Certificate generation method and apparatus, device, and readable storage medium

Abstract

A certificate generation method and apparatus, a device, and a readable storage medium. The method comprises: using a text generation model to automatically generate corresponding summary information for an object to be authenticated uploaded by an application end; filling in a target description template corresponding to said object with the summary information to obtain a target description file; correcting the target description file on the basis of a user correction operation at the application end to obtain a file to be authenticated, wherein the file to be authenticated is used to distinguish and mark the automatically-generated summary information and information corrected on the basis of the user correction operation; and sending the file to be authenticated and said object to a corresponding authentication end, and obtaining a corresponding digital certificate issued by the authentication end after reviewing and approving the file to be authenticated and said object.

Description

Method, device, equipment and readable storage medium for generating a certificate

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to the Chinese patent application filed with the China Patent Office on June 16, 2023, with application number 202310722075.X, and application name “A method, device, equipment and readable storage medium for generating a certificate”, the entire contents of which are incorporated by reference into this application. Technical Field

[0003] The present application relates to a credential generation method, apparatus, device and readable storage medium. Background Art

[0004] Currently, when users apply for a certificate, they need to manually fill in the product profile, product functions and other information of the certified product, which makes the application process time-consuming and labor-intensive and prone to errors. At the same time, the lack of unified filling standards makes the review work of the certificate issuer difficult.

[0005] Therefore, how to simplify user operations during the application and issuance of digital certificates is a problem that those skilled in the art need to solve.

[0006] Summary of the Invention

[0007] According to an embodiment disclosed in the present application, in a first aspect, a method for generating a credential is provided, comprising:

[0008] Use the text generation model to automatically generate corresponding summary information for the object to be authenticated uploaded by the applicant;

[0009] Fill the summary information into the target description template corresponding to the object to be authenticated to obtain the target description file;

[0010] Modify the target description file according to the user modification operation of the application end to obtain the file to be authenticated; the file to be authenticated distinguishes and marks the automatically generated summary information and the information modified by the user modification operation; and

[0011] Send the document to be authenticated and the object to be authenticated to the corresponding authentication end, and obtain the corresponding digital certificate issued by the authentication end after reviewing the document to be authenticated and the object to be authenticated.

[0012] According to an embodiment disclosed in the present application, in a second aspect, a credential generation device is provided, comprising:

[0013] The generation module is used to automatically generate corresponding summary information for the object to be authenticated uploaded by the applicant using the text generation model;

[0014] The filling module is used to fill the summary information into the target description template corresponding to the object to be authenticated to obtain the target description file;

[0015] A correction module is used to correct the target description file according to the correction operation of the user at the application end to obtain a file to be authenticated; the file to be authenticated distinguishes and marks the automatically generated summary information and the information corrected by the user correction operation; and

[0016] The sending module is used to send the document to be authenticated and the object to be authenticated to the corresponding authentication end, and obtain the corresponding digital certificate issued by the authentication end after reviewing the document to be authenticated and the object to be authenticated.

[0017] According to an embodiment disclosed in this application, in a third aspect, an electronic device is provided, including:

[0018] a memory for storing computer-readable instructions; and

[0019] A processor is configured to execute computer-readable instructions to implement the credential generation method of any embodiment of the first aspect.

[0020] According to the embodiments disclosed in the present application, in a fourth aspect, a non-volatile computer-readable storage medium is provided for storing computer-readable instructions, wherein the computer-readable instructions, when executed by a processor, implement the credential generation method of any embodiment of the first aspect.

[0021] The details of one or more embodiments of the present application are set forth in the accompanying drawings and the description below. Other features and advantages of the present application will become apparent from the description, drawings, and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without any creative work.

[0023] FIG1 is a flow chart of a method for generating a credential according to some embodiments of the present application;

[0024] FIG2 is a schematic diagram of a blockchain-based trusted credential system according to some embodiments of the present application;

[0025] FIG3 is a schematic diagram of obtaining a text summary in some embodiments of the present application;

[0026] FIG4 is a schematic diagram of a CLIP training process according to some embodiments of the present application;

[0027] FIG5 is a schematic diagram of generating text descriptions of images in some embodiments of the present application;

[0028] FIG6 is a schematic diagram of obtaining text summaries in some other embodiments of the present application;

[0029] FIG7 is a schematic diagram of generating a text description of a video according to some embodiments of the present application;

[0030] FIG8 is a schematic diagram of a credential generation device according to some embodiments of the present application;

[0031] FIG9 is a schematic diagram of an electronic device according to some embodiments of the present application;

[0032] FIG10 is a schematic diagram of a server structure in some embodiments of the present application;

[0033] FIG11 is a schematic diagram of a terminal structure in some embodiments of the present application;

[0034] FIG12 is a schematic diagram of application scenarios of some embodiments of the present application;

[0035] FIG13 is a schematic diagram of the structure of a non-volatile computer-readable storage medium according to some embodiments of the present application. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0037] Currently, when applying for a certificate, users are required to manually fill in information such as the product overview and features of the certified product, making the application process time-consuming and prone to errors. Furthermore, the lack of a unified filling standard makes the review process difficult for the certificate issuer. Therefore, this application provides a certificate generation solution that simplifies user operations during the digital certificate application and issuance process, improving the efficiency and accuracy of certificate issuance.

[0038] Refer to Figure 12, which is a schematic diagram of the application scenario of an embodiment of the present application, wherein the application end 1201, the authentication end 1202, the blockchain system 1203 and the trusted certificate issuing system 1204 in Figure 12 communicate through the network, wherein the application end 1201, the authentication end 1201 and the trusted certificate issuing system 1204 in the blockchain system 1203 can implement the certificate generation method involved in the embodiment of the present application based on the computer device and its functions.

[0039] Referring to FIG. 12 and FIG. 1 , an embodiment of the present application discloses a credential generation method that can be applied to a blockchain-based trusted credential issuance system 1204, including:

[0040] S101: Use a text generation model to automatically generate corresponding summary information for the object to be authenticated uploaded by the applicant.

[0041] In this embodiment, the objects to be authenticated can include literary works, paintings, digital collections, medical certificates, charity application certificates, government certificates, contracts, design documents, and the like. These objects can be submitted to the trusted credential issuance system in the form of text, images, and / or videos. This system can be a web-based system, with the application end being a user terminal provided by the system for applying for digital certificates. Accordingly, the summary information of the object to be authenticated is a textual description of the object, intended to briefly introduce the object's main content, functions, and purpose.

[0042] In some examples, the text generation model can be a neural network-based text key information extraction model, a natural language processing-based content generation model, or a neural network-based image and text matching model. Therefore, in some embodiments, the text generation model is used to automatically generate corresponding summary information for the object to be authenticated uploaded by the applicant. This includes: if the object to be authenticated is text, the text generation model is used to extract key information from the text and obtain summary information based on the key information; in this case, the text generation model is a neural network-based text key information extraction model. If the object to be authenticated is text, the text generation model is used to perform natural language processing on the text to obtain summary information; in this case, the text generation model is a natural language processing-based content generation model. If the object to be authenticated is an image and / or video, the text generation model is used to extract pixel features from the image and / or video, match the pixel features with corresponding text description information, and obtain summary information based on the text description information; in this case, the text generation model is a neural network-based image and text matching model. The specific structure and training process of the text generation model can be referenced in current related art and will not be further described in this embodiment.

[0043] In some embodiments, a text generation model is used to extract key information from a text, including: using the text generation model to split the text into sentences, and constructing a similarity matrix for calculating the similarity between sentences based on the vector representation of each sentence; constructing a similarity graph based on the similarity matrix with sentences as nodes and similarities as edges; and selecting the top N sentences with similarity rankings as key information based on the similarity graph. It can be seen that the process of generating summary information for a text using a text generation model includes: inputting text information into a text generation model, and the text generation model splitting the input text information into sentences; determining a vector representation for each sentence to establish a similarity matrix to calculate the similarity between sentences; constructing a similarity graph (a graph structure with sentences as nodes and similarity scores as edges) based on the similarity matrix; calculating sentence rankings based on the similarity graph; selecting the top N sentences for aggregation, generating summary information for the text, and outputting it. In this way, the key information for generating summary information can be obtained using the text extraction method. The value of N is configurable.

[0044] In some embodiments, extracting key information from a text using a text generation model includes: using the text generation model to split the text into sentences and generate a term sequence for each sentence; determining a vector representation for each sentence based on the term sequence of each sentence; constructing a document vector based on the vector representation of each sentence, and determining summary information corresponding to the document vector using a beam search algorithm; and dynamically selecting key information from the summary information and each sentence. The document vector includes contextual information. Thus, using text extraction and abstraction methods, key information for generating summary information can be obtained. In this case, the text generation model may include: a sentence extractor, a tokenizer, a sentence encoder, a Transformer network, an autoregressive decoder, and a pointer network. Specifically, the sentence extractor splits the text into sentences; the tokenizer generates a term sequence for each sentence; the sentence encoder determines a vector representation for each sentence based on the term sequence of each sentence; the Transformer network constructs a document vector based on the vector representation of each sentence using a mean pooling operation; the autoregressive decoder determines the summary information corresponding to the document vector using a beam search algorithm; and the pointer network dynamically selects key information from the summary information and each sentence. Because sentence vectors contain contextual information, the Transformer network's self-attention mechanism further refines the sentence vector's features. This is then fed into an autoregressive encoder through a mean pooling operation to form a document vector, effectively combining extraction and abstraction. This approach uses the sentence extractor, tokenizer, and sentence encoder to first anchor the text context using an extraction approach. Then, through the Transformer network, autoregressive decoder, and pointer network, key information is determined using an abstraction approach, constrained by the previously anchored text context. This ensures that the final key information does not deviate from the original text, combining the advantages of both extraction and abstraction, resulting in a more accurate and natural description.

[0045] Among them, the process of using the text generation model to generate summary information for the image includes: inputting the image into the text generation model, the text generation model extracts pixel features from the input image, and obtains a set of candidate text prompts that match the pixel features; selects the texts with the top N matching degrees from the candidate text prompt set, and organizes them into text descriptions, thereby obtaining the summary information of the image.

[0046] In some embodiments, a text generation model is used to automatically generate corresponding summary information for the object to be authenticated uploaded by the applicant, including: if the object to be authenticated is a video, then using the text generation model to extract video frames from the video to obtain a video frame sequence; matching corresponding text description information for each video frame in the video frame sequence; summarizing all text description information to obtain a text file; extracting key information from the text file, and obtaining summary information based on the key information. The method of extracting video frames from the video to obtain a video frame sequence using the text generation model includes: using the text generation model to divide the video into multiple video segments, extracting a single video frame from each video segment to obtain a video frame sequence; or using the text generation model to determine adjacent video frames in the video as adjacent video frame groups; if the difference between two video frames in any adjacent video frame group is greater than a preset threshold, retaining the next video frame in the current adjacent video frame group; otherwise, deleting the next video frame in the current adjacent video frame group; and summarizing all retained video frames to obtain a video frame sequence.

[0047] As can be seen, this embodiment matches corresponding text description information to each video frame in the video frame sequence, and aggregates all text description information to obtain a text file. This allows videos that are difficult to process to be converted into text for processing, reducing the processing difficulty. It also supports the processing of multiple data modalities (text, images, and videos), making it more widely applicable. After converting the video into a text file, the corresponding key information can be obtained according to the aforementioned processing method for text, namely: using a text generation model to split the text file into sentences and generate a word sequence for each sentence; determining a vector representation of each sentence based on the word sequence of each sentence; constructing a document vector based on the vector representation of each sentence, and using a beam search algorithm to determine the summary information corresponding to the document vector; and dynamically selecting key information from the summary information and each sentence. This can achieve: a combination of data modality conversion, extraction, and abstraction methods, and can also reuse the aforementioned sentence extractor, word segmenter, sentence encoder, Transformer network, autoregressive decoder, and pointer network modules to improve their utilization.

[0048] In some examples, the text generation model can also divide the video into frames, extract pixel features for each frame, and then match the features obtained by summarizing the pixel features of all video frames with corresponding text description information to obtain video summary information. Specifically, pixel features are extracted for each video frame, and a set of candidate text prompts matching the pixel features is obtained. The top N matching text prompts are selected from the candidate text prompts and organized into a text description, thereby obtaining a text description for each video frame.

[0049] If necessary, you can load a multilingual translation model as needed, automatically translate the text output by the text generation model into the target language, and then return it to the user.

[0050] S102: Fill the summary information into the target description template corresponding to the object to be authenticated to obtain a target description file.

[0051] It should be noted that the target description template is a file with a standardized format developed by the relevant issuing agency. Therefore, by filling the summary information into the target description template, a target description file with a standardized format can be obtained, thereby ensuring the standardization of subsequent documents to be authenticated and facilitating the issuance of digital certificates.

[0052] As mentioned above, the objects requiring authentication can include literary works, paintings, digital collectibles, medical certificates, charitable application certificates, government certificates, contracts, design documents, and so on. Therefore, the corresponding issuing organizations are also diverse. Each issuing organization can develop a corresponding description template and upload it to the trusted certificate issuance system for users who need to access it. The user who applies for a digital certificate is the user on the application end. The authentication end is the user end provided by the trusted certificate issuance system for the issuing organization. Therefore, the relevant issuing organization can upload the description template through the authentication end and conduct the review process.

[0053] In some embodiments, before automatically generating corresponding summary information for the object to be authenticated uploaded by the applicant using the text generation model, the process also includes: receiving a credential application request sent by the applicant; returning at least one description template to the applicant based on the credential application request, so that the applicant displays at least one description template; at least one description template is uniformly set by the relevant issuing authority; obtaining the user's selection operation on the target description template and the uploaded object to be authenticated on the applicant; the object to be authenticated is text, image, and / or video. The user applying for a digital credential can flexibly select the description template corresponding to the object to be authenticated as the target description template, and submit the selected target description template and the object to be authenticated to the trusted credential issuance system.

[0054] For works in text, image, and / or video form, the method of this embodiment can be used to obtain their summary description information without providing watermarked thumbnails of the works, thereby reducing the risk of exposing confidential information of the works and having the effect of protecting the content of the works.

[0055] S103. Modify the target description file according to the user modification operation of the application end to obtain a file to be authenticated; the file to be authenticated distinguishes and marks the automatically generated summary information and the information modified by the user modification operation.

[0056] S104: Send the document to be authenticated and the object to be authenticated to the corresponding authentication end, and obtain the corresponding digital certificate issued by the authentication end after reviewing the document to be authenticated and the object to be authenticated.

[0057] It should be noted that the automatically generated summary information and the information modified by the user's correction operation are distinguished and marked in the file to be authenticated. Then, when the authentication end displays the file to be authenticated, it can clearly show the difference before and after the correction to assist the audit work and improve the audit efficiency.

[0058] After the document to be authenticated and the object to be authenticated are sent to the corresponding authentication terminal, the corresponding authentication terminal uses program software and / or relevant personnel to review the document to be authenticated and the object to be authenticated. If the review passes, the corresponding authentication terminal issues a corresponding digital certificate for the object to be authenticated and sends the corresponding digital certificate to the trusted certificate issuance system. If the review fails, the corresponding authentication terminal sends a failure result to the trusted certificate issuance system. After receiving the digital certificate or failure result, the trusted certificate issuance system sends this digital certificate or failure result to the original applicant to inform the original applicant of the relevant results. In some embodiments, the failure result sent by the authentication terminal after the review of the document to be authenticated and the object to be authenticated fails is obtained; the failure result is sent to the applicant terminal, so that the applicant terminal user can re-upload the object to be authenticated and / or correct the document to be authenticated; the failure result is stored in a database and stored in a blockchain system by calling a smart contract. The failure result may include: the object to be authenticated, the target description template, the target description file, the document to be authenticated, the identification information of the applicant terminal (such as the applicant terminal user ID), and the identification information of the authentication terminal (such as the authentication terminal user ID).

[0059] In this embodiment, all information records, documents, and other information involved in the credential application, issuance, verification, and authorization processes can be stored in a database and a blockchain system to ensure the immutability of the relevant data. Therefore, in some embodiments, the object to be authenticated, the target description template, the target description file, the document to be authenticated, and the digital certificate are stored in a database, and a smart contract is invoked to store them in the blockchain system. The database is the database of the trusted credential issuance system, and the blockchain system can be considered as a backup storage for this database. Accordingly, in some embodiments, the process further includes: generating an application record including the digital certificate, its validity period, its certificate number, identification information of the authenticating end, and identification information of the target description template; storing the application record in the database, and invoking a smart contract to store it in the blockchain system. The validity period of the digital certificate is set by the relevant issuing authority in accordance with relevant guidance documents, and the certificate number of the digital certificate is automatically generated by the trusted credential issuance system using a random algorithm or other algorithm. Accordingly, in some embodiments, the process further includes: querying the blockchain system for the object to be authenticated, the target description template, the target description file, the document to be authenticated, the digital certificate, and / or the application record.

[0060] After the digital certificate is issued, any user can check it. In some embodiments, the trusted credential issuance system also receives a verification request for the digital certificate; forwards the verification request to the applicant, so that the applicant displays the verification request; if the user of the applicant receives an authorization notice for the verification request, pushes the verifiable information specified in the authorization notice to the initiator of the verification request, so that the initiator displays the verifiable information; generates a verification and authorization record including the digital certificate, verifiable information, and identification information of the initiator; stores the verification and authorization record in a database, and calls a smart contract to store it in the blockchain system so that the verification and authorization record can be queried in the blockchain system. Among them, the initiator of the verification request can be any applicant, any authentication terminal, or the user portal used by a third-party verification user.

[0061] It should be noted that the certificate owner can independently specify the information that can be seen during third-party verification. Users with digital certificates can independently specify the verifiable information when authorizing third-party verification. The trusted certificate issuance system will push this information together with the authorization notification to the initiator of the verification request.

[0062] This embodiment can utilize a text generation model to automatically generate corresponding summary information for the object to be authenticated uploaded by the applicant, and automatically fill it into the target description template with a standard format corresponding to the object to be authenticated, thereby automatically obtaining a target description file with a standard format. The user of the applicant then only needs to modify the target description file to obtain the document to be authenticated, without having to manually fill in all the content. In addition, the document to be authenticated distinguishes between the automatically generated summary information and the user-corrected information. Therefore, after sending the document to be authenticated and the object to be authenticated to the corresponding authentication terminal, the authentication terminal can easily compare the automatically generated summary information with the user-corrected information, thereby facilitating the authentication review. If the authentication terminal reviews the document to be authenticated and the object to be authenticated and approves it, the corresponding digital certificate issued by the authentication terminal for the object to be authenticated can be obtained.

[0063] The trusted credential issuance system of this embodiment is implemented based on Verifiable Credentials (VC). VC is a W3C (World Wide Web Consortium) standard and specification that defines credential data formats and addresses credential trust interaction. VC constructs a verifiable credential data model using encryption algorithms and digital signatures. Its declared content, signatures, and metadata can be quickly verified by the verifier. This verification process verifies whether the credential conforms to the standard data request format, is validly signed, and has not been tampered with, thereby proving its validity and the authenticity of the subject. VC allows users to store and manage their own data. Only with explicit user consent and approval, and after selecting data according to a disclosure policy, can a ciphertext certificate be generated and presented to the verifier for verification. VC allows for selective disclosure of information based on data attributes, minimizing information disclosure. For example, when a company applies for a loan from a bank, if the bank's loan eligibility requirement is an annual income exceeding 3 million, the company only needs to prove whether its annual income is greater than 3 million, without providing detailed annual income data.

[0064] It can be seen that this solution can automatically obtain a target description file with a standardized format. The user only needs to modify the target description file to obtain the file to be authenticated, thereby simplifying the user operations in the digital certificate application process and improving the application efficiency and accuracy; since the file to be authenticated is obtained based on a target description template with a standardized format and the file to be authenticated distinguishes and marks the automatically generated summary information and the user-corrected information, it provides convenience for the authentication review work, is conducive to simplifying the user operations in the digital certificate issuance process, and can improve the issuance efficiency and accuracy.

[0065] It should be noted that the use cases of verifiable digital credentials involve the credential issuer, holder, and verifier. The issuer is typically an authoritative institution (such as a school, hospital, financial institution, or government department), an entity capable of issuing certificates. The holder is the owner of the certificate and has full control over the credential data. They can decide how to manage the certificate, with whom to share it, and even revoke it. The verifier obtains a verifiable presentation from the holder and verifies the authenticity and validity of the credential data, as well as the authenticity of the issuer and holder.

[0066] Generally speaking, the main process of creating a verifiable digital credential includes the following: (1) The issuing authority (i.e., the issuer) designs and defines the credential template based on the business characteristics. In the template, it can specify which attributes the declaration content (i.e., the summary information) includes, which attributes are required, and which are optional. (2) The issuing authority publishes the credential template and provides the corresponding credential issuance service; (3) The credential applicant (i.e., the holder) fills in the credential corresponding data (including: own identity, declaration content, etc.) according to the selected credential template and submits the credential creation application; (4) After obtaining the credential creation application, the issuing authority reviews the relevant data provided by the corresponding applicant. When the review is passed, the credential is issued and signed. (5) The credential holder can create a verifiable representation of selective information disclosure based on the issued credential, such as specifying that certain attributes that need to be disclosed are disclosed in plain text, while attributes that do not need to be disclosed are masked with ciphertext, or only disclose the proof of the declaration (zero-knowledge proof) without disclosing the entire verifiable credential.

[0067] A trusted credential issuance system can be built based on the data model for verifiable digital credentials defined by the W3C specification. Specifically, different cryptographic algorithms are applied to the data model to implement different VC credential applications based on business characteristics. In practical applications, interfaces for interaction with upper-layer applications can be implemented based on specific user needs, building a trusted credential issuance and interaction system. To ensure high data credibility and traceability, blockchain technology can be incorporated to power this trusted credential issuance and interaction system. Specifically, operations such as credential data confirmation and authorization can be stored on-chain, ensuring traceability, verifiability, and tamper-proofness of credential data interactions. For example, when creating a credential, data such as the credential declaration content can be traced back to its source on the blockchain.

[0068] Accordingly, this embodiment provides a blockchain-based trusted credential system, as shown in Figure 2, comprising modules such as a service interface (e.g., a Web service RESTful interface), identity authentication, credential template management, credential management, credential verification, data storage, data on-chain services, and credential declaration content generation. Based on these modules, the blockchain-based trusted credential system can provide users with Web services such as login authentication, template registration, credential application, and credential verification via a RESTful interface. The identity authentication module implements user identity login and permission control. The credential template management module provides credential template addition, deletion, modification, and query functions for authorized credential users (credential issuers). The credential management module provides credential application (creation), viewing, and deletion functions for general business users. The credential verification module provides credential verification, verifying the authenticity and validity of credential content. The data storage module provides storage functions (including file system storage and database storage) for various data (including credential templates, credential files, and related data). The data on-chain service stores automatically generated declaration content, user-modified content, the credential review and issuance process, and relevant data for credential verification and authorization on-chain. As can be seen, the data stored in the data on-chain service and the data storage module can back up each other. The credential declaration content generation module can be independently developed as an engine that provides automatic declaration content generation, used as a web service unit, and specifically provides automatic declaration content generation services for the trusted credential system through RESTful APIs. It can also be a module that encapsulates the automatic declaration content generation function.

[0069] In this embodiment, the data to be stored can be written to the blockchain ledger via a predefined smart contract interface. This indicates that data exchange between blockchain applications and the blockchain ledger is achieved through smart contracts, which serve as a bridge between blockchain applications and ledger data. A smart contract is essentially a computing program that executes contract terms. It runs on the blockchain and manipulates data on the ledger, allowing blockchain applications to interact with the blockchain ledger through it. Smart contracts are designed to automatically execute certain pre-defined actions (such as contract terms) when certain pre-defined conditions (such as payment terms, compliance terms, and time constraints) are met, completing a transaction and writing the transaction data to the ledger. The execution of smart contracts requires no human intervention; it is entirely automated by the contract program, eliminating the need for trusted intermediaries. This reduces transaction costs, mitigates the risk of malicious tampering, and significantly improves transaction efficiency.

[0070] Smart contracts can be written using smart contract programming languages ​​such as SCL, Solidity, Liquid, Go, Node.js, and Java. Different blockchain platforms support different smart contract programming languages. For example, the FISCO BCOS blockchain platform currently supports Solidity and Liquid, while another widely used blockchain platform, Hyperledger Fabric, supports Go, Node.js, and Java for writing smart contract code. This example uses the FISCO BCOS blockchain as an example. First, paste the smart contract's Solidity code into the Contract IDE (under Contract Management) on the WebBase platform. Then, go through the compilation and deployment steps to deploy the smart contract on the blockchain. Once the contract is successfully deployed, information such as the contract address, contract name, and contract ABI will be displayed below the IDE. Finally, the contract program and contract interface can be tested and verified using the Contract IDE's contract testing interface. Alternatively, the contract can be called through WebBase's RESTFul interface to upload data to the blockchain or retrieve data from the blockchain.

[0071] The trusted credential system and blockchain provided by this embodiment include at least the following contract interfaces: Contract interface 1: the system writes the automatically generated declaration content for the text into the blockchain account book by calling the contract interface; Contract interface 2: the system writes the modified content of the user's declaration content for interface 1 into the blockchain account book by calling the contract interface; Contract interface 3: the system writes the automatically generated text description information for the image into the blockchain account book by calling the contract interface; Contract interface 4: the system writes the modified content or data of the user's text description information for interface 3 into the blockchain account book by calling the contract interface; Contract interface 5: the system writes the user's application for creating a credential into the blockchain account book by calling the contract interface. Please record and write it into the blockchain account book; Contract interface 6: The system writes the audit results of a certain certificate by the authorized agency into the blockchain account book by calling this contract interface; Contract interface 7: The system obtains all content or data generated during the creation process of a certain certificate from the blockchain by calling this contract interface; Contract interface 8: The system writes the user's application and authorization record for viewing or verifying the content of a certain certificate into the blockchain account book by calling this contract interface; Contract interface 9: The system writes the certificate holder's authorization record for a certain inspection application into the blockchain account book by calling this contract interface; Contract interface 10: The system obtains records and data related to the authorization of a certain certificate from the blockchain by calling this contract interface.

[0072] This embodiment encapsulates a module with the function of automatically generating declaration content, which can automatically generate declaration content based on text or images and allow users to modify or edit it. The automatically generated declaration content will be stored in the blockchain through the data chain service, and the modifications made by the user to the automatically generated declaration content will also be stored on the chain through the data chain service, thereby ensuring that the source of all data is traceable and improving the authenticity and reliability of the data. At the same time, after receiving the certificate creation application, the authorized agency (certificate issuing agency) can also quickly distinguish which data is automatically generated by the system and which data is modified by the user, and can make a quick judgment on the declaration content of the certificate based on these data sources and processing traces, thereby improving the efficiency of certificate issuance.

[0073] Generally, credential declaration content includes two types: (a) certain attribute information possessed by the entity, and (b) summary information that needs to be obtained from the data. This information can be extracted from text and images. Therefore, the credential declaration content generation module mainly has text summary acquisition function and image text description function.

[0074] There are two main methods for obtaining text summarization: extractive and abstract. The extractive method selects the most important subset of words based on ranking technology to summarize the article without modifying the original content. The abstract method uses natural language understanding (NLU) and natural language generation (NLG) to organize language wording from the perspective of semantic understanding, summarize the text content to generate a new short description. The disadvantages of the extractive method are that the generated content is redundant, contains unnecessary or non-critical information, and the connection between sentences is abrupt; the abstract method generation process often lacks the control and guidance of key information, which can easily lead to large deviations in the generated content.

[0075] Taking the extraction method as an example, the workflow of the extraction method shown in Figure 3 includes: inputting text information; splitting the input text information into sentences; determining the vector representation for each sentence to establish a similarity matrix (cosine similarity can be used, that is, using the cosine of the angle between two vectors to measure the similarity between vectors of different sentences) to calculate the similarity between sentences; constructing a similarity graph based on the similarity matrix (a graph structure with sentences as nodes and similarity scores as edges); calculating the sentence ranking based on the similarity graph, and the calculation method can use the PageRank or TextRank method; selecting the top N sentences for summary, generating a summary of the text and outputting it.

[0076] As mentioned above, when the credential declaration content module is called, it can be divided into two modes, including: functional module and Web service unit; when the module is integrated into the system as a functional module, an interface function String TextExtract(String input,int type) can be defined. The input parameter input can be a text content string or a local path of a text file, and the type of the parameter input is specified by the parameter type; the output of the interface function is a text string, that is, a text summary.

[0077] When the credential declaration content module provides a call interface as a Web service unit, a RESTful interface can be defined as the interaction interface, supporting both POST and GET methods. Taking POST as an example, the interface URL is: http: / / localhost:port / text-extract (Note: HTTPS is also supported), where localhost is the host address where the service unit is deployed, such as DNS or IP; port is the listening port number of the Web server, which defaults to 80; the call method is HTTP POST; and the request parameters are shown in Table 1.

[0078] Table 1

[0079] The returned text summary string is as follows:

[0080] {

[0081] textAbstract:"A sample abstract"

[0082] }

[0083] When the text size exceeds 8K, it needs to be saved as a text file and then POSTed to the Web service unit for processing. The interface URL is: http: / / localhost:port / long-text-extract; the calling method is: HTTP POST; the request parameters are as shown in Table 2, and the file object is usually submitted in the FormData form.

[0084] Table 2

[0085] The returned text summary string is as follows:

[0086] {

[0087] textAbstract:"A sample abstract"

[0088] }

[0089] After obtaining the text summary information, the system will automatically call the data on-chain service to write the automatically generated summary information into the blockchain through a pre-defined contract interface (i.e., contract interface 1) for subsequent data tracing and data source confirmation.

[0090] The function of generating text descriptions for images can generate text descriptions that humans can understand given an image. This task is relatively easy for humans to complete, but it is very challenging for machines because it requires both understanding the content of the image and converting this understanding into natural language. The CLIP (Contrastive Language-Image Pre-training) model can achieve this function. CLIP can align text and images in the feature domain. As shown in Figure 4, in order to establish a connection between images and texts, CLIP extracts features from images and texts separately during training. The backbone model for image feature extraction can use the ResNet series model or the VIT (Vision Transformer) series model, and the text feature extraction uses the BERT (Bidirectional Encoder Representation from Transformers) model; after feature extraction, all text features T1, T2, T3...T N , and image features I1, I2, I3...I N The inner products are calculated separately to form a matrix. From the perspective of the image, the rows represent a classifier, while from the perspective of the text, the columns represent a classifier. Given the matching relationship between text and images in a batch, the training objective is to maximize the inner product of the same pair of image and text features (i.e., the elements on the matrix diagonal), while minimizing the inner product of unrelated features (i.e., the elements on the off-diagonal lines of the matrix). Through continuous iterative training, pairings between text features and image features are established, forming a mapping relationship between image and text.

[0091] Specifically, this embodiment implements the function of generating a text description for a given image based on CLIP-Interrogator. The text generation module is developed and implemented based on the open source Python code library of CLIP-Interrogator. As shown in Figure 5, an image file is input to the text generation module, and the text generation module can output the corresponding text description. The image file can be obtained from the storage path entered by the user (which can be a resource descriptor such as a URL, or a file path in the local file system), then saved to a specified processing directory, and the corresponding storage path is obtained as the input of CLIP-Interrogator; CLIP-Interrogator loads the pre-trained CLIP model according to the configuration information, performs inference processing on the input image file, and obtains a set of candidate text prompts that match it; the final text prompts are selected from the candidate text prompt set (such as sorting by matching degree to obtain the top N) and organized into a text description. Here, a multilingual translation model can be loaded as needed to automatically translate the output text into the target language and finally return it to the caller.

[0092] After obtaining the image description information, the system will automatically call the data on-chain service to write the automatically generated image description information into the blockchain through a pre-defined contract interface (i.e., contract interface 3) for subsequent data tracing and data source confirmation.

[0093] This embodiment can also combine the extraction method and the abstraction method. The disadvantage of the extraction method is that the generated content is redundant, contains unnecessary or non-critical information, and the connection between sentences is stiff; the generation process of the abstraction method often lacks the control and guidance of key information, which easily leads to large deviations in the generated content. Therefore, combining the extraction method and the abstraction method can solve the problems of the extraction method and the abstraction method respectively, and combine the advantages of the two to make the final generation result more accurate and the description more natural. As shown in Figure 6, this embodiment combines the extraction method and the abstraction method to realize a text generation model including a sentence extractor, a word segmenter, a sentence encoder, a Transformer network, an autoregressive decoder and a pointer network. The text generation process in the model includes: inputting a document into the sentence extractor to obtain a set of candidate sentences: S1, S2, S3 and S4; inputting the candidate sentence set into the word segmenter to obtain a word sequence for each sentence; inputting the word sequence for each sentence into the sentence encoder to obtain a corresponding sentence vector based on the word sequence of each sentence: x i,1 、x i,2 、x i,3 、x i,4 ; Among them, the single sentence vector x integrates the information of each sentence, such as: x i,1 is the sentence vector of sentence S1 that combines the information of sentences S1, S2, S3 and S4, x i,2The sentence vector for sentence S2 is a fusion of information from sentences S1, S2, S3, and S4, so a single sentence vector contains contextual information. All sentence vectors are fed into a Transformer network, where mean pooling is performed on all sentence vectors to produce a document vector. The document vector is fed into an autoregressive decoder to obtain candidate summaries using a beam search algorithm. A pointer network dynamically extracts words from the sentence and candidate summaries based on the context to produce the final summary. The autoregressive decoder can be implemented using the Transformer. This approach allows for the general direction of textual context to be anchored before content generation is performed using an abstraction approach. This ensures that the final key information, constrained by the previously anchored textual context, aligns with the true meaning of the original text.

[0094] In Figure 6, h j,1 、h j,2 、h j,3 、h j,4 is an input to the autoregressive decoder, which is processed by the autoregressive decoder together with the document vector. When the autoregressive decoder is first used, h j,1 is a special character: <sos>:start of sentence, which and the document vector are input to the autoregressive decoder, so that the autoregressive decoder outputs the word y with the highest probability j,1 ; Then based on y j,1 Get h j,2 ; Then h j,1 、h j,2 And the document vector is input to the autoregressive decoder, so that the autoregressive decoder outputs the word y with the highest probability j,2 ; The next step is based on y j,2 Get h j,3 ; Then h j,1 、h j,2 、h j,3 And the document vector is input to the autoregressive decoder, so that the autoregressive decoder outputs the word y with the highest probability j,3 ; The next step is based on y j,3 Get h j,4 ; Then h j,1 、h j,2 、h j,3 、h j,4 And the document vector is input to the autoregressive decoder, so that the autoregressive decoder outputs the word y with the highest probability j,4 ; The autoregressive decoder will then output <eos>:end-of-sentence, then stop iteration. j,1 、y j,2 、y j,3 、y j,4 It is the word vector obtained after the autoregressive decoder performs masked multi-head attention processing on the input data.

[0095] As mentioned above, the input image can be a picture or a video. Please refer to Figure 7. When the image file is transferred to the processing engine server through the RESTful API interface, the server program first determines whether the input file is a video file. If not, it directly generates the corresponding text description content for the image file; if it is, it extracts the image from the video file according to a certain strategy to generate a set of video frame sequences; the image extraction strategy supports extracting one video frame every N (the value is adjustable); or judging based on the pixel difference between the adjacent previous and next frames, if the pixel difference ratio between the current frame and the previous frame exceeds a certain threshold (configurable), the current frame is extracted; otherwise, the current frame is not extracted. For the obtained video frame sequence, a text description content is generated for each video frame therein, and all the generated text description contents are summarized into a text file. Then, the text file is used as the text that needs to be extracted, and it is processed according to the text summary acquisition method provided in this application until the final summary is obtained. The final summary is returned to the caller through the processing engine server. Therefore, the video is first converted into text and then processed using extraction and abstraction methods. This not only supports the processing of multiple data modalities and improves applicability, but also converts difficult-to-process videos into simpler text types, reducing processing difficulty and improving processing efficiency.

[0096] The blockchain-based trusted credential system provided in this embodiment can automatically generate credential declaration content. It not only supports the automatic acquisition of summary information for text content, but also supports the automatic generation of text descriptions for images. The automatically generated credential declaration content can then be stored on-chain via a data on-chain service, ensuring data source traceability and ensuring data quality. Furthermore, the automatically generated credential declaration content can be further modified or edited by the user, and the modified data is also stored on-chain, ensuring traceability of data processing operations. Automatically generating credential declaration content reduces the risk of user input errors, facilitating rapid entry of credential declaration content and improving input efficiency. Credential issuing users can view the source of all credential declaration content and data processing traces (e.g., user modification traces) and make quick decisions based on this information regarding credential issuance, thereby improving the efficiency of credential issuance review. This embodiment is applicable to scenarios such as digital work storage, digital collections, medical certificates, charitable application certificates, and government certificates.

[0097] The following introduces a credential generation device provided in an embodiment of the present application. The credential generation device described below can be referenced with other embodiments described in this document.

[0098] As shown in FIG8 , the embodiment of the present application discloses a credential generation device, including:

[0099] The generation module 801 is used to automatically generate corresponding summary information for the object to be authenticated uploaded by the applicant using a text generation model;

[0100] Filling module 802, used to fill the summary information into the target description template corresponding to the object to be authenticated to obtain a target description file;

[0101] The correction module 803 is used to correct the target description file according to the correction operation of the user at the application end to obtain the file to be authenticated; the file to be authenticated distinguishes and marks the automatically generated summary information and the information corrected by the user correction operation;

[0102] The sending module 804 is used to send the document to be authenticated and the object to be authenticated to the corresponding authentication end, and obtain the corresponding digital certificate issued by the authentication end after the authentication end reviews the document to be authenticated and the object to be authenticated.

[0103] In some embodiments, the application receiving module is used to receive a credential application request sent by the applicant before automatically generating corresponding summary information for the object to be authenticated uploaded by the applicant using a text generation model; return at least one description template to the applicant based on the credential application request so that the applicant displays at least one description template; at least one description template is uniformly set by the relevant issuing agency; obtain the user's selection operation on the target description template and the uploaded object to be authenticated on the applicant; the object to be authenticated is text, picture and / or video.

[0104] In some embodiments, the generation module is specifically configured to:

[0105] If the object to be authenticated is text, the text generation model is used to extract key information from the text, and summary information is obtained based on the key information.

[0106] In some embodiments, the generation module is specifically configured to:

[0107] The text is split into sentences using a text generation model, and a similarity matrix is ​​constructed based on the vector representation of each sentence to calculate the similarity between sentences. A similarity graph is constructed based on the similarity matrix with sentences as nodes and similarities as edges. Based on the similarity graph, the top N sentences with the highest similarity are selected as key information.

[0108] In some embodiments, the generation module is specifically configured to:

[0109] Use a text generation model to split the text into sentences and generate a word sequence for each sentence; determine the vector representation of each sentence based on the word sequence of each sentence; construct a document vector based on the vector representation of each sentence, and use a beam search algorithm to determine the summary information corresponding to the document vector; dynamically select key information from the summary information and each sentence.

[0110] In some embodiments, the generation module is specifically configured to:

[0111] If the object to be authenticated is text, the text generation model is used to perform natural language processing on the text to obtain summary information.

[0112] In some embodiments, the generation module is specifically configured to:

[0113] If the object to be authenticated is a picture and / or video, the text generation model is used to extract pixel features in the picture and / or video, and the corresponding text description information is matched to the pixel features, and summary information is obtained based on the text description information.

[0114] In some embodiments, the generation module is specifically configured to:

[0115] If the object to be authenticated is a video, a text generation model is used to extract video frames from the video to obtain a video frame sequence; corresponding text description information is matched for each video frame in the video frame sequence; all text description information is summarized to obtain a text file; key information is extracted from the text file, and summary information is obtained based on the key information.

[0116] In some embodiments, the generation module is specifically configured to:

[0117] The text generation model is used to divide the video into multiple video segments, and a single video frame is extracted from each video segment to obtain a video frame sequence.

[0118] In some embodiments, the generation module is specifically configured to:

[0119] Adjacent video frames in a video are determined as adjacent video frame groups using a text generation model. If the difference between two video frames in any adjacent video frame group is greater than a preset threshold, the next video frame in the current adjacent video frame group is retained. Otherwise, the next video frame in the current adjacent video frame group is deleted. All retained video frames are aggregated to obtain a video frame sequence.

[0120] In some embodiments, further comprising:

[0121] The initial information storage module is used to store the object to be authenticated, the target description template, the target description file, the file to be authenticated and the digital certificate in the blockchain system.

[0122] In some embodiments, further comprising:

[0123] The application record generation module is used to generate an application record including a digital certificate, a validity period of the digital certificate, a certificate number of the digital certificate, identification information of the authentication end, and identification information of the target description template.

[0124] In some embodiments, further comprising:

[0125] The application record storage module is used to store application records in the blockchain system.

[0126] In some embodiments, further comprising:

[0127] The query module is used to query the objects to be authenticated, target description templates, target description files, files to be authenticated, digital certificates and / or application records in the blockchain system.

[0128] In some embodiments, further comprising:

[0129] The failure prompt module is used to obtain the failure result sent by the authentication end after the authentication end fails to pass the review of the authentication file and the authentication object; and send the failure result to the application end so that the user of the application end can re-upload the authentication object and / or correct the authentication file;

[0130] The failed result storage module is used to store the failed results in the blockchain system.

[0131] In some embodiments, further comprising:

[0132] The verification module is used to receive verification requests for digital certificates; forward the verification requests to the applicant end so that the applicant end displays the verification requests; if the user of the applicant end obtains the authorization notification for the verification request, the verifiable information specified in the authorization notification is pushed to the initiator of the verification request so that the initiator end displays the verifiable information.

[0133] In some embodiments, further comprising:

[0134] The verification and authorization record storage module is used to generate verification and authorization records including digital credentials, verifiable information and identification information of the initiator; and store the verification and authorization records in the blockchain system.

[0135] In some embodiments, further comprising:

[0136] The verification and authorization record query module is used to query the verification and authorization records in the blockchain system.

[0137] Among them, for more specific working processes of each module and unit in this embodiment, reference can be made to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0138] It can be seen that this embodiment provides a credential generation device that can simplify user operations during the digital credential application and issuance process, and improve the efficiency and accuracy of credential issuance.

[0139] An electronic device provided in an embodiment of the present application is introduced below. The electronic device described below can be referenced with other embodiments described herein.

[0140] As shown in FIG9 , an embodiment of the present application discloses an electronic device, including:

[0141] Memory 901, for storing computer-readable instructions;

[0142] The processor 902 is configured to execute the computer-readable instructions to implement the method disclosed in any of the above embodiments.

[0143] Furthermore, an embodiment of the present application also provides an electronic device. The electronic device can be either a server 50 as shown in FIG10 or a terminal 60 as shown in FIG11 . FIG10 and FIG11 are both structural diagrams of electronic devices according to an exemplary embodiment, and the contents of the diagrams are not to be construed as limiting the scope of use of the present application.

[0144] Figure 10 is a schematic diagram of the structure of a server provided in an embodiment of the present application. The server 50 may include: at least one processor 51, at least one memory 52, a power supply 53, a communication interface 54, an input / output interface 55, and a communication bus 56. The memory 52 is used to store computer-readable instructions, which are loaded and executed by the processor 51 to implement the relevant steps of credential generation disclosed in any of the aforementioned embodiments.

[0145] In this embodiment, the power supply 53 is used to provide operating voltage for each hardware device on the server 50; the communication interface 54 can create a data transmission channel between the server 50 and external devices. The communication protocol it follows is any communication protocol that can be applied to the technical solution of this application and is not specifically limited here; the input and output interface 55 is used to obtain external input data or output data to the outside world. Its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0146] In addition, the memory 52, as a carrier for resource storage, can be a read-only memory, random access memory, disk or CD, etc. The resources stored thereon include an operating system 521, computer-readable instructions 522 and data 523, etc. The storage method can be temporary storage or permanent storage.

[0147] The operating system 521 is used to manage and control the hardware devices on the server 50 and the computer-readable instructions 522, enabling the processor 51 to operate and process data 523 in the memory 52. ​​The operating system 521 may be Windows Server, NetWare, Unix, Linux, or the like. In addition to including computer-readable instructions capable of performing the credential generation method disclosed in any of the aforementioned embodiments, the computer-readable instructions 522 may further include computer-readable instructions capable of performing other specific tasks. Data 523 may include data such as application update information and information about the application developer.

[0148] FIG11 is a schematic structural diagram of a terminal provided in an embodiment of the present application. The terminal 60 may specifically include but is not limited to a smart phone, a tablet computer, a laptop computer, or a desktop computer.

[0149] Generally, the terminal 60 in this embodiment includes a processor 61 and a memory 62 .

[0150] Among them, the processor 61 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 61 can be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 61 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 61 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 61 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0151] The memory 62 may include one or more computer-readable storage media, which may be non-transitory. The memory 62 may also include high-speed random access memory, and non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In this embodiment, the memory 62 is at least used to store the following computer-readable instructions 621, wherein, after the computer-readable instructions are loaded and executed by the processor 61, the relevant steps in the credential generation method executed by the terminal side disclosed in any of the aforementioned embodiments can be implemented. In addition, the resources stored in the memory 62 may also include an operating system 622 and data 623, etc., and the storage method may be temporary storage or permanent storage. Among them, the operating system 622 may include Windows, Unix, Linux, etc. The data 623 may include but is not limited to update information of the application.

[0152] In some embodiments, the terminal 60 may further include a display screen 63 , an input / output interface 64 , a communication interface 65 , a sensor 66 , a power supply 67 , and a communication bus 68 .

[0153] Those skilled in the art will appreciate that the structure shown in FIG. 11 does not limit the terminal 60 and may include more or fewer components than shown in the figure.

[0154] A non-volatile computer-readable storage medium provided in an embodiment of the present application is introduced below. The non-volatile computer-readable storage medium described below can be referenced with other embodiments described herein.

[0155] A non-volatile computer-readable storage medium for storing computer-readable instructions, wherein the computer-readable instructions, when executed by a processor, implement the credential generation method disclosed in the aforementioned embodiment. The computer-readable storage medium is a computer-readable storage medium that, as a carrier for storing resources, can be a read-only memory, random access memory, a magnetic disk, or an optical disk. The resources stored thereon include an operating system, computer-readable instructions, and data, and the storage method can be either temporary or permanent.

[0156] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0157] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of readable storage medium known in the art.

[0158] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.< / eos> < / sos>

Claims

1. A method for generating a voucher, characterized in that: include: Use the text generation model to automatically generate corresponding summary information for the object to be authenticated uploaded by the applicant; Fill the summary information into the target description template corresponding to the object to be authenticated to obtain a target description file; Modify the target description file according to the user modification operation of the application end to obtain a file to be authenticated; The document to be authenticated distinguishes and marks the automatically generated summary information and the information modified by the user modification operation; as well as The document to be authenticated and the object to be authenticated are sent to the corresponding authentication end, and the corresponding digital certificate issued by the authentication end after the authentication end reviews the document to be authenticated and the object to be authenticated.

2. The method according to claim 1, characterized in that Before automatically generating corresponding summary information for the object to be authenticated uploaded by the applicant by using the text generation model, the method further includes: Receiving a certificate application request sent by the applicant; Returning at least one description template to the applicant according to the credential application request, so that the applicant displays the at least one description template; and The selection operation of the target description template and the uploaded object requiring authentication by the user of the application end are obtained; the object requiring authentication is text, picture and / or video.

3. The method according to claim 2, characterized in that The text generation model automatically generates corresponding summary information for the object to be authenticated uploaded by the applicant, including: In response to the object to be authenticated being a first text, key information is extracted from the first text using the text generation model, and the summary information is obtained based on the key information.

4. The method according to claim 3, characterized in that The extracting key information from the first text using the text generation model includes: The first text is split into sentences using the text generation model, and a similarity matrix for calculating the similarity between sentences is constructed based on the vector representation of each sentence; a similarity graph is constructed based on the similarity matrix with sentences as nodes and similarities as edges; and the top N sentences ranked by similarity are selected as the key information based on the similarity graph.

5. The method according to claim 3, characterized in that The extracting key information from the first text using the text generation model includes: The text is split into sentences using the text generation model, and a word sequence is generated for each sentence; a vector representation of each sentence is determined based on the word sequence of each sentence; a document vector is constructed based on the vector representation of each sentence, and summary information corresponding to the document vector is determined using a beam search algorithm; and the key information is dynamically selected from the summary information and each sentence.

6. The method according to claim 2, characterized in that The text generation model automatically generates corresponding summary information for the object to be authenticated uploaded by the applicant, including: In response to the object to be authenticated being a second text, natural language processing is performed on the second text using the text generation model to obtain the summary information.

7. The method according to claim 1, characterized in that The text generation model automatically generates corresponding summary information for the object to be authenticated uploaded by the applicant, including: In response to the object to be authenticated being a picture and / or video, the text generation model is used to extract pixel features in the picture and / or video, and corresponding text description information is matched to the pixel features, and the summary information is obtained based on the text description information.

8. The method according to claim 7, characterized in that The text generation model automatically generates corresponding summary information for the object to be authenticated uploaded by the applicant, including: In response to the object to be authenticated being a video, the text generation model is used to extract video frames from the video to obtain a video frame sequence; corresponding text description information is matched for each video frame in the video frame sequence; all text description information is aggregated to obtain a text file; key information is extracted from the text file, and the summary information is obtained based on the key information.

9. The method according to claim 8, characterized in that The extracting video frames from the video using the text generation model to obtain a video frame sequence includes: The video is divided into a plurality of video segments using the text generation model, and a single video frame is extracted from each video segment to obtain the video frame sequence.

10. The method according to claim 8, characterized in that The extracting video frames from the video using the text generation model to obtain a video frame sequence includes: Determining adjacent video frames in the video as an adjacent video frame group using the text generation model; In response to a difference between two video frames in any adjacent video frame group being greater than a preset threshold, retaining a subsequent video frame in the current adjacent video frame group; and In response to a difference between two video frames in any adjacent video frame group being smaller than a preset threshold, a subsequent video frame in the current adjacent video frame group is deleted; and all retained video frames are aggregated to obtain the video frame sequence.

11. The credential generation method according to claim 1, wherein: Also includes: The object to be authenticated, the target description template, the target description file, the file to be authenticated, and the digital certificate are stored in a blockchain system.

12. The method according to claim 1, characterized in that Also includes: An application record is generated, which includes the digital certificate, the validity period of the digital certificate, the certificate number of the digital certificate, the identification information of the authentication end, and the identification information of the target description template.

13. The credential generation method according to claim 12, characterized in that: Also includes: The application record is stored in the blockchain system.

14. The method according to claim 12, characterized in that Also includes: The object to be authenticated, the target description template, the target description file, the file to be authenticated, the digital certificate and / or the application record are queried in the blockchain system.

15. The method according to any one of claims 1 to 14, characterized in that Also includes: Obtaining a failure result sent by the authentication end after the authentication end reviews the file to be authenticated and the object to be authenticated and fails to pass; Sending the failure result to the application end, so that the user of the application end can re-upload the object to be authenticated and / or correct the file to be authenticated; as well as The failed result is stored in the blockchain system.

16. The method according to any one of claims 1 to 14, characterized in that Also includes: Receiving a request for verification of the digital certificate; forwarding the verification request to the applicant so that the applicant displays the verification request; In response to obtaining the authorization notification of the verification request from the user of the application end, the verifiable information specified in the authorization notification is pushed to the initiator of the verification request, so that the initiator displays the verifiable information.

17. The method according to claim 16, characterized in that Also includes: Generating a verification and authorization record including the digital certificate, the verifiable information, and the identification information of the initiating end; as well as The verification and authorization records are stored in the blockchain system.

18. The method according to claim 17, characterized in that Also includes: The verification and authorization records are queried in the blockchain system.

19. A credential generating device, characterized in that: include: The generation module is used to automatically generate corresponding summary information for the object to be authenticated uploaded by the applicant using the text generation model; A filling module, configured to fill the summary information into a target description template corresponding to the object to be authenticated to obtain a target description file; A correction module, configured to correct the target description file according to the correction operation of the user at the application end to obtain a file to be authenticated; The document to be authenticated distinguishes and marks the automatically generated summary information and the information modified by the user modification operation; as well as The sending module is used to send the document to be authenticated and the object to be authenticated to the corresponding authentication end, and obtain the corresponding digital certificate issued by the authentication end after the authentication end reviews the document to be authenticated and the object to be authenticated.

20. An electronic device, characterized in that: include: a memory for storing computer-readable instructions; as well as A processor, configured to execute the computer-readable instructions to implement the method according to any one of claims 1 to 18.

21. A non-volatile computer-readable storage medium, characterized in that Used to store computer-readable instructions, wherein when the computer-readable instructions are executed by a processor, the method according to any one of claims 1 to 18 is implemented.

Images