​​are calculated. The scoring formula is as follows: DamageScore=(std(gray)×0.5)+((128-mean(gray))×0.01) Set a damage threshold; a score greater than the threshold is considered damage, and the higher the score, the more severe the damage. The damage assessment and reporting submodule combines damage classification results, damage segmentation area proportions, and heuristic damage scores to generate an equipment damage assessment table, which includes equipment ID, category, damage type, damage severity, and confidence level. The confidence level is the same as the confidence level output by the model. The system automatically generates a damage report for returned equipment, which includes the damage assessment table, the original and segmented images of the damaged equipment, and damage score details. The report hash value is uploaded to the blockchain, and the original report is stored in IPFS. If the damage detection confidence level is <0.8, the system marks the device as "awaiting manual review" and provides a manual annotation interface; After the manual review results are updated, a new report is generated and uploaded to the blockchain.

8. The online housing and real estate rental system based on blockchain technology according to claim 1, characterized in that: In the data encryption submodule, off-chain data is stored using the AES-256 encryption algorithm, and the key is generated jointly by the user's DID and the system master key; on-chain data is encrypted using the national cryptographic SM2 algorithm to ensure that the data cannot be leaked; all data transmission in the system uses the SSL / TLS1.3 protocol, and API interface calls use token authentication + signature verification to prevent data from being stolen or tampered with. The access control submodule is based on role-based access control (RBAC): it defines roles such as administrator, landlord, tenant, regulatory agency, and bank, and assigns different operation permissions to each role; permission change records are recorded on the blockchain to ensure traceability; users can only access their own related data, regulatory agencies can only view the data required for regulation, and must obtain authorization from the data owner to access other data, and the authorization record is recorded on the blockchain; The security detection and vulnerability remediation submodule employs an Intrusion Detection System (IDS) and an Intrusion Prevention System (IPS) to monitor system access logs and transaction logs in real time, detect abnormal operations, and trigger alarms when anomalies are detected. Weekly vulnerability scans are performed on the system using the Nessus vulnerability scanning tool to conduct security checks on blockchain nodes, servers, and databases. Quarterly penetration tests are conducted to simulate attack scenarios, identify, and remediate security vulnerabilities. A security emergency response mechanism is established to immediately activate emergency plans upon discovering vulnerabilities or data breaches, and to record the incident handling process and results on the blockchain.

9. The online housing and real estate rental system based on blockchain technology according to claim 1, characterized in that: In the user account management submodule, administrators can create accounts with different roles, while ordinary users create accounts through the registration process, and the accounts are bound to DID; Supports account password reset, account freezing / unfreezing, and permission adjustment. Operation records are stored in the system log, and the log hash is uploaded to the blockchain. The parameter configuration submodule configures the parameters of the blockchain consensus algorithm, smart contract, and recommendation algorithm; it also configures the parameters of the data acquisition device and supports device calibration and status monitoring. The log management submodule collects system operation logs, transaction logs, and node logs. The logs include the operator's DID, operation time, operation content, and operation result. The logs are stored in the time-series database InfluxDB and support queries by time, operation type, and user DID. The hash values ​​of key logs are uploaded to the blockchain to ensure that the logs are tamper-proof. It supports visual analysis of logs and triggers alarms when abnormal logs are detected.

10. A rental method for an online housing and real estate rental system based on blockchain technology according to any one of claims 1-9, characterized in that: Includes the following steps: Step 1: System Initialization: Configure consortium blockchain nodes, consensus / recommendation / damage identification parameters, deploy smart contracts, and create initial accounts; Step 2: User authentication: The user submits information, completes authentication with UnionPay via CTID, generates DID, and hashes the authentication result onto the blockchain; Step 3: Property Listing and Verification: Landlords upload property information and verify it through government API / bank / AI. After verification, key property information is uploaded to the blockchain. Step 4: Intelligent Recommendation and Matching: Generate user and property vectors, filter, match similarity, perform collaborative filtering and diversity adjustment to output recommended properties, and communicate with both parties to reach an agreement; Step 5: Contract Signing and Payment: Generate an electronic contract and digitally sign it; the contract hash is uploaded to the blockchain. The tenant pays the deposit and rent; the deposit is held in escrow by the smart contract; the rent deduction record is uploaded to the blockchain. Step 6: Damage identification and deposit processing upon termination of lease: Collect and preprocess equipment images, locate the equipment, detect damage and generate a report; if there is no dispute, the deposit will be refunded; if there is a dispute, it will be handled through arbitration. Step 7: Data storage and update: Key data hashes are stored on-chain, and raw data is stored off-chain to update user credit and property status; Step 8: System monitoring: Collect full-process logs, perform real-time security checks, and regularly back up and patch vulnerabilities.

Images