Cross-platform data collaboration method and system based on intelligent API
By generating de-identified metadata in cross-platform data collaboration and using intelligent APIs to generate mapping logic with cloud-based large models, the problems of manual coding dependence and interaction delay in cross-platform data collaboration are solved, achieving efficient and secure data conversion and transmission.
Patent Information
- Application Number
- CN202610227805.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-02-26
- Publication Date
- 2026-05-15
AI Technical Summary
Existing technologies lack semantic understanding capabilities in cross-platform data collaboration, leading to reliance on manual hard coding for data cleaning and format conversion. This makes it impossible to automatically handle unknown heterogeneous data structures, and it is difficult to balance the contradiction between open data sharing and secure desensitization, resulting in high latency in interaction.
By collecting interface definition information and data structure characteristics from the source platform, de-identified metadata description text is generated. This text is then sent to a cloud-based large model using an intelligent API to generate cross-platform mapping logic. A deterministic execution script is loaded in the local environment to perform data transformation and generate the target data package. Combined with a script verification mechanism, the security and integrity of data transmission are ensured.
It significantly reduces development costs, improves cross-platform collaboration efficiency, enhances system expansion flexibility, ensures data security without requiring out-of-domain computation, effectively intercepts abnormal data, and improves data utilization and business flow efficiency.
Smart Images
Figure CN122053594A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a cross-platform data collaboration method and system based on intelligent APIs, belonging to the field of data integration and ETL technology. Background Technology
[0002] Cross-platform data collaboration based on intelligent APIs refers to breaking down the technical barriers between heterogeneous operating systems, databases, and application systems. Through standardized interfaces and intelligent conversion mechanisms, it enables seamless data flow, real-time sharing, and deep integration across different platforms. Its core significance lies in effectively eliminating "data silos," significantly improving the utilization rate of data assets and the efficiency of business processes, thereby laying a solid data foundation for enterprises to build a globally unified data view, empower cross-domain agile decision-making, and drive digital business innovation.
[0003] Traditional methods of cross-platform data collaboration mainly rely on manually written, hard-coded ETL scripts for data cleaning and format conversion. This approach lacks semantic understanding capabilities, cannot automatically handle unknown heterogeneous data structures, and struggles to balance the conflict between open data sharing and secure data anonymization, resulting in high latency issues in cross-platform interactions. Summary of the Invention
[0004] This invention provides a cross-platform data collaboration method and system based on intelligent APIs, the main purpose of which is to improve the efficiency of cross-platform data collaboration.
[0005] To achieve the above objectives, this invention provides a cross-platform data collaboration method based on an intelligent API, comprising:
[0006] The interface definition information and data structure characteristics are collected on the source platform side, and the de-identified metadata description text of the source platform is generated based on the interface definition information and data structure characteristics.
[0007] The metadata description text is sent to the cloud-based big model via the intelligent API, enabling the cloud-based big model to generate cross-platform mapping logic based on the metadata description text.
[0008] The mapping logic is encapsulated into a structured deterministic execution script;
[0009] The deterministic execution script is loaded in the local environment, and the sensitive business data of the source platform is transformed and processed according to the deterministic execution script to generate the target data packet;
[0010] After the target data packet passes verification, the target data packet is pushed to the target platform.
[0011] Optionally, based on the interface definition information and data structure characteristics, a de-identified metadata description text for the source platform is generated, including:
[0012] Based on the data structure characteristics, extract the key field attributes of the interface definition information and the logical relationships between the key field attributes;
[0013] The key field attributes are matched and identified with a pre-set sensitive feature library to obtain the sensitive type labels of the key field attributes;
[0014] In response to the sensitive type tag, a masking process is performed on the business example data in the target field attribute to obtain the de-sensitized placeholder of the business example data;
[0015] The key field attributes, logical relationships, and de-identified placeholders are assembled to obtain the de-identified metadata description text of the source platform.
[0016] Optionally, the construction of the sensitive feature library includes:
[0017] Collect privacy compliance documents and historical API documentation from multiple sources to build a basic corpus;
[0018] The named entity recognition model is used to perform semantic analysis on the basic corpus to extract candidate features with potential sensitive attributes from the basic corpus;
[0019] Determine the sensitivity type label corresponding to the candidate feature;
[0020] Candidate features carrying the aforementioned sensitive type labels are used as standard sensitive features, serialized and stored in the database to generate the sensitive feature library.
[0021] Optionally, in response to the sensitive type label, masking processing is performed on the business example data in the target field attribute to obtain the de-identified placeholder of the business example data, including:
[0022] The target masking strategy corresponding to the business example data is determined based on the sensitive type label. The target masking strategy includes data retention rules and a character replacement mapping table.
[0023] Based on the target masking strategy, the structural feature characters in the business example data are retained, and the remaining characters in the business example data other than the structural feature characters are replaced with preset mask symbols to generate the de-identified placeholders;
[0024] The structural feature characters include prefixes, suffixes, and separators in the data format, which are used to maintain the recognizability of the data structure semantics while masking the real content.
[0025] Optionally, the metadata description text can be sent to a large cloud model via a smart API, including:
[0026] The interface authentication module of the intelligent API is invoked to establish an encrypted transmission channel between the local client corresponding to the metadata description text and the cloud-based large model;
[0027] The metadata description text is serialized and encoded to obtain the encoded metadata description text;
[0028] The encoded metadata description text is then encapsulated into a model request message according to the predefined protocol format of the intelligent API.
[0029] The model request message is sent to the inference interface of the large cloud model through the encrypted transmission channel;
[0030] Listen to the first response data returned by the inference interface, and perform deserialization and parsing on the first response data to obtain the target code fragment generated by the cloud-based large model.
[0031] Optionally, the cloud-based large model generates cross-platform mapping logic based on the metadata description text, including:
[0032] Obtain the platform feature identifier of the target platform to be adapted;
[0033] Using the target code snippet as the input and the platform feature identifier as the adaptation constraint, a secondary inference request for the cloud-based large model is constructed.
[0034] Receive the second response data returned by the cloud-based large model based on the secondary inference request, and extract the transformed logic implementation code from the second response data;
[0035] The logic implementation code is encapsulated into the cross-platform mapping logic.
[0036] Optionally, the mapping logic is encapsulated as a structured deterministic execution script, including:
[0037] Parse the input parameters and output results of the mapping logic to generate a script metadata description header that defines the binding relationship between the input parameters and the output results;
[0038] Extract the core algorithm code of the mapping logic, and concatenate the core algorithm code with the script metadata description header to obtain the original script text;
[0039] The original script text is encrypted and signed to obtain the structured deterministic executable script.
[0040] Optionally, the sensitive business data of the source platform is transformed according to the deterministic execution script to generate a target data packet, including:
[0041] Construct an empty parameter object that matches the deterministic execution script;
[0042] The sensitive business data is preprocessed to obtain the source data stream to be converted;
[0043] The source data stream is mapped and filled into the empty parameter object to generate a set of parameters to be executed that conforms to the deterministic execution script calling specification;
[0044] In the local sandbox environment, the set of parameters to be executed is passed as input variables into the core algorithm code of the deterministic execution script, and the logical operation results returned after execution are captured.
[0045] The result of the logical operation is serialized and encapsulated to generate the target data packet.
[0046] Optionally, the target data packet verification process includes:
[0047] Calculate the data checksum of the target data packet;
[0048] The data verification code is compared with the original verification field encapsulated in the target data packet;
[0049] If the comparison matches, the target data packet is deemed to have passed verification; if the comparison does not match, the target data packet is deemed to have failed verification, the target data packet is discarded, and an error log is generated.
[0050] To address the aforementioned problems, this invention also provides a cross-platform data collaboration system based on intelligent APIs, the system comprising:
[0051] The descriptive text generation module is used to collect interface definition information and data structure characteristics on the source platform side, and generate desensitized metadata descriptive text for the source platform based on the interface definition information and data structure characteristics.
[0052] The mapping logic determination module is used to send the metadata description text to the cloud big model through the intelligent API, so that the cloud big model can generate cross-platform mapping logic based on the metadata description text.
[0053] An execution script encapsulation module is used to encapsulate the mapping logic into a structured deterministic execution script;
[0054] The data packet generation module is used to load the deterministic execution script in the local environment, and to transform the sensitive business data of the source platform according to the deterministic execution script to generate the target data packet;
[0055] The data packet push module is used to push the target data packet to the target platform after the target data packet passes the verification.
[0056] This invention collects source platform interface features and generates anonymized metadata. It then uses a cloud-based large-scale model to intelligently generate cross-platform mapping logic, automatically adapting to heterogeneous systems without manual intervention. This significantly reduces development costs and improves scalability. Simultaneously, the generated deterministic execution script transforms sensitive business data in a local sandbox environment, ensuring that calculations are completed without the source data leaving the domain, physically eliminating the risk of data leakage. Furthermore, by combining script verification and data packet verification mechanisms, it effectively intercepts abnormal data, guaranteeing the integrity of data transmission and the security of the system. Therefore, this invention can improve the efficiency of cross-platform data collaboration. Attached Figure Description
[0057] Figure 1 This is a flowchart illustrating a cross-platform data collaboration method based on an intelligent API, as provided in an embodiment of the present invention.
[0058] Figure 2 A functional block diagram of a cross-platform data collaboration system based on intelligent APIs provided in an embodiment of the present invention;
[0059] Figure 3 A schematic diagram of a computer device for a cross-platform data collaboration method based on an intelligent API, provided in an embodiment of the present invention;
[0060] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0061] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0062] This application provides a cross-platform data collaboration method based on a smart API. The executing entity of this cross-platform data collaboration method based on a smart API includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, this cross-platform data collaboration method based on a smart API can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.
[0063] Reference Figure 1 The diagram shown illustrates a flowchart of a cross-platform data collaboration method based on an intelligent API, according to an embodiment of the present invention. In this embodiment, the cross-platform data collaboration method based on an intelligent API includes:
[0064] S1. Collect interface definition information and data structure characteristics on the source platform side, and generate de-identified metadata description text of the source platform based on the interface definition information and data structure characteristics.
[0065] It should be explained that the source platform refers to the original business system, database server, or SaaS application that generates data and provides data service interfaces. The interface definition information is used to describe the technical specifications of the data interaction method, including but not limited to the URL address of the interface, request method (such as GET / POST), authentication method (such as API Key, OAuth token), call frequency limit, and transmission protocol type. The data structure features are used to define the metadata information of the data organization form and semantic attributes, including but not limited to the name of the data field, data type (such as integer, string, date), field length limit, required field constraint, hierarchical nesting relationship between fields, value range enumeration, and business semantic meaning of the field.
[0066] This invention generates desensitized metadata description text for the source platform based on the interface definition information and data structure features, providing a unified input basis for subsequent intelligent processing of source platform data, significantly reducing network transmission bandwidth consumption, and improving the response speed of subsequent model processing.
[0067] Specifically, the step of generating the desensitized metadata description text of the source platform based on the interface definition information and data structure characteristics includes:
[0068] Based on the data structure characteristics, extract the key field attributes of the interface definition information and the logical relationships between the key field attributes;
[0069] The key field attributes are matched and identified with a pre-set sensitive feature library to obtain the sensitive type labels of the key field attributes;
[0070] In response to the sensitive type tag, a masking process is performed on the business example data in the target field attribute to obtain the de-sensitized placeholder of the business example data;
[0071] The key field attributes, logical relationships, and de-identified placeholders are assembled to obtain the de-identified metadata description text of the source platform.
[0072] The key field attributes are a set of technical parameters describing the essential characteristics of a data field, including but not limited to the field name and data type. The logical relationships refer to information describing the dependency structure and constraint rules between different data fields, including but not limited to the nesting depth of parent and child fields, primary and foreign key reference constraints, and dependency conditions between fields. The sensitive feature library refers to a pre-built and maintained rule database for data security auditing. The sensitive type label refers to a marker assigned to the key field attributes regarding data privacy categories; this marker defines the sensitive data category to which the field belongs, such as, but not limited to, personal identification information, financial account information, and medical health information. The data includes Kang data, trade secrets, or general privacy data. The business example data refers to sample values used to demonstrate the format of field content. The anonymized placeholders refer to alternative data generated after algorithmic transformation of the business example data. The alternative data is consistent with the original business example data in terms of data format, length, and character type, but has covered the real numerical content. For example, the mobile phone number example "13800000000" is converted to "138XXXX0000" or the name example "Zhang San" is converted to "*San". The metadata description text refers to the document generated by serializing the key field attributes, logical relationships, and anonymized placeholders according to a preset standardized structure.
[0073] Furthermore, the construction of the sensitive feature library includes:
[0074] Collect privacy compliance documents and historical API documentation from multiple sources to build a basic corpus;
[0075] The named entity recognition model is used to perform semantic analysis on the basic corpus to extract candidate features with potential sensitive attributes from the basic corpus;
[0076] Determine the sensitivity type label corresponding to the candidate feature;
[0077] Candidate features carrying the aforementioned sensitive type labels are used as standard sensitive features, serialized and stored in the database to generate the sensitive feature library.
[0078] The multi-source privacy compliance documents refer to a collection of legal provisions, technical standards, and compliance guidance documents on data security and privacy protection issued by different industry associations or organizations, including but not limited to the Personal Information Protection Law and the General Data Protection Regulation. The historical interface documents refer to information such as interface design documents, API dictionaries, data dictionary version records, and Web service description language files generated and archived by the source platform in its past business operations. The basic corpus refers to a collection of text data formed after the multi-source privacy compliance documents and the historical interface documents have undergone structured cleaning, format unification, and deduplication processing. The named entity recognition model refers to a machine learning model specifically for text sequence labeling trained based on deep learning algorithms (such as BiLSTM-CRF, BERT, etc.). The candidate features refer to the original strings that may point to sensitive data and are initially screened and extracted by the named entity recognition model. The standard sensitive features refer to the final feature entries after sensitivity classification verification, deambiguation, and format normalization processing.
[0079] Further, the step of performing masking processing on the business example data in the target field attribute in response to the sensitive type tag to obtain the de-identified placeholder of the business example data includes:
[0080] The target masking strategy corresponding to the business example data is determined based on the sensitive type label. The target masking strategy includes data retention rules and a character replacement mapping table.
[0081] Based on the target masking strategy, the structural feature characters in the business example data are retained, and the remaining characters in the business example data other than the structural feature characters are replaced with preset mask symbols to generate the de-identified placeholders;
[0082] The structural feature characters include prefixes, suffixes, and separators in the data format, which are used to maintain the recognizability of the data structure semantics while masking the real content.
[0083] The data retention rules refer to the pre-defined data segment retention strategies for different sensitive type labels. These rules define the character range logic that needs to be extracted from the business sample data and preserved in its original form during the desensitization process, including but not limited to retaining the first N characters, the last M characters, specific separator symbols, and data format verification bits. The character replacement mapping table is a configuration set storing the conversion relationship between original characters and mask symbols, defining character replacement logic for different data types. For example, for pure numeric types, non-reserved bits are mapped to the characters "#" or "*". The structural feature characters refer to key character fragments in the business sample data that are not replaced during the data desensitization process. The pre-defined mask symbols refer to specific characters used to mask sensitive information content. The prefix refers to a fixed-length character sequence at the beginning of the data string, such as the operator code in a mobile phone number or the region code in an ID card number. The suffix refers to a fixed-length character sequence at the end of the data string, such as the sequence code or check code in an ID card number or the last digit of the card type identifier in a bank card number. The separator is a specific symbol used to connect different data segments, such as "-" or " / " in a date.
[0084] S2. Send the metadata description text to the cloud big model through the smart API, so that the cloud big model can generate cross-platform mapping logic based on the metadata description text.
[0085] This invention sends the metadata description text to a cloud-based large model via an intelligent API, enabling the cloud-based large model to generate cross-platform mapping logic based on the metadata description text. Compared to traditional technologies that rely on manually written code, this method can autonomously identify and parse complex data structures and business semantic differences between heterogeneous platforms, generating accurate conversion logic without manual intervention. This significantly reduces the human resource costs of system development and maintenance, while greatly improving the system's adaptability and flexibility to unknown platforms or new interfaces.
[0086] Specifically, sending the metadata description text to the cloud-based large model via the intelligent API includes:
[0087] The interface authentication module of the intelligent API is invoked to establish an encrypted transmission channel between the local client corresponding to the metadata description text and the cloud-based large model;
[0088] The metadata description text is serialized and encoded to obtain the encoded metadata description text;
[0089] The encoded metadata description text is then encapsulated into a model request message according to the predefined protocol format of the intelligent API.
[0090] The model request message is sent to the inference interface of the large cloud model through the encrypted transmission channel;
[0091] Listen to the first response data returned by the inference interface, and perform deserialization and parsing on the first response data to obtain the target code fragment generated by the cloud-based large model.
[0092] The interface authentication module refers to the logical unit integrated in the smart API client for handling identity verification and security authentication. The encrypted transmission channel refers to the virtual link established between the local client and the cloud-based large model server based on the secure transport layer protocol. The encoded metadata description text refers to the binary data stream after serialization. The protocol format refers to the data interaction specifications and structural standards defined by the smart API. It defines the specific field layout of the model request message, including the required Header and Body parameter structures. The model request message refers to the complete data packet conforming to the protocol format. The cloud-based large model refers to the deep learning language model deployed on the cloud server cluster. The inference interface refers to the service endpoint exposed by the cloud-based large model for receiving inference requests submitted by the client. The first response data refers to the data sequence returned continuously and in real time by the cloud-based large model during the generation of the target code fragment. The target code fragment refers to the source code text generated by the cloud-based large model based on the business logic and data structure features in the metadata description text.
[0093] Optionally, the monitoring of the first response data returned by the inference interface is implemented using streaming technology.
[0094] Specifically, the logic for enabling the cloud-based large model to generate a cross-platform mapping based on the metadata description text includes:
[0095] Obtain the platform feature identifier of the target platform to be adapted;
[0096] Using the target code snippet as the input and the platform feature identifier as the adaptation constraint, a secondary inference request for the cloud-based large model is constructed.
[0097] Receive the second response data returned by the cloud-based large model based on the secondary inference request, and extract the transformed logic implementation code from the second response data;
[0098] The logic implementation code is encapsulated into the cross-platform mapping logic.
[0099] Wherein, the platform feature identifier refers to a set of technical parameters used to uniquely determine the target operating environment; the adaptation constraint refers to the logical transformation boundary generated based on the platform feature identifier; the secondary inference request refers to a task instruction initiated to the cloud-based large model again after obtaining the initial target code fragment; the second response data refers to the data packet returned by the cloud-based large model after executing the secondary inference request; the logical implementation code refers to the source code text that can be directly executed by a computer extracted from the second response data; and the mapping logic refers to the functional logical unit formed by encapsulating the logical implementation code.
[0100] Optionally, the encapsulation of the logic implementation code into the cross-platform mapping logic can be obtained by a runtime compiler in real time.
[0101] S3. Encapsulate the mapping logic into a structured deterministic execution script.
[0102] This invention encapsulates the mapping logic into a structured deterministic execution script, eliminating the nondeterminism caused by directly outputting results from a large model, which facilitates subsequent debugging, reuse, and canary release.
[0103] Specifically, encapsulating the mapping logic into a structured deterministic execution script includes:
[0104] Parse the input parameters and output results of the mapping logic to generate a script metadata description header that defines the binding relationship between the input parameters and the output results;
[0105] Extract the core algorithm code of the mapping logic, and concatenate the core algorithm code with the script metadata description header to obtain the original script text;
[0106] The original script text is encrypted and signed to obtain the structured deterministic executable script.
[0107] The input parameters include external input data required to execute the mapping logic, specifically defining the data name, data type, value range, and reference identifier in the execution script. The output result refers to the target data object generated after processing by the mapping logic. The script metadata description header refers to the standardized configuration information block located at the top of the script file. The core algorithm code refers to the executable source code fragment that implements specific data transformation and business logic processing. The original script text refers to the unencrypted string sequence formed by sequentially concatenating the script metadata description header and the core algorithm code. The deterministic execution script refers to the final script entity after encryption and signature processing.
[0108] Optionally, the process of encrypting and signing the original script text to obtain the structured deterministic execution script can be implemented using a hash digest algorithm.
[0109] S4. Load the deterministic execution script in the local environment, and perform transformation processing on the sensitive business data of the source platform according to the deterministic execution script to generate the target data packet.
[0110] This invention loads the deterministic execution script in a local environment and transforms sensitive business data from the source platform based on the script to generate a target data packet. By loading the script and performing the transformation in the local environment, the sensitive business data from the source platform does not need to be uploaded to the cloud or other external servers for computation, physically preventing the leakage of sensitive data during network transmission. It should be explained that loading the deterministic execution script in the local environment refers to injecting a structured script file distributed from the cloud into the runtime container of the local client or edge device, and performing integrity verification and initialization of the script.
[0111] In detail, the step of transforming the sensitive business data of the source platform according to the deterministic execution script to generate the target data packet includes:
[0112] Construct an empty parameter object that matches the deterministic execution script;
[0113] The sensitive business data is preprocessed to obtain the source data stream to be converted;
[0114] The source data stream is mapped and filled into the empty parameter object to generate a set of parameters to be executed that conforms to the deterministic execution script calling specification;
[0115] In the local sandbox environment, the set of parameters to be executed is passed as input variables into the core algorithm code of the deterministic execution script, and the logical operation results returned after execution are captured.
[0116] The result of the logical operation is serialized and encapsulated to generate the target data packet.
[0117] Wherein, the empty parameter object refers to a data structure carrier instantiated in memory according to the interface specifications defined in the deterministic execution script; the source data stream to be converted refers to the set of original sensitive business data extracted from the database of the source platform; the set of parameters to be executed refers to a stateful data object that fully conforms to the script execution input parameter requirements after the source data stream to be converted is filled into the empty parameter object according to the field mapping relationship; the core algorithm code refers to the instruction sequence part in the deterministic execution script that implements the specific business logic conversion; the logical operation result refers to the original return value output after the core algorithm code is executed in the local sandbox environment; and the target data packet refers to the text data stream generated after the logical operation result is serialized and encoded according to the communication protocol standard of the target platform.
[0118] It should be further clarified that the local sandbox environment refers to:
[0119] An isolated process created in user space of the source platform's local operating system;
[0120] The isolated process has an independent memory address space and interacts with the host process and other applications through inter-process communication mechanisms. When the deterministic execution script finishes execution or an exception occurs, the isolated process is forcibly terminated and all memory resources it occupies are released.
[0121] S5. After the target data packet passes the verification, the target data packet is pushed to the target platform.
[0122] After the target data packet passes verification, the present invention pushes the target data packet to the target platform, which can effectively intercept abnormal data caused by misunderstanding of large models or script logic errors.
[0123] In detail, the target data packet verification process includes:
[0124] Calculate the data checksum of the target data packet;
[0125] The data verification code is compared with the original verification field encapsulated in the target data packet;
[0126] If the comparison matches, the target data packet is deemed to have passed verification; if the comparison does not match, the target data packet is deemed to have failed verification, the target data packet is discarded, and an error log is generated.
[0127] The data checksum is used as a verification benchmark for the integrity of the current data packet before transmission or after reception. The original checksum field refers to the checksum data pre-calculated by the sending end and encapsulated in a specific protocol header or trailer area of the data packet when the target data packet is generated. The error log refers to the structured fault record automatically generated by the system when the data checksum does not match the original checksum field; it includes the timestamp of the checksum failure, the unique identifier of the failed data packet, the specific error type, and the exception stack information.
[0128] Optionally, the data checksum of the target data packet can be calculated using a cyclic redundancy check algorithm.
[0129] Specifically, pushing the target data packet to the target platform can be achieved through a RESTful API interface.
[0130] like Figure 2 The diagram shown is a functional block diagram of a cross-platform data collaboration system based on intelligent APIs according to the present invention.
[0131] The cross-platform data collaboration system 200 based on intelligent APIs described in this invention can be installed in electronic devices. Depending on the functions implemented, the cross-platform data collaboration system based on intelligent APIs includes a description text generation module 201, a mapping logic determination module 202, an execution script encapsulation module 203, a data packet generation module 204, and a data packet push module 205. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.
[0132] In this embodiment of the invention, the functions of each module / unit are as follows:
[0133] The description text generation module 201 is used to collect interface definition information and data structure features on the source platform side, and generate desensitized metadata description text of the source platform based on the interface definition information and data structure features.
[0134] The mapping logic determination module 202 is used to send the metadata description text to the cloud big model through the smart API, so that the cloud big model generates cross-platform mapping logic based on the metadata description text.
[0135] The execution script encapsulation module 203 is used to encapsulate the mapping logic into a structured deterministic execution script;
[0136] The data packet generation module 204 is used to load the deterministic execution script in the local environment, and to perform transformation processing on the sensitive business data of the source platform according to the deterministic execution script to generate the target data packet;
[0137] The data packet push module 205 is used to push the target data packet to the target platform after the target data packet passes the verification.
[0138] In detail, the modules in the cross-platform data collaboration system 200 based on intelligent API described in this embodiment of the invention adopt the same approach as described above when in use. Figure 1 This method employs the same technical means as the cross-platform data collaboration method based on intelligent APIs described above, and can produce the same technical effects, so it will not be elaborated here.
[0139] In one embodiment, a computer device is provided, which may be a server or a client, and its internal structure diagram may be as follows: Figure 3 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements functions or steps on the server or client side of a cross-platform data collaboration method based on an intelligent API.
[0140] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps:
[0141] The descriptive text generation module is used to collect interface definition information and data structure characteristics on the source platform side, and generate desensitized metadata descriptive text for the source platform based on the interface definition information and data structure characteristics.
[0142] The mapping logic determination module is used to send the metadata description text to the cloud big model through the intelligent API, so that the cloud big model can generate cross-platform mapping logic based on the metadata description text.
[0143] An execution script encapsulation module is used to encapsulate the mapping logic into a structured deterministic execution script;
[0144] The data packet generation module is used to load the deterministic execution script in the local environment, and to transform the sensitive business data of the source platform according to the deterministic execution script to generate the target data packet;
[0145] The data packet push module is used to push the target data packet to the target platform after the target data packet passes the verification.
[0146] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:
[0147] The descriptive text generation module is used to collect interface definition information and data structure characteristics on the source platform side, and generate desensitized metadata descriptive text for the source platform based on the interface definition information and data structure characteristics.
[0148] The mapping logic determination module is used to send the metadata description text to the cloud big model through the intelligent API, so that the cloud big model can generate cross-platform mapping logic based on the metadata description text.
[0149] An execution script encapsulation module is used to encapsulate the mapping logic into a structured deterministic execution script;
[0150] The data packet generation module is used to load the deterministic execution script in the local environment, and to transform the sensitive business data of the source platform according to the deterministic execution script to generate the target data packet;
[0151] The data packet push module is used to push the target data packet to the target platform after the target data packet passes the verification.
[0152] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.
[0153] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0154] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.
[0155] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0156] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims
1. A cross-platform data collaboration method based on intelligent APIs, characterized in that, The method includes: The interface definition information and data structure characteristics are collected on the source platform side, and the de-identified metadata description text of the source platform is generated based on the interface definition information and data structure characteristics. The metadata description text is sent to the cloud-based big model via the intelligent API, enabling the cloud-based big model to generate cross-platform mapping logic based on the metadata description text. The mapping logic is encapsulated into a structured deterministic execution script; The deterministic execution script is loaded in the local environment, and the sensitive business data of the source platform is transformed and processed according to the deterministic execution script to generate the target data packet; After the target data packet passes verification, the target data packet is pushed to the target platform.
2. The cross-platform data collaboration method based on intelligent API as described in claim 1, characterized in that, Based on the interface definition information and data structure characteristics, a de-identified metadata description text for the source platform is generated, including: Based on the data structure characteristics, extract the key field attributes of the interface definition information and the logical relationships between the key field attributes; The key field attributes are matched and identified with a pre-set sensitive feature library to obtain the sensitive type labels of the key field attributes; In response to the sensitive type tag, a masking process is performed on the business example data in the target field attribute to obtain the de-sensitized placeholder of the business example data; The key field attributes, logical relationships, and de-identified placeholders are assembled to obtain the de-identified metadata description text of the source platform.
3. The cross-platform data collaboration method based on intelligent API as described in claim 2, characterized in that, The construction of the sensitive feature library includes: Collect privacy compliance documents and historical API documentation from multiple sources to build a basic corpus; The named entity recognition model is used to perform semantic analysis on the basic corpus to extract candidate features with potential sensitive attributes from the basic corpus; Determine the sensitivity type label corresponding to the candidate feature; Candidate features carrying the aforementioned sensitive type labels are used as standard sensitive features, serialized and stored in the database to generate the sensitive feature library.
4. The cross-platform data collaboration method based on intelligent API as described in claim 2, characterized in that, In response to the sensitive type tag, the business example data in the target field attribute is masked to obtain the de-sensitized placeholders for the business example data, including: The target masking strategy corresponding to the business example data is determined based on the sensitive type label. The target masking strategy includes data retention rules and a character replacement mapping table. Based on the target masking strategy, the structural feature characters in the business example data are retained, and the remaining characters in the business example data other than the structural feature characters are replaced with preset mask symbols to generate the de-identified placeholders; The structural feature characters include prefixes, suffixes, and separators in the data format, which are used to maintain the recognizability of the data structure semantics while masking the real content.
5. The cross-platform data collaboration method based on intelligent API as described in claim 1, characterized in that, Sending the metadata description text to the cloud-based large model via the intelligent API includes: The interface authentication module of the intelligent API is invoked to establish an encrypted transmission channel between the local client corresponding to the metadata description text and the cloud-based large model; The metadata description text is serialized and encoded to obtain the encoded metadata description text; The encoded metadata description text is then encapsulated to generate a model request message according to the predefined protocol format of the intelligent API. The model request message is sent to the inference interface of the large cloud model through the encrypted transmission channel; Listen to the first response data returned by the inference interface, and perform deserialization and parsing on the first response data to obtain the target code fragment generated by the cloud-based large model.
6. The cross-platform data collaboration method based on intelligent API as described in claim 1, characterized in that, The logic for generating cross-platform mappings based on the metadata description text in the cloud-based large model includes: Obtain the platform feature identifier of the target platform to be adapted; Using the target code snippet as the input and the platform feature identifier as the adaptation constraint, a secondary inference request for the cloud-based large model is constructed. Receive the second response data returned by the cloud-based large model based on the secondary inference request, and extract the transformed logic implementation code from the second response data; The logic implementation code is encapsulated into the cross-platform mapping logic.
7. The cross-platform data collaboration method based on intelligent API as described in claim 1, characterized in that, The mapping logic is encapsulated into a structured deterministic execution script, including: Parse the input parameters and output results of the mapping logic to generate a script metadata description header that defines the binding relationship between the input parameters and the output results; Extract the core algorithm code of the mapping logic, and concatenate the core algorithm code with the script metadata description header to obtain the original script text; The original script text is encrypted and signed to obtain the structured deterministic executable script.
8. The cross-platform data collaboration method based on intelligent API as described in claim 1, characterized in that, The deterministic execution script is used to transform sensitive business data from the source platform to generate a target data packet, including: Construct an empty parameter object that matches the deterministic execution script; The sensitive business data is preprocessed to obtain the source data stream to be converted; The source data stream is mapped and filled into the empty parameter object to generate a set of parameters to be executed that conforms to the deterministic execution script calling specification; In the local sandbox environment, the set of parameters to be executed is passed as input variables into the core algorithm code of the deterministic execution script, and the logical operation results returned after execution are captured. The result of the logical operation is serialized and encapsulated to generate the target data packet.
9. A cross-platform data collaboration method based on intelligent APIs as described in claim 1, characterized in that, The process of verifying the target data packet includes: Calculate the data checksum of the target data packet; The data verification code is compared with the original verification field encapsulated in the target data packet; If the comparison matches, the target data packet is deemed to have passed verification; if the comparison does not match, the target data packet is deemed to have failed verification, the target data packet is discarded, and an error log is generated.
10. A cross-platform data collaboration system based on intelligent APIs, characterized in that, The system is used to execute a cross-platform data collaboration method based on a smart API as described in any one of claims 1-9, the system comprising: The descriptive text generation module is used to collect interface definition information and data structure characteristics on the source platform side, and generate desensitized metadata descriptive text for the source platform based on the interface definition information and data structure characteristics. The mapping logic determination module is used to send the metadata description text to the cloud big model through the intelligent API, so that the cloud big model can generate cross-platform mapping logic based on the metadata description text. An execution script encapsulation module is used to encapsulate the mapping logic into a structured deterministic execution script; The data packet generation module is used to load the deterministic execution script in the local environment, and to transform the sensitive business data of the source platform according to the deterministic execution script to generate the target data packet; The data packet push module is used to push the target data packet to the target platform after the target data packet passes the verification.