Task Processing Method and Device for a Privacy Computing Platform Based on Algorithm Interconnection

Abstract

The present application provides a task processing method and apparatus for a privacy computing platform based on algorithm interconnection. The method includes: invoking a pre-configured target algorithm process according to a privacy computing task; the target algorithm process is obtained by orchestrating a plurality of target algorithm components; the plurality of target algorithm components are formed according to a pre-agreed algorithm concept model and attribute fields; sending the target algorithm process to a second privacy computing node in the network to cooperate with the second privacy computing node to complete the privacy computing task. By orchestrating the target algorithm components formed according to the pre-agreed algorithm concept model and attribute fields, since the algorithm concept model and attribute fields are jointly agreed upon by the first privacy computing node and the second privacy computing node, the target algorithm process is applicable to the first privacy computing node and the second privacy computing node, and can enable multiple privacy computing nodes to cooperate through a series of standard interaction interfaces to complete a unified privacy computing task.

Description

Technical Field

[0001] This application relates to the field of big data privacy computing technology. Specifically, it relates to a task processing method and device for a privacy computing platform based on algorithm interconnection. Background Art

[0002] With the continuous expansion of the privacy computing market scale and the entry of many manufacturers in various fields, multiple privacy computing manufacturers have each designed different privacy computing algorithms and developed different privacy computing platforms. For the convenience of description, this application refers to privacy computing platforms from different privacy computing manufacturers or different versions of the same manufacturer as heterogeneous privacy computing platforms.

[0003] Due to the complete differences in reference standards, technical implementation methods, algorithm designs, etc. among different manufacturers, different developers, or different versions, in a business scenario, each participating party needs to deploy and use a privacy computing platform of the same privacy computing manufacturer and the same version, otherwise the parties will not be able to cooperate to complete the same privacy computing task. Summary of the Invention

[0004] The purpose of the embodiments of this application is to provide a task processing method and device for a privacy computing platform based on algorithm interconnection, so as to achieve algorithm interconnection of heterogeneous privacy computing platforms and complete the processing of privacy computing tasks.

[0005] In a first aspect, the embodiments of this application provide a task processing method for a privacy computing platform based on algorithm interconnection, including: calling a pre-configured target algorithm process according to a privacy computing task; where the target algorithm process is obtained by orchestrating a plurality of target algorithm components; the plurality of target algorithm components are formed according to a pre-agreed algorithm concept model and attribute fields; sending the target algorithm process to a second privacy computing node in the network to cooperate with the second privacy computing node to complete the privacy computing task.

[0006] The embodiments of this application orchestrate target algorithm components formed according to a pre-agreed algorithm concept model and attribute fields to obtain a target algorithm process, and send the target algorithm process to a second privacy computing node to cooperate in processing a privacy computing task. Since the algorithm concept model and attribute fields are jointly agreed upon by the first privacy computing node and the second privacy computing node, the target algorithm process is applicable to the first privacy computing node and the second privacy computing node, and enables multiple privacy computing nodes to cooperate to complete a unified privacy computing task through a series of standard interaction interfaces.

[0007] In any embodiment, invoking multiple pre-configured target algorithm components according to a privacy computing task and orchestrating the multiple target algorithm components to obtain a target algorithm process includes: reading an algorithm component list; the algorithm component list includes algorithm components local to the first privacy computing node and security-certified algorithm components published and authorized by the second privacy computing node; selecting multiple target algorithm components from the algorithm component list and orchestrating the multiple target algorithm components according to the input-output order of each target algorithm component to obtain the target algorithm process. In the embodiments of the present application, since the algorithm components in the algorithm component list are formed by each privacy computing node according to a pre-agreed algorithm concept model and attribute fields, therefore, selecting target algorithm components from the algorithm component list for orchestration, the obtained target algorithm process can run in each privacy computing node, and data can be exchanged between each privacy computing node.

[0008] In any embodiment, the method further includes: receiving a pre-agreed algorithm concept model and attribute fields, and forming algorithm components according to the algorithm concept model and the attribute fields; wherein, the algorithm concept model includes an algorithm component object, a component task object, an algorithm process object, and a process task object; the attribute fields include description fields respectively corresponding to the algorithm component object, the component task object, the algorithm process object, and the process task object. In the embodiments of the present application, by uniformly defining the algorithm concept model and attribute fields with each privacy computing node, compatibility is achieved when different privacy computing nodes develop privacy algorithms.

[0009] In any embodiment, the method further includes: receiving an algorithm management request and performing corresponding operations according to the algorithm management request; wherein, the algorithm management request includes an algorithm component management request, a component task management request, an algorithm process management request, and a process task management request; the algorithm component management request includes a publishing request, a taking-offline request, an update request, a deletion request, and an authorization request for an algorithm component; the component task management request includes a list query request, a start request, a disabling request, a parameter modification request, an input query request, and an output query request for a component task; the algorithm process management request includes a publishing request, a taking-offline request, an update request, a deletion request, an authorization request, and a component rearrangement request for an algorithm process; the process task management request includes a list query request, an execution request, a pause request, a parameter modification request, a re-execution request, an input query request, and an output query request for a process task. In the embodiments of the present application, by designing the core interaction process and functional modules of algorithm collaboration, multiple privacy computing nodes can complete the same privacy computing task through a series of standard interaction interfaces systems.

[0010] In any embodiment, after forming the algorithm component, the method further includes: sending a release request of the algorithm component to the second privacy computing node, where the release request includes authentication information and an attribute field of the algorithm component, so that the second privacy computing node authenticates the algorithm component according to the authentication information, and after the authentication is passed, adds the algorithm component to the local algorithm component list. By releasing the developed algorithm component to the second privacy computing node in the embodiments of the present application, the second privacy computing node can include the algorithm component in the local algorithm component list for later use after verifying the security of the algorithm component.

[0011] In any embodiment, the cooperation with the second privacy computing node to complete the privacy computing task includes: the first privacy computing node and the second privacy computing node respectively configure algorithm parameters corresponding to the received target algorithm process, and run the configured target algorithm process to cooperate to complete the privacy computing task. Different privacy computing nodes will configure different algorithm parameters due to different input data or actual requirements. By running algorithm processes configured with different algorithm parameters, multiple component task and process task running instances can be obtained. These instances support the secure exchange of encrypted intermediate data between privacy computing nodes during the processing of privacy algorithm tasks, so as to complete the processing of privacy computing tasks.

[0012] In any embodiment, running the configured target algorithm process includes: if the first privacy computing node needs to use external resources of the second privacy computing node during the process of running the configured target algorithm process, sending a resource acquisition request to the second privacy computing node, so that when the second privacy computing node determines that the external resources are in an authorized state according to the resource acquisition request, it returns the external resources to the first privacy computing node. By authorizing some data by the privacy computing node in the embodiments of the present application, other privacy computing nodes can use the authorized data.

[0013] In any embodiment, the method further includes: obtaining log information generated during the execution of the privacy computing task; encrypting and storing the log information. This is convenient for discovering and tracking privacy computing nodes that violate the agreement.

[0014] In a second aspect, an embodiment of the present application provides a task processing device for a privacy computing platform based on algorithm interconnection, including: an algorithm component orchestration module for invoking a pre-configured target algorithm process according to a privacy computing task; wherein, the target algorithm process is obtained by orchestrating a plurality of target algorithm components; the plurality of target algorithm components are formed by the privacy computing node according to a pre-agreed algorithm concept model and attribute fields; a collaborative processing module for sending the target algorithm process to a second privacy computing node in the network to collaborate with the second privacy computing node to complete the privacy computing task.

[0015] In a third aspect, an embodiment of the present application provides an electronic device, including: a processor, a memory, and a bus, wherein the processor and the memory communicate with each other through the bus; the memory stores program instructions executable by the processor, and the processor can execute the method of the first aspect by invoking the program instructions.

[0016] In a fourth aspect, an embodiment of the present application provides a non-transitory computer-readable storage medium, including: the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the method of the first aspect.

[0017] Other features and advantages of the present application will be described in the subsequent specification, and part of them will become obvious from the specification, or can be understood by implementing the embodiments of the present application. The objectives and other advantages of the present application can be achieved and obtained through the structures specifically pointed out in the written specification, claims, and drawings. Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application, so they should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0019] Figure 1 It is a schematic diagram of the concept model provided by the embodiment of the present application;

[0020] Figure 2 It is a schematic diagram of the process flow of a task processing method for a privacy computing platform based on algorithm interconnection provided by the embodiment of the present application;

[0021] Figure 3 It is a schematic diagram of the structure of a task processing device for a privacy computing platform based on algorithm interconnection provided by the embodiment of the present application;

[0022] Figure 4Schematic diagram of the physical structure of the electronic device provided by the embodiment of the present application. Detailed implementation manners

[0023] Developers have found that the current privacy computing industry lacks a unified conceptual model and interaction process for core privacy computing algorithms such as secure statistics, joint feature engineering, joint modeling, and joint prediction in heterogeneous privacy computing platforms. At the same time, a general description dimension and description scope of privacy computing algorithms have not been summarized, so it is difficult to solve the problem that heterogeneous privacy computing platforms of different manufacturers, different developers, and different versions cannot cooperate to complete the same privacy computing task.

[0024] In actual business, it will bring great inconvenience to users. For example, a large business group purchases and integrates privacy computing platforms of multiple manufacturers, but it is difficult to realize the cooperation of multiple platforms to carry out privacy computing tasks. Another example is that due to upgrades and iterations in aspects such as architecture, algorithms, and communication of a certain manufacturer, the two versions of privacy computing platforms deployed successively cannot be compatible with each other.

[0025] To solve the above problems, the embodiment of the present application provides a task processing method for a privacy computing platform based on algorithm interconnection. This method uses a pre-agreed algorithm conceptual model and attribute fields to form a target algorithm component, arranges the target algorithm component to obtain a target algorithm process, and sends the target algorithm process to a second privacy computing node that cooperates to process privacy computing tasks. Since the target algorithm component is pre-agreed by multiple privacy nodes, the target algorithm process can be used by multiple privacy nodes, thus realizing the interconnection and interoperability of multiple privacy nodes.

[0026] Next, the technical solutions in the embodiments of the present application will be described in conjunction with the accompanying drawings in the embodiments of the present application.

[0027] Before forming the target algorithm component, multiple privacy computing nodes in the network negotiate and define the conceptual model and attribute fields of the privacy computing algorithm. Among them, the conceptual model includes algorithm component objects, component task objects, algorithm process objects, and process task objects. The above-mentioned multiple privacy computing nodes run platforms belonging to different privacy computing manufacturers.

[0028] Among them, the algorithm component and the algorithm process are static objects before the algorithm runs. The algorithm component is the code implementation of a single privacy computing algorithm, and usually a single algorithm component cannot run independently; multiple algorithm components are arranged to form an algorithm process. For example, a data preprocessing component, a feature engineering component, a logistic regression modeling component, a model evaluation component, and a logistic regression prediction component are arranged to form a risk control algorithm process.

[0029] Component tasks and process tasks are dynamic objects after the algorithm runs. A component task is an execution instance of a single algorithm component; a process task is an execution instance of an algorithm process. An algorithm process can be run multiple times with different parameter configurations to generate multiple different process tasks. Figure 1 It is a schematic diagram of the conceptual model provided by the embodiment of the present application, as Figure 1 shown.

[0030] The attribute fields include the description field of the algorithm component, the description field of the component task, the description field of the algorithm process, and the description field of the process task.

[0031] Table 1 shows the description field of the algorithm component provided by the embodiment of the present application, as specifically shown in Table 1:

[0032] Table 1

[0033]

[0034]

[0035]

[0036] Table 2 shows the description field of the component task provided by the embodiment of the present application, as shown in Table 2:

[0037] Table 2

[0038]

[0039]

[0040] Table 3 shows the description field of the algorithm process provided by the embodiment of the present application, as shown in Table 3:

[0041] Table 3

[0042]

[0043]

[0044] Table 4 shows the description field of the process task provided by the embodiment of the present application, as shown in Table 4:

[0045] Table 4

[0046]

[0047]

[0048] It should be noted that the specific fields in Table 1 - Table 4 above are only examples. In actual implementation, more or fewer fields than those in Table 1 - Table 4 can be pre - agreed, and the attributes, names, and specific contents of the descriptions of the fields can be adjusted according to the actual situation. The embodiments of this application do not make specific limitations on this.

[0049] The embodiments of this application have uniformly defined the algorithm concept model and attribute fields with each privacy - computing node, enabling multiple privacy - computing nodes to have the same understanding of algorithms (what objects are divided into), descriptions (which fields are included), code implementations (how to define the attributes of class in code), etc., providing a feasible interface design idea and technical implementation solution for node connection, resource sharing, and task collaboration among heterogeneous privacy - computing platforms.

[0050] In addition, multiple privacy - computing nodes also pre - agree on the content related to algorithm management. Among them, algorithm management refers to the basic ability to maintain the algorithm components and algorithm processes for normal use and ensure the normal execution of component tasks and process tasks. Specifically, it includes:

[0051] 1) Algorithm component and component task management: including the release, offline, update, deletion, authorization application, etc. of algorithm components; the list query, component task start, disable, parameter modification, input / output query, etc. of component tasks.

[0052] 2) Algorithm process and process task management: including the release, offline, update, deletion, authorization application, component rearrangement, etc. of algorithm processes; the list query, execution, pause, parameter modification, re - execution, input / output query, etc. of process tasks.

[0053] 3) Algorithm information management: including the addition, deletion, modification, query, etc. of description field information of components, component tasks, algorithm processes, and process tasks.

[0054] 4) Algorithm list management: Manage and maintain the algorithm component list and algorithm process list after authentication and security review. Among them, algorithm component list management supports functions such as authorization application for externally released algorithm components, filtering components by conditions, viewing component information, etc.; for self - owned algorithm components, it supports functions such as authorization review, filtering components by conditions, viewing and modifying component information, adding components, deleting components, configuration modification, component orchestration, etc. Algorithm process list management supports functions such as cooperation application for externally released algorithm processes, filtering processes by conditions, viewing process information, etc.; for self - owned algorithm processes, it supports functions such as cooperation review, filtering processes by conditions, viewing and modifying process information, adding new processes, deleting processes, configuration modification, process component rearrangement, etc.

[0055] Based on algorithm component management, users can submit algorithm component management requests through privacy computing nodes. For example, they can publish a certain algorithm component through a publish request, take a certain algorithm component offline through a decommission request, update a certain algorithm component through an update request, delete a certain algorithm component through a delete request, authorize a certain algorithm component to other privacy computing nodes through an authorization request, and so on. After receiving the component management request, the privacy computing node performs corresponding operations on the component management request.

[0056] Based on component task management, users can submit component task management requests through privacy computing nodes. For example, they can query the component task list through a list query request, start a component task through a start request, disable a component task through a disable request, modify a certain parameter in the component task through a parameter modification request, query input data through an input query request, query output data through an output query request, etc. After receiving the component task management request, the privacy computing node performs corresponding operations on the component task.

[0057] Among them, it should be noted that in order to ensure the trustworthiness of algorithm components, it is necessary to authenticate the algorithm components through a security certification agency before publishing the algorithm components to other privacy computing nodes in the network. After passing the authentication, a security certification certificate is obtained. Among them, the security certification certificate can be a digital certificate with a signature, and this digital certificate is used to ensure the trustworthiness and security of the algorithm components. The privacy computing node can send the algorithm component publish request to other privacy computing nodes in the network. The publish request includes authentication information and attribute fields of the algorithm component, where the authentication information is the above-mentioned digital certificate. After receiving the publish request, other privacy computing nodes verify the authenticity of the digital certificate. When it is determined that the digital certificate is true, the algorithm component is added to the local algorithm component list for later use in data interaction and process orchestration.

[0058] Among them, the digital certificate of the algorithm component should include the algorithm name, algorithm description, algorithm version number, validity period, algorithm privacy and security statement, developer signature, security certification agency signature, etc., and be unique. The algorithm privacy and security statement in the algorithm component includes algorithm security definition, algorithm security principle description, usage specifications and usage scope that meet security requirements, etc.

[0059] Based on algorithm process management, users can submit algorithm process management requests through privacy computing nodes. For example, they can publish an algorithm process through a publish request, take the algorithm process offline through a decommission request, update the algorithm process through an update request, delete the algorithm process through a delete request, authorize the algorithm process through an authorization request, and re - orchestrate the algorithm components through a component rearrangement request. After receiving the algorithm process management request, the privacy computing node performs corresponding operations on the algorithm process.

[0060] Based on process task management, users can submit process task management requests through privacy computing nodes. For example: query the list of process tasks through a list query request, run a process task through an execution request, pause the execution of a process task through a pause request, re-execute a process task through a re-execution request, query the input data of a process task through an input query request, query the output data of a process task through an output query request, etc. After the privacy process task receives the algorithm process management request, it performs corresponding operations on the process task.

[0061] After multiple privacy computing nodes have agreed on the above content, when developers develop privacy algorithms, they all develop according to the above agreement. As a result, the algorithm components obtained, the component tasks after running the algorithm components, the algorithm processes after orchestrating the algorithm components, and the process tasks after running the algorithm processes can be implemented in each privacy computing node, improving the algorithm management efficiency in a large-scale privacy computing network and standardizing the privacy computing-related computing operations.

[0062] Next, a method for processing privacy computing tasks based on the above negotiation results will be specifically introduced. As Figure 2 shown, this method is applied to the first privacy computing node in the network. It can be understood that the network also includes a second privacy computing node. Among them, the second privacy computing node can be one or multiple, and the privacy computing platforms running in the first privacy computing node and the second privacy computing node can be the same or different. The same means belonging to the same privacy computing manufacturer and the same version of the privacy computing platform; different means privacy computing platforms with different privacy computing manufacturers and / or versions. This method includes:

[0063] Step 201: Invoke a pre-configured target algorithm process according to the privacy computing task; wherein, the target algorithm process is obtained by orchestrating multiple target algorithm components; the multiple target algorithm components are formed according to a pre-agreed algorithm concept model and attribute fields.

[0064] Step 202: Send the target algorithm process to the second privacy computing node in the network to collaborate with the second privacy computing node to complete the privacy computing task.

[0065] Among them, in step 201, after receiving the privacy computing task, the first privacy computing node obtains the target algorithm process according to the actual requirements of the privacy computing task. It can be understood that the first privacy computing node can obtain the pre-configured algorithm process from the algorithm process list managed locally. If there is no target algorithm process applicable to the privacy computing task in the algorithm process list, it can obtain multiple target algorithm components required for the target algorithm process and orchestrate the multiple target algorithm components to obtain the target algorithm process. Among them, the target algorithm components are written using the above-mentioned negotiated algorithm concept model and attribute fields. Since a single target algorithm component cannot run independently, it is necessary to orchestrate multiple target algorithm components, and the orchestration rule is to orchestrate them in the order of input and output of each target algorithm component according to business requirements. For example: a certain business needs to use a data reading component, a secure intersection component, and a linear regression component. Then, according to the actual business execution requirements, they are orchestrated in the order of data reading component - secure intersection component - linear regression component and written into the "flow_dag" field to complete the orchestration of algorithm components.

[0066] In step 202, after obtaining the target algorithm process by orchestrating algorithm components, the first privacy computing node can send algorithm process cooperation request information to the second privacy computing node in the network. If the second privacy computing node has the willingness to cooperate, it will agree to the request and subsequently participate in the privacy computing task to collaborate on the processing of the privacy computing task. For example, collaborate to complete secure operations, secure intersections, secure statistics, feature engineering, joint modeling, joint prediction, etc.

[0067] In the embodiment of the present application, the target algorithm process is obtained by orchestrating the target algorithm components formed by the pre-agreed algorithm concept model and attribute fields, and the target algorithm process is sent to the second privacy computing node to collaborate on processing the privacy computing task. Since the algorithm concept model and attribute fields are jointly agreed upon by the first privacy computing node and the second privacy computing node, the target algorithm process is applicable to the first privacy computing node and the second privacy computing node, and can enable multiple privacy computing nodes to collaborate to complete a unified privacy computing task through a series of standard interaction interfaces.

[0068] Based on the above embodiment, the collaborating with the second privacy computing node to complete the privacy computing task includes:

[0069] The first privacy computing node and the second privacy computing node respectively configure corresponding algorithm parameters for the target algorithm process they receive and run the configured target algorithm process to collaborate to complete the privacy computing task.

[0070] In a specific implementation process, the first privacy computing node and the second privacy computing node respectively configure corresponding algorithm parameters for the received target algorithm according to their actual situations. For example, the algorithm process includes four algorithm components: data import, secure intersection, logistic regression, and algorithm evaluation. If the first privacy computing node has a higher requirement for data security, the parameter of the secure intersection can be set relatively high, for example, it can be 0.8. If the second privacy computing node has a lower requirement for data security, the parameter of the secure intersection can be set relatively low, for example, it can be 0.5. In addition, since the data sources participating in the privacy computing task processing in the first privacy computing node are different from those in the second privacy computing node, the model parameters obtained after training the logistic regression algorithm component are also different. Therefore, the first privacy computing node and the second privacy computing node respectively configure different algorithm parameters for the received algorithm process. After the first privacy computing node runs the target algorithm process with the configured algorithm parameters, it obtains the component tasks corresponding to each algorithm component and the process task corresponding to the algorithm process. Similarly, after the second privacy computing node runs the target algorithm process with the configured algorithm parameters, it obtains the component tasks corresponding to each algorithm component and the process task corresponding to the algorithm process.

[0071] In the process of the first privacy computing node and the second privacy computing node collaborating to complete the target algorithm process, still taking the above-listed algorithm process as an example, if the algorithm component list in the first privacy computing node includes data import, secure intersection, logistic regression, and algorithm evaluation, and the algorithm component list in the second privacy computing node includes data import, secure intersection, and logistic regression. Then, in the process of executing the target algorithm process, the two parties cooperate to run data import, secure intersection, and logistic regression. The first privacy computing node runs the algorithm evaluation and synchronizes the result after running the algorithm evaluation to the second privacy computing node. At this time, the second privacy computing node does not need to run the algorithm evaluation component. In addition, when the two parties collaborate to run the algorithm process, the privacy computing nodes can securely exchange encrypted intermediate data with each other. For example, encrypted intermediate gradients, encrypted intermediate histogram information, etc. It can be understood that the secure protection technology of the intermediate data can be pre-negotiated among the privacy computing nodes. When transmitting the intermediate data, the intermediate data is securely processed to ensure the security when the intermediate data is exchanged. Among them, there are various secure protection technologies for the intermediate data, such as homomorphic encryption, multi-party secure computing, differential privacy, etc. The embodiments of the present application do not make specific limitations on the secure protection technology.

[0072] In addition, only the final agreed result party (i.e., the first privacy computing node) can obtain the operation result of the entire algorithm process, and can view and download the operation result. Of course, the first privacy computing node can also authorize the operation result to the second privacy computing node for the second privacy computing node to view and / or use. For example, the second privacy computing node can use the operation result as a new resource and input it into other algorithm processes. Specifically, the prediction result can be used as a training sample for a new algorithm process, or the model in the algorithm process can be used as the initial model in a new algorithm process.

[0073] In the embodiments of the present application, since different privacy computing nodes have different input data or different actual requirements, different algorithm parameters will be configured. By running algorithm processes with different algorithm parameter configurations, multiple component tasks and process task running instances can be obtained. These instances support secure exchange of encrypted intermediate data between privacy computing nodes during the processing of privacy algorithm tasks, so as to complete the processing of privacy computing tasks.

[0074] Based on the above embodiments, running the configured target algorithm process includes:

[0075] If the first privacy computing node needs to use the external data of the second privacy computing node during the process of running the configured target algorithm process, it sends a data acquisition request to the second privacy computing node, so that when the second privacy computing node determines that the external data is in an authorized state according to the data acquisition request, it returns the external data to the first privacy computing node.

[0076] In a specific implementation process, the first privacy computing node may need to use external resources such as external data or models in the algorithm process to participate in algorithm collaborative computing. Taking the need to use external resources in the second privacy computing node as an example, the second privacy computing node needs to authorize the data or model resources required by the first privacy computing node. Only authorized and non-sensitive resources can be used across platforms in privacy computing tasks. The authorized state should not be permanent. After authorization by both nodes, the authorization can be cancelled at any time.

[0077] Therefore, when the first privacy computing node needs to use the external data of the second privacy computing node, it can send a data acquisition request to the second privacy computing node. After receiving the data acquisition request, the second privacy computing node determines whether the corresponding data is among the data that has been authorized for the first privacy computing node to use. If so, it can return the data to the first privacy computing node; if it is determined that the first privacy computing node has no permission to use the data, it can evaluate whether authorization is required. After evaluating that authorization is possible and authorizing, it returns the data to the first privacy computing node. If it is learned after evaluation that the first privacy computing node has no permission, it can return a message indicating unauthorized access to the first privacy computing node.

[0078] Among them, the algorithm component installation package, algorithm component container, etc. can also be regarded as a special type of resource. The first privacy computing node retrieves and views the algorithm component information published by the second privacy computing node. If necessary, it submits an application for algorithm component authorization. After being authorized, the algorithm components published by the second privacy computing node can be orchestrated, configured, and used across platforms, and participate in subsequent process tasks for collaborative computing.

[0079] The embodiments of the present application can use the resources in the second privacy computing node across platforms. For resources that are input data, the accuracy of the results obtained by the first privacy computing node during the operation of the algorithm process is improved; for resources that are algorithm components, it is not necessary for the developers of the first privacy computing node to re-develop the algorithm components. While protecting the resource security of data, algorithms, and model providers, it also greatly improves the efficiency of privacy computing task processing.

[0080] Based on the above embodiments, the method further includes:

[0081] Obtain the log information generated during the execution of the privacy computing task;

[0082] Encrypt and store the log information.

[0083] In a specific implementation process, the first privacy computing node can perform algorithmic evidence preservation for the privacy computing task. Algorithmic evidence preservation means that the first privacy computing node can record the log information of static and dynamic algorithm inputs, outputs, and key operations in the privacy computing interconnection network to meet subsequent internal and external regulatory audits. And the log information is encrypted and has the property of non-tamperability, without exposing the privacy data involved in the privacy computing task. For parties that violate the agreement, discovery and tracking can be carried out.

[0084] The embodiments of the present application record the key data and key operations in the privacy computing task through the setting of an evidence preservation mechanism, providing a basis for subsequent internal and external audits, and a task contribution degree calculation method can also be further designed with reference to the evidence preservation.

[0085] Figure 3 The following is a schematic structural diagram of a task processing device for a privacy computing platform based on algorithm interconnection provided by an embodiment of the present application. The device may be a module, a program segment, or code on an electronic device. It should be understood that the device corresponds to the above Figure 2 method embodiment and can execute Figure 2 each step involved in the method embodiment. The specific functions of the device can be referred to the description above. To avoid repetition, the detailed description is appropriately omitted here. The device includes: an algorithm component orchestration module 301 and a collaborative processing module 302, where:

[0086] The algorithm component orchestration module 301 is used to call a pre-configured target algorithm process according to a privacy computing task; wherein, the target algorithm process is obtained by orchestrating a plurality of target algorithm components; the plurality of target algorithm components are formed by the privacy computing node according to a pre-agreed algorithm concept model and attribute fields; the collaborative processing module 302 is used to send the target algorithm process to a second privacy computing node in the network to collaborate with the second privacy computing node to complete the privacy computing task.

[0087] Based on the above embodiment, the collaborative processing module 302 is specifically used for:

[0088] Configuring corresponding algorithm parameters for the received target algorithm process respectively, and running the configured target algorithm process to collaboratively complete the privacy computing task.

[0089] Based on the above embodiment, the algorithm component orchestration module 301 is specifically used for:

[0090] Reading an algorithm component list; the algorithm component list includes algorithm components local to the first privacy computing node and security-certified algorithm components published and authorized by the second privacy computing node;

[0091] Selecting a plurality of target algorithm components from the algorithm component list, and orchestrating the plurality of target algorithm components according to the input-output order of each target algorithm component to obtain the target algorithm process.

[0092] Based on the above embodiment, the device further includes an algorithm component generation module for:

[0093] Receiving a pre-agreed algorithm concept model and attribute fields, and forming algorithm components according to the algorithm concept model and the attribute fields; wherein, the algorithm concept model includes an algorithm component object, a component task object, an algorithm process object, and a process task object; the attribute fields include description fields corresponding to the algorithm component object, the component task object, the algorithm process object, and the process task object respectively.

[0094] Based on the above embodiments, the device further includes an algorithm management module for:

[0095] Receiving an algorithm management request and performing corresponding operations according to the algorithm management request;

[0096] Wherein, the algorithm management request includes an algorithm component management request, a component task management request, an algorithm process management request, and a process task management request;

[0097] The algorithm component management request includes a release request, a decommissioning request, an update request, a deletion request, and an authorization request for the algorithm component;

[0098] The component task management request includes a list query request, a start request, a disable request, a parameter modification request, an input query request, and an output query request for the component task;

[0099] The algorithm process management request includes a release request, a decommissioning request, an update request, a deletion request, an authorization request, and a component rearrangement request for the algorithm process;

[0100] The process task management request includes a list query request, an execution request, a pause request, a parameter modification request, a re-execution request, an input query request, and an output query request for the process task.

[0101] Based on the above embodiments, the device further includes a component release module for:

[0102] Sending a release request for the algorithm component to the second privacy computing node, where the release request includes authentication information and an attribute field of the algorithm component, so that the second privacy computing node authenticates the algorithm component according to the authentication information, and after the authentication is passed, adds the algorithm component to the local algorithm component list.

[0103] Based on the above embodiments, the device further includes a resource authorization module for:

[0104] If the first privacy computing node needs to use external resources of the second privacy computing node during the execution of a configured target algorithm process, sending a resource acquisition request to the second privacy computing node, so that when the second privacy computing node determines that the external resources are in an authorized state according to the resource acquisition request, it returns the external resources to the first privacy computing node.

[0105] Based on the above embodiments, the device further includes a log retention module for:

[0106] Obtaining log information generated during the execution of the privacy computing task;

[0107] Encrypt and store the log information.

[0108] Figure 4 Schematic diagram of the physical structure of the electronic device provided by the embodiment of the present application, as Figure 4 shown, the electronic device includes: a processor 401, a memory 402, and a bus 403; wherein,

[0109] The processor 401 and the memory 402 communicate with each other through the bus 403;

[0110] The processor 401 is configured to call program instructions in the memory 402 to execute the methods provided by the above method embodiments, for example, including: calling a plurality of pre-configured target algorithm components according to a privacy computing task, and orchestrating the plurality of target algorithm components to obtain a target algorithm process; wherein, the plurality of target algorithm components are formed according to a pre-agreed algorithm concept model and attribute fields; sending the target algorithm process to a second privacy computing node in the network to cooperate with the second privacy computing node to complete the privacy computing task.

[0111] The processor 401 may be an integrated circuit chip with signal processing capabilities. The above-mentioned processor 401 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0112] The memory 402 may include, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc.

[0113] This embodiment discloses a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the methods provided in the above method embodiments, for example, including: invoking a pre-configured target algorithm process according to a privacy computing task; wherein, the target algorithm process is obtained by orchestrating a plurality of target algorithm components; the plurality of target algorithm components are formed according to a pre-agreed algorithm concept model and attribute fields; sending the target algorithm process to a second privacy computing node in the network to cooperate with the second privacy computing node to complete the privacy computing task.

[0114] This embodiment provides a non-transitory computer-readable storage medium, which stores computer instructions. The computer instructions cause the computer to execute the methods provided in the above method embodiments, for example, including: invoking a pre-configured target algorithm process according to a privacy computing task; wherein, the target algorithm process is obtained by orchestrating a plurality of target algorithm components; the plurality of target algorithm components are formed according to a pre-agreed algorithm concept model and attribute fields; sending the target algorithm process to a second privacy computing node in the network to cooperate with the second privacy computing node to complete the privacy computing task.

[0115] In the embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections between each other can be through some communication interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical or other forms.

[0116] In addition, the units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0117] Furthermore, in each embodiment of this application, the functional modules can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.

[0118] In this document, relational terms such as first and second are used solely to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0119] The above description is only for the embodiments of this application and is not intended to limit the scope of protection of this application. For those skilled in the art, various changes and modifications can be made to this application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application shall be included within the scope of protection of this application.

Claims

1. A task processing method for a privacy computing platform based on algorithm interconnection, applied to a first privacy computing node in a network, characterized in that Including: Invoking multiple pre-configured target algorithm components according to a privacy computing task, and orchestrating the multiple target algorithm components to obtain a target algorithm process; The multiple target algorithm components are formed according to a pre-agreed algorithm concept model and attribute fields; Sending the target algorithm process to a second privacy computing node in the network to cooperate with the second privacy computing node to complete the privacy computing task; The step of invoking multiple pre-configured target algorithm components according to a privacy computing task and orchestrating the multiple target algorithm components to obtain a target algorithm process includes: Reading an algorithm component list; the algorithm component list includes algorithm components local to the first privacy computing node and security-certified algorithm components published and authorized by the second privacy computing node; Selecting multiple target algorithm components from the algorithm component list, and orchestrating the multiple target algorithm components according to the input-output order of each target algorithm component to obtain the target algorithm process; The method further includes: Receiving a pre-agreed algorithm concept model and attribute fields, and forming algorithm components according to the algorithm concept model and the attribute fields; wherein, the algorithm concept model includes an algorithm component object, a component task object, an algorithm process object, and a process task object; the attribute fields include description fields corresponding to the algorithm component object, the component task object, the algorithm process object, and the process task object respectively; Wherein, the algorithm components and the algorithm process are static objects before algorithm execution. The algorithm component is the code implementation of a single privacy computing algorithm, and multiple algorithm components are orchestrated to form one algorithm process; The component task and the process task are dynamic objects after algorithm execution. The component task is a running instance of a single algorithm component, and the process task is a running instance of the algorithm process.

2. The method according to claim 1, wherein The method further includes: Receiving an algorithm management request, and performing corresponding operations according to the algorithm management request; Wherein, the algorithm management request includes an algorithm component management request, a component task management request, an algorithm process management request, and a process task management request; The algorithm component management request includes a release request, a take-offline request, an update request, a deletion request, and an authorization request for an algorithm component; The component task management request includes a list query request, a start request, a disable request, a parameter modification request, an input query request, and an output query request for a component task; The algorithm process management request includes a release request, a take-offline request, an update request, a deletion request, an authorization request, and a component rearrangement request for an algorithm process; The process task management request includes a list query request, an execution request, a pause request, a parameter modification request, a re-execution request, an input query request, and an output query request for a process task.

3. The method according to claim 1, wherein After generating the algorithm components, the method further includes: Send a release request of the algorithm component to the second privacy computing node, where the release request includes authentication information and an attribute field of the algorithm component, so that the second privacy computing node authenticates the algorithm component according to the authentication information, and after the authentication is passed, adds the algorithm component to the local algorithm component list.

4. The method according to claim 1, wherein The cooperation with the second privacy computing node to complete the privacy computing task includes: The first privacy computing node and the second privacy computing node respectively configure algorithm parameters corresponding to the received target algorithm process, and run the configured target algorithm process to cooperate to complete the privacy computing task.

5. The method according to claim 4, characterized in that The running of the configured target algorithm process includes: If the first privacy computing node needs to use external resources of the second privacy computing node during the process of running the configured target algorithm process, send a resource acquisition request to the second privacy computing node, so that when the second privacy computing node determines that the external resources are in an authorized state according to the resource acquisition request, return the external resources to the first privacy computing node.

6. The method according to any one of claims 1-5, characterized in that, The method further includes: Obtain log information generated during the execution of the privacy computing task; Encrypt and store the log information.

7. A task processing device for a privacy computing platform based on algorithm interconnection, which is applied to a first privacy computing node in a network, and is characterized in that It includes: An algorithm component orchestration module, configured to call a plurality of pre-configured target algorithm components according to a privacy computing task, and orchestrate the plurality of target algorithm components to obtain a target algorithm process; the plurality of target algorithm components are formed by the privacy computing node according to a pre-agreed algorithm concept model and attribute fields; A cooperative processing module, configured to send the target algorithm process to a second privacy computing node in the network to cooperate with the second privacy computing node to complete the privacy computing task; Specifically, the algorithm component orchestration module is configured to: Read the algorithm component list; the algorithm component list includes algorithm components local to the first privacy computing node and security-authenticated algorithm components released and authorized by the second privacy computing node; Select a plurality of target algorithm components from the algorithm component list, and orchestrate the plurality of target algorithm components according to the input-output order of each target algorithm component to obtain the target algorithm process; The device further includes an algorithm component generation module, configured to: Receive a pre-agreed algorithm concept model and attribute fields, and form algorithm components according to the algorithm concept model and the attribute fields; wherein, the algorithm concept model includes an algorithm component object, a component task object, an algorithm process object, and a process task object; the attribute fields include description fields corresponding to the algorithm component object, the component task object, the algorithm process object, and the process task object respectively; Wherein, the algorithm component and the algorithm process are static objects before algorithm operation, the algorithm component is the code implementation of a single privacy computing algorithm, and a plurality of the algorithm components are orchestrated to form one algorithm process; The component task and the process task are dynamic objects after the algorithm runs. The component task is a running instance of a single algorithm component, and the process task is a running instance of the algorithm process.

8. An electronic device, characterized in that, Including: A processor, a memory, and a bus. Among them, The processor and the memory complete communication with each other through the bus; The memory stores program instructions executable by the processor, and the processor can execute the method according to any one of claims 1-7 by calling the program instructions.

9. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer instructions. When the computer instructions are run by the computer, the computer is caused to execute the method according to any one of claims 1-7.

Images