Confidentiality Method and System for AI Workflow Model Trading Platform

Abstract

Multiple embodiments of this specification relate to the field of information technology, and specifically to a confidentiality method and system for an AI workflow model trading platform. The method includes the steps of: receiving a trial loading request for an AI workflow model sent by a client and generating an identification code; converting the interaction code into a code forwarded by the server; adding an end reporting module to the AI workflow model and sending the AI workflow model code and the identification code to the client; when the end reporting module runs on the client, periodically generating a reporting identification code according to the identification code and the current time and storing it in memory, and the client attaches the reporting identification code when executing the instruction to upload the code; when the server verifies that the reporting identification code passes, sending the instruction upload code to the AI server; after receiving the data fed back by the AI server, storing the fed-back data at a storage address matching the result acquisition code.

Description

Technical Field

[0001] Multiple embodiments of this specification relate to the field of information technology, and specifically to a confidentiality method and system for an AI workflow model trading platform. Background Art

[0002] With the development of AI technology, the fields and ways of its application have been continuously enriched. The AI workflow model is one of the ways to apply AI technology. An AI workflow model refers to organizing a series of artificial intelligence algorithms, data processing steps, and business logics into an executable process to solve specific problems or complete specific tasks. Such a model usually includes multiple stages such as data preprocessing, feature engineering, model training, evaluation, and deployment. Each stage may involve different technology stacks and tool sets, and these stages can be linear or iterative, depending on the application requirements. Currently, there are platforms specifically for trading AI workflow models in the market. These platforms allow users to publish the workflow models they create for others to purchase or download and use. At the same time, users can also search for models suitable for their needs on the platform for secondary development or directly apply them to actual scenarios. In this way, not only the sharing of knowledge and technology is promoted, but also the threshold for small and medium-sized enterprises and individual developers to enter the AI field is greatly reduced. However, the current AI workflow trading platforms do not support the trial function, resulting in an increase in the difficulty and risk of AI workflow trading. Summary of the Invention

[0003] Multiple embodiments of this specification describe a confidentiality method and system for an AI workflow model trading platform.

[0004] In a first aspect, an embodiment of this specification provides a confidentiality method for an AI workflow model trading platform, which runs on the server where the platform is located. The server is connected to an AI server, and at least one AI service runs on the AI server. The method includes the steps of:

[0005] Receiving an AI workflow model trial loading request sent by a client and generating an identification code for the client;

[0006] Converting the interaction code in the AI workflow model with the AI server into a forwarding code forwarded by the server, where the forwarding code includes an instruction upload code and a result acquisition code;

[0007] Adding an end reporting module to the AI workflow model and sending the AI workflow model code and the identification code to the client;

[0008] When the end reporting module runs on the client, it periodically generates a reporting identification code according to the identification code and the current time and stores it in the memory. When the client executes the instruction upload code, it attaches the reporting identification code;

[0009] When the server receives the instruction to upload the code, it verifies the reported identification code. When the verification passes, it sends the instruction upload code to the AI server;

[0010] After receiving the data fed back by the AI server, it stores the fed-back data at the storage address matching the result acquisition code.

[0011] In a second aspect, an embodiment of the present specification provides a confidentiality system for an AI workflow model trading platform, including:

[0012] A server and an AI server, the platform runs on the server, the server is connected to the AI server, and at least one AI service runs on the AI server.

[0013] The server includes:

[0014] A receiving module, which receives the AI workflow model trial loading request sent by the client and generates an identification code for the client;

[0015] A conversion module, which converts the interaction code with the AI server in the AI workflow model into a forwarding code forwarded by the server. The forwarding code includes an instruction upload code and a result acquisition code;

[0016] A sending module, which adds an end reporting module to the AI workflow model and sends the AI workflow model code and the identification code to the client. When the end reporting module runs on the client, it periodically generates a reported identification code according to the identification code and the current time and stores it in the memory. When the client executes the instruction upload code, it attaches the reported identification code;

[0017] A verification module, which verifies the reported identification code when receiving the instruction upload code. When the verification passes, it sends the instruction upload code to the AI server;

[0018] A feedback module, which stores the fed-back data at the storage address matching the result acquisition code after receiving the data fed back by the AI server.

[0019] In a third aspect, an embodiment of the present specification provides an electronic device, including a processor and a memory;

[0020] The processor is connected to the memory;

[0021] The memory is used to store executable program code;

[0022] The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method described in any of the above aspects.

[0023] In a fourth aspect, an embodiment of the present specification provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method described in any of the above aspects is implemented.

[0024] In a fifth aspect, an embodiment of the present specification provides a computer program product, including a computer program, and when the computer program is executed by a processor, the method described in any of the above aspects is implemented.

[0025] The beneficial effects brought by the technical solutions provided in some embodiments of the present specification at least include:

[0026] In multiple embodiments of the present specification, the confidentiality method for the AI workflow model trading platform provided can provide a trial function for the AI workflow model trading platform, enabling users to experience the functions of the AI workflow model, but not being able to copy the code of the AI workflow model, thus protecting the rights and interests of the creators; by modifying the code of the AI workflow model and realizing the invocation of the AI service through the server forwarding, the security of the core code of the AI workflow model can be effectively protected; by means of the reverse dependency injection technology, the dynamic loading of the code can be realized, enabling some code to be obtained by the client only during runtime, that is, it can ensure that the AI workflow model can normally implement functions during the trial and make the code of the AI workflow model more secure.

[0027] Other features and advantages of multiple embodiments of the present specification will be further revealed in the following specific embodiments and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present specification, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present specification. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0029] Figure 1 It is a schematic diagram of the AI workflow model trading platform applied in the embodiment of the present specification.

[0030] Figure 2 It is a schematic diagram of the AI workflow model trading details page applied in the embodiment of the present specification.

[0031] Figure 3 It is a schematic diagram of the confidentiality of the AI workflow model trading provided in the embodiment of the present specification.

[0032] Figure 4 Schematic diagram of the AI workflow model transaction confidentiality method provided in the embodiments of this specification.

[0033] Figure 5 Schematic diagram of the method for sending an identification code to a client provided in the embodiments of this specification.

[0034] Figure 6 Schematic diagram of the cache duration adjustment method provided in the embodiments of this specification.

[0035] Figure 7 Schematic diagram of the AI workflow model transaction confidentiality system provided in the embodiments of this specification.

[0036] Figure 8 Schematic diagram of an electronic device provided in the embodiments of this specification. Detailed implementation manners

[0037] The technical solutions of the embodiments of this specification will be explained and described below with reference to the accompanying drawings of the embodiments of this specification. However, the following embodiments are only the preferred embodiments of this specification and not all of them. Based on the embodiments in the implementation manners, other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of this specification.

[0038] Terms such as "first", "second", "third", etc. in the specification, claims and the above-mentioned drawings of this specification are used to distinguish different objects rather than to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices.

[0039] In the following description, terms indicating orientation or positional relationships such as "inside", "outside", "above", "below", "left", "right", etc. are only for the convenience of describing the embodiments and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation of this specification.

[0040] The data involved in this application are all information and data authorized by users or fully authorized by all parties, and the collection of relevant data complies with the relevant laws, regulations and standards of relevant countries and regions.

[0041] Before introducing the technical solutions described in this specification, the application scenarios of the technical solutions and related technologies will be introduced.

[0042] With the rapid development of artificial intelligence technology, AI Workflow Models have become an indispensable part of modern data science and machine learning projects. AI Workflow Models organize and manage a series of complex tasks and steps in a systematic way. Simply put, an AI Workflow Model can be regarded as a workflow model that uses AI to complete some steps on the basis of a traditional automated processing flow. For example, in a traditional data processing flow, data cleaning and preprocessing usually require a lot of manual operations, such as removing duplicate values, filling in missing data, standardizing numerical values, etc. By introducing AI technology, these steps can become more intelligent and efficient. Machine learning algorithms are used to identify and correct errors or inconsistencies in the data. For example, natural language processing (NLP) technology can be used to clean spelling mistakes or grammar problems in text data. Prediction models (such as K-nearest neighbor algorithm, random forest, etc.) are used to predict and fill in missing values based on existing data, rather than simply filling them with the mean or median.

[0043] Another example is that in a typical customer service automation system, the traditional workflow may include steps such as receiving customer requests, classifying problem types, and assigning them to the appropriate customer service staff for handling. After introducing AI, the functions of automatic classification and intelligent response can be realized. Automatic classification uses natural language processing technology to automatically classify customer requests, identify problem types (such as technical support, bill inquiry, etc.), and forward the requests to the corresponding departments or system modules. Intelligent response is based on historical conversation records and a common question library, and uses chatbots to automatically generate response suggestions for customer service staff to refer to or directly send to customers.

[0044] In the e-commerce industry, image processing is an important link to improve the user experience and increase the sales conversion rate. In order to show all angles of a product, a large number of photos usually need to be taken, which is a tedious task for photographers and post-production staff. The AI Workflow Model can generate 3D models of products through 3D scanning technology and rendering engines. Then, these models are used to automatically generate product images taken from different angles without actual shooting.

[0045] In short, the AI Workflow Model technology can not only significantly improve development efficiency, reduce error rates, but also enable non-professionals to participate in the creation process of AI solutions, effectively accelerating the application of AI technology. Currently, there are platforms specifically for trading AI Workflow Models. As shown Figure 1 below, these platforms allow users to publish the workflow models they create for others to purchase or download and use. The trading platform for AI Workflow Models generally includes a screening area 11 and a list area 12. Please refer to the appendix Figure 2, after clicking on an item in the list area 12, the details page of the AI workflow model will be expanded. The details page will display the name 13 of the AI workflow model, the function introduction, the configuration area 14, and the price, and provide a trial button 15 and a purchase button 16. When the user clicks the purchase button 16, the server 21 where the platform is located will provide the usage page of the AI workflow model, or the deployment code, or the APP installation file to the client 23. When the user clicks the trial button 15, the usage page will be displayed, but the code of the AI workflow model loaded on this usage page is processed and protected so that the code of the AI workflow model cannot be copied. That is, the function of the AI workflow model can only be realized on the usage page provided for the trial. This process makes use of the confidentiality method and system of the AI workflow model trading platform provided in this specification.

[0046] The method provided in this specification is applied to a system architecture such as Figure 3 shown. Figure 3 This is a schematic diagram of an architecture of the system architecture in an embodiment of this specification. As Figure 3 shown, the system architecture includes a server 21, an AI server 22, and a client 23, and the client 23 is provided with an interaction interface. Among them, the interaction interface can run on the device of the client 23 in the form of a browser, or can run on the client 23 in the form of an independent application (APP), etc. For the specific display form of the client 23, no limitation is made here.

[0047] In this specification, after the purchase button 16 is clicked, the code of the AI workflow model will be downloaded from the server 21 and will directly interact with the AI server 22 during the subsequent work process to complete the function of the AI workflow model. After the trial button 15 is clicked, the code of the AI workflow model will be processed by the server 21 and provided to the client 23. When the client 23 runs the AI workflow model, it will indirectly interact with the AI server 22 through the server 21 as a medium. At this time, the client 23 will not directly interact with the AI server 22 at all.

[0048] The server 21 and the AI server 22 involved in this specification can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms. The client 23 can be a smart phone, a tablet computer, a notebook computer, a handheld computer, a personal computer, a smart speaker, a smart TV, a smart watch, a vehicle-mounted device, a wearable device, etc., but is not limited thereto. The client 23 and the server can be directly or indirectly connected through wired or wireless communication methods, which are not limited in this specification. The numbers of the server 21 and the client 23 are also not limited. Among them, the response documents involved in this specification can be stored in a database. Among them, the database can be regarded as an electronic filing cabinet in short - a place for storing electronic files, and users can perform operations such as adding, querying, updating, and deleting data in the files.

[0049] A so - called "database" is a collection of data stored together in a certain way, shared by multiple users, with as little redundancy as possible, and independent of application programs. A Database Management System (DBMS) is a computer software system designed to manage a database, generally having basic functions such as storage, interception, security guarantee, backup, etc. The DBMS can be classified according to the database model it supports, such as relational, Extensible Markup Language (XML); or according to the type of computer it supports, such as server clusters, mobile phones; or according to the query language it uses, such as Structured Query Language (SQL), XQuery; or according to the performance impulse focus, such as maximum scale, highest running speed; or other classification methods. No matter which classification method is used, some DBMSs can cross categories. For example, they can support multiple query languages at the same time.

[0050] Please refer to the appendix Figure 4 , this specification first provides a confidentiality method for an AI workflow model trading platform, which runs on the server 21 where the platform is located. The server 21 is connected to the AI server 22, and at least one AI service runs on the AI server 22, including the steps:

[0051] Step S101) Receive the AI workflow model trial loading request sent by the client 23 and generate an identification code for the client 23.

[0052] The method for generating a reporting identification code based on the identification code and the current time includes:

[0053] The initial value of the generated serial number is 0, and every time a preset time period passes, the serial number is incremented by 1.

[0054] Extract the hash values of the identification code and the serial number as the reporting identification code.

[0055] The preset time period can be every minute, every 10 minutes, every hour, every day, etc. Exemplarily, taking one minute as the preset time period, the serial number is incremented by 1. The serial number can be represented in hexadecimal (representation in other number systems does not affect implementation). Combining the original identification code and the current serial number, calculating their hash value can obtain the reporting identification code. Exemplarily, the identification code is: ABC123, the initial value of the serial number is 0, and combining the identification code and the serial number is: ABC123_0. Using the SHA-256 algorithm to calculate the hash value of "ABC123_0", the result obtained is: 676E1BBDCAD44B87B6EB076FB5AEB88E5CDD4176852E5273F044AAC408F2E800, which is the reporting identification code. One minute later, the serial number is updated to 1, and using the SHA-256 algorithm to calculate the hash value of "ABC123_1", a new reporting identification code is obtained. Thus, the reporting identification code is updated every minute.

[0056] When the code of the AI workflow model is copied and executed, an operation of setting the initial value of the serial number to 0 will be performed, which will cause the subsequent generated reporting identification code to be different from that on the server 21.

[0057] Step S102) Convert the code for interacting with the AI server 22 in the AI workflow model into forwarding code forwarded by the server 21. The forwarding code includes instruction upload code and result acquisition code.

[0058] Exemplarily, the AI workflow model is used to process image classification tasks. The original client 23 code directly communicates with the AI server 2221. When the user clicks the trial button 15 to obtain the code of the AI workflow model, the code includes an intermediate layer that is responsible for receiving the client 23 request, forwarding it to the AI server 22, and returning the result to the client 23. The instruction upload code is responsible for receiving the client 23 request and forwarding it to the AI server 22, and the result acquisition code is responsible for obtaining the processing result from the AI server 22 and returning the result to the client 23.

[0059] Exemplarily, the client 23 code uses the send_request_to_forward_server function to be responsible for reading the image file and sending it to the forward server via an HTTP POST request. The URL of the forward server is http: / / localhost:5000 / classify, which is the address where the forward server listens.

[0060] In the server where the platform is located, a simple web service is created using the Flask framework, listening for POST requests on the / classify path. When a request from the client 23 is received, it checks whether an image file has been uploaded. The received image file is forwarded to the AI server 22 (http: / / ai-server:8000 / classify). The response from the AI server 22 is received and returned to the client 23.

[0061] Similarly, a web service is created on the AI server 22 using the Flask framework, listening for POST requests on the / classify path. It receives the image file and calls the simulated image classification model, and finally returns the classification result to the forward server.

[0062] In this way, even if the code of the AI workflow model is copied, without the forwarding by the server where the platform is located, it cannot interact with the AI server 22 and cannot complete the functions of the AI workflow model. Therefore, after the user clicks the trial button 15, by marking the client 23 with trial permission and only forwarding the client 23 with trial permission, it can be achieved that the copied AI workflow model code cannot complete the corresponding functions during execution, thus protecting the code security of the AI workflow model.

[0063] Step S103) Add an end reporting module to the AI workflow model, and send the AI workflow model code and the identification code to the client 23.

[0064] Please refer to the appendix Figure 5 , the method of sending the AI workflow model code and the identification code to the client 23 includes:

[0065] Before sending the AI workflow model code and the identification code to the client 23, the server adds a program injection module to the AI workflow model and sets an injection code for the program injection module. The program injection module includes several lines of code for injecting code. Exemplarily, dependency injection is used for code injection. Its general composition and process include: a basic application program, a dependency injection module, a custom processor, and modification of the basic application program. Among them, the basic application program defines a TaskProcessor class and its subclass ImageTaskProcessor for processing image data. Its main function uses the default ImageTaskProcessor to process image data and print the result.

[0066] The dependency injection module provides an Injector class for registering and obtaining different processor instances. It provides a register_processor method for registering processor classes. It provides a get_processor method for obtaining a processor instance according to the name.

[0067] The custom processor defines a new processor class CustomImageTaskProcessor, which implements different processing logics. The injector.register_processor method is used to register the custom processor into the dependency injection module.

[0068] In the modification of the basic application program, the main function needs to be modified so that it can select different processor instances according to the passed parameters. For example, if the 'custom' parameter is specified, the custom processor is used; otherwise, the default processor is used. Among them, the custom processor is the code to be injected.

[0069] Step S202) The server generates a setting statement for writing the identification code into the memory and associating the corresponding memory location with a preset variable, caches the setting statement after associating it with the injection code.

[0070] Step S203) When the program injection module is executed on the client 23, it connects to the port specified by the server and provides the injection code. The corresponding setting statement is retrieved through the injection code.

[0071] Step S204) The server sends the corresponding setting statement to the client 23 according to the injection code and clears the cached setting statement.

[0072] Step S205) The program injection module obtains the setting statement from the server, reads the setting statement into the memory for execution, and then deletes the setting statement.

[0073] In the custom processor, several statements are set to allocate a storage space in memory and associate it with a variable name. Then, in this storage space, the identification codes directly attached in the code are written into the corresponding memory spaces. Among them, the several statements set are stored in a variable or an object. After execution, the relevant values of the variable or the object are overwritten, so that the set statements will disappear after execution. Exemplarily, the preset variable is Id_code_container. The variable Id_code_container is associated with a storage space in memory. After the identification code is written into the memory, the code can call the variable Id_code_container to use the identification code. At this time, even if the code of the AI workflow model is copied, the set statements cannot be obtained, so the correct identification code cannot be set, that is, the copied code cannot use the AI workflow model properly.

[0074] When the reporting module described in step S104 runs on the client 23, it periodically generates a reporting identification code based on the identification code and the current time and stores it in memory. When the client 23 executes the instruction to upload the code, the reporting identification code is attached.

[0075] When the server receives the instruction to upload the code in step S105, it verifies the reporting identification code. When the verification passes, it sends the instruction to upload the code to the AI server 22.

[0076] Among them, the method for the server to verify the reporting identification code includes:

[0077] Record the time when the AI workflow model code and the identification code are sent to the client 23;

[0078] According to the current time and the preset duration, calculate the restored value of the current sequence number value;

[0079] Extract the hash values of the identification code and the restored value of the sequence number value, and compare them with the reporting identification code. When the comparison is consistent, it is determined that the verification of the reporting identification code passes.

[0080] By agreeing on the same hash function, the verification of the identification code and the sequence number value can be achieved. The server and the client 23 have the same identification code, and the server records the timestamp when the client 23 obtains the AI workflow model code. After the code of the AI workflow model is loaded, it will be automatically executed immediately, so that the server obtains the moment when the sequence number value on the client 23 is set to the initial value 0. Thus, at a subsequent moment, the same sequence number value as that of the client 23 can be calculated by tracking the timestamp.

[0081] When the copy of the code of the AI workflow model is obtained, the server will not know the time when the copied code is first executed, that is, it will not know the moment when the sequence number value in the code is set to the initial value 0, so it is impossible to calculate the subsequent sequence number values. Therefore, the server can verify the source of the code of the AI workflow model by verifying the reported identification code generated by means of the sequence number value.

[0082] Step S106) After receiving the data fed back by the AI server 22, store the fed-back data at the storage address matching the result acquisition code. The result can be obtained by the result acquisition code reading the content from the corresponding storage address.

[0083] On the other hand, in another implementation, the server stores the identifier of each client 23 and records the interaction history with the client 23;

[0084] After the server caches the setting statement for a preset cache duration, delete the setting statement.

[0085] When the server receives the instruction to upload the code, verify the reported identification code. If the verification fails, please refer to the appendix Figure 6 , and execute the following steps:

[0086] Step S301) Obtain the client 23 identifier and read the recorded interaction history according to the client 23 identifier.

[0087] Step S302) Obtain the function usage rate of the user for the functions of the AI workflow model, the time concentration of the user using the AI workflow model, the total number of AI workflow models used by the user, and the parameter repetition degree of the user using the AI workflow model according to the interaction history.

[0088] Step S303) Obtain a trial score according to the function usage rate, time concentration, total number, and parameter repetition degree.

[0089] Step S304) When the trial score is greater than a preset threshold, reduce the cache duration corresponding to the setting statement.

[0090] The recorded interaction history can reflect the usage of the AI workflow model by the user corresponding to the client 23. The function usage rate of the functions of the AI workflow model represents the ratio of the number of functions used by the user to the total number of all functions. Exemplarily, the AI workflow model has four functions: intelligent matting, multi-angle image generation, background replacement, and partial replacement. If the user uses intelligent matting and multi-angle image generation, the function usage rate is 0.5. Generally speaking, the user's demand for the functions of the AI workflow model is based on their work content, and the normal user's demand basically does not cover all the functions of the AI workflow model. When the function usage rate is too high, the user may be using the AI workflow model for unconventional purposes.

[0091] The time concentration of the user's use of the AI workflow model represents the ratio of the length of time when the user uses the AI workflow model relatively concentrated in a day to 24 hours. Generally, the time period when the user uses the AI workflow model is basically fixed, and the concentrated time will not last too long. When the user uses the AI workflow model for a long time and concentratedly every day, it means that their use of the AI workflow model far exceeds their work needs and may be for unconventional use. The total number of AI workflow models used by the user represents the total number of AI workflow models that the user has tried and purchased. When the user has tried too many AI workflow models, it may not be based on actual needs. The parameter repeatability of the user's use of the AI workflow model represents the ratio of the number of times the repeated parameters are provided to the total number of times all parameters are provided among all the parameters provided to all the AI workflow models when the user tries and uses the purchased AI workflow models. Exemplarily, when using multiple AI workflow models, providing the same picture indicates that it is not using the AI workflow model based on actual needs. The trial score is equal to the weighted sum of the function usage rate, time concentration, total number, and parameter repeatability. When the trial score exceeds the preset threshold, it means that the probability that the user will copy the code of the tried AI workflow model is relatively high. At this time, the cache duration corresponding to the setting statement of this user should be reduced, so that the time for this user to obtain the setting statement is shortened, the time window for copying the code of the AI workflow model is shortened, and the difficulty increases.

[0092] On the other hand, in another embodiment, before sending the AI workflow model code and the identification code to the client 23, the server adds an encryption module and a decryption module to the AI workflow model;

[0093] When the encryption module runs on the client 23, it encrypts some code segments of the AI workflow model at a preset period;

[0094] After the decryption module writes the decryption key pre-stored in the file into the memory and associates it with a preset variable, it deletes the file where the decryption key is located;

[0095] When the client 23 runs the encrypted code segment, the decryption module reads the decryption key from the memory and decrypts the code segment.

[0096] When the encryption module runs on the client 23, it automatically encrypts some code segments of the workflow model at a preset time period (such as every ten minutes, several hours, or once a day). The way of regular encryption increases the difficulty for attackers to obtain and utilize these codes.

[0097] The decryption module reads the key required for decryption from a pre-stored file and writes this key into the memory. At the same time, it associates it with a specific variable for code calling. To further improve security, after the key is loaded into the memory, the original file storing the key is deleted, so that even if the storage space is accessed, the attacker cannot obtain the key. Thus, the encrypted code cannot be decrypted.

[0098] On the other hand, in another embodiment, before sending the AI workflow model code and the identification code to the client 23, the server adds a sequence module to the AI workflow model. The sequence module is provided with a statement array, and the statements of the code of the AI workflow model are stored in the statement array in a scrambled order. The sequence module is provided with a record file for recording the correct statement order;

[0099] When the container module runs on the client 23, it reads the sorting table into the memory and then deletes the record file;

[0100] When the client 23 runs the AI workflow model, a copy statement array is newly created in the persistent storage space, and a copy sorting table is newly created in the memory. According to the sorting table, statements are taken out from the statement array one by one;

[0101] The taken-out statements are randomly stored in the positions in the copy statement array and recorded in the copy sorting table;

[0102] After all statements are taken out and executed, the copy statement array and the copy sorting table are used to overwrite the statement array and the sorting table respectively.

[0103] When all the code is stored in the statement array according to the statements, that is, all the code is stored in a variable. When the code is copied, the value of the variable will not be directly copied. And the sorting of the code in the variable is incorrect, and the statements of the code need to rely on the sorting table stored in the memory to be executed correctly. The sorting table exists in the memory, so it will not be directly copied. Thus, the security of the code is guaranteed. That is, when the user uses the trial function, they cannot obtain the complete executable code of the AI workflow model.

[0104] On the other hand, this specification provides a confidentiality system for an AI workflow model trading platform, including:

[0105] A server and an AI server 22. The platform runs on the server, and the server is connected to the AI server 22. At least one AI service runs on the AI server 22.

[0106] Please refer to the appendix Figure 7 , and the server includes:

[0107] A receiving module 100 that receives a trial loading request for an AI workflow model sent by a client 23 and generates an identification code for the client 23.

[0108] A conversion module 200 that converts the interaction code with the AI server 22 in the AI workflow model into a forwarding code forwarded by the server. The forwarding code includes an instruction upload code and a result acquisition code.

[0109] A sending module 300 that adds an end reporting module to the AI workflow model and sends the AI workflow model code and the identification code to the client 23. When the end reporting module runs on the client 23, it periodically generates a reporting identification code according to the identification code and the current time and stores it in the memory. When the client 23 executes the instruction upload code, it attaches the reporting identification code.

[0110] A verification module 400 that verifies the reporting identification code when receiving the instruction upload code. When the verification passes, it sends the instruction upload code to the AI server 22.

[0111] A feedback module 500 that stores the feedback data received from the AI server 22 at a storage address matching the result acquisition code.

[0112] Please refer to Figure 8 The structural schematic diagram of an electronic device provided by an embodiment of this specification shown.

[0113] As Figure 8As shown, the electronic device 1100 may include: at least one processor 1101, at least one network interface 1104, a user interface 1103, a memory 1105, and at least one communication bus 1102. Among them, the communication bus 1102 can be used to realize the connection and communication of the above-mentioned components. Among them, the user interface 1103 may include buttons, and the optional user interface may further include a standard wired interface and a wireless interface. Among them, the network interface 1104 may include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, etc. Among them, the processor 1101 may include one or more processing cores. The processor 1101 connects various parts within the entire electronic device 1100 through various interfaces and lines, and executes various functions of the routing device 1100 and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 1105, and calling the data stored in the memory 1105. Optionally, the processor 1101 may be implemented in at least one of the hardware forms of DSP, FPGA, and PLA. The processor 1101 may integrate one or a combination of several of CPU, GPU, and modem, etc. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communication.

[0114] It can be understood that the above-mentioned modem may not be integrated into the processor 1101 and may be implemented separately by a single chip.

[0115] Among them, the memory 1105 may include RAM and may also include ROM. Optionally, the memory 1105 includes a non-transitory computer-readable medium. The memory 1105 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 1105 may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-mentioned method embodiments, etc.; the data storage area may store the data involved in the above-mentioned method embodiments. Optionally, the memory 1105 may further be at least one storage device located far from the aforementioned processor 1101. The memory 1105, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and application programs. The processor 1101 can be used to call the application programs stored in the memory 1105 and execute the methods in the above-mentioned multiple embodiments.

[0116] The embodiments of this specification also provide a computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When they run on a computer or a processor, the computer or the processor is caused to execute multiple steps in the above embodiments. If each component module of the above electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in the computer-readable storage medium.

[0117] The embodiments of this specification also provide a computer program product, including a computer program. When the computer program is executed by a processor, multiple steps in the above embodiments are implemented.

[0118] Without conflict, the technical features in this embodiment and the implementation scheme can be combined arbitrarily.

[0119] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes multiple computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this specification are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired manner (such as coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or a wireless manner (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or a data center that integrates multiple available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a Digital Versatile Disc (DVD)), or a semiconductor medium (for example, a Solid State Disk (SSD)), etc.

[0120] When implemented through hardware or firmware, the foregoing method flow is programmed into a hardware circuit to obtain a corresponding hardware circuit structure and implement corresponding functions. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is an integrated circuit whose logic function is determined by a user's programming of the device. A designer can program on their own to "integrate" a digital system onto a PLD, without having to ask a chip manufacturer to design and fabricate a dedicated integrated circuit chip. Moreover, nowadays, instead of manually fabricating an integrated circuit chip, this programming is mostly implemented using "logic compiler" software, which is similar to the software compiler used in program development and writing. The original code before compilation also has to be written in a specific programming language, which is called a hardware description language (HDL), and there are not only one but many kinds of HDLs. Those skilled in the art should also be clear that as long as the method flow is slightly logically programmed in the above-mentioned several hardware description languages and programmed into an integrated circuit, it is easy to obtain a hardware circuit that implements the logical method flow.

[0121] The embodiments described above are merely described in the preferred embodiment manner of this specification, and do not limit the scope of this specification. Without departing from the design spirit of this specification, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of this specification shall fall within the protection scope determined by the claims of this specification.

Claims

1. A confidentiality method for an AI workflow model trading platform, which runs on the server where the platform is located. The server is connected to an AI server, and at least one AI service runs on the AI server. It is characterized in that, Including the steps: Receiving an AI workflow model trial loading request sent by a client and generating an identification code for the client; Converting the code in the AI workflow model that interacts with the AI server into a forwarding code forwarded by the server, where the forwarding code includes an instruction upload code and a result acquisition code; Adding an end reporting module to the AI workflow model and sending the code of the AI workflow model and the identification code to the client; When the end reporting module runs on the client, periodically generating a reporting identification code based on the identification code and the current time and storing it in memory, and the client attaches the reporting identification code when executing the instruction upload code; When the server receives the instruction upload code, it verifies the reporting identification code, and if the verification passes, it sends the instruction upload code to the AI server; After receiving the data fed back by the AI server, storing the fed-back data at a storage address matching the result acquisition code, and the result acquisition code can obtain the fed-back data by reading the content from the corresponding storage address; The step of sending the code of the AI workflow model and the identification code to the client includes: Before sending the code of the AI workflow model and the identification code to the client, the server adds a program injection module to the AI workflow model and sets an injection code for the program injection module; The server generates a setting statement for writing the identification code into memory and associating the corresponding memory location with a preset variable, caches the setting statement after associating it with the injection code; When the program injection module is executed on the client, it connects to the port specified by the server and provides the injection code; The server sends the corresponding setting statement to the client according to the injection code and clears the cached setting statement; The program injection module obtains the setting statement from the server, reads the setting statement into memory for execution, and then deletes the setting statement.

2. The AI workflow model transaction platform confidentiality method according to claim 1, characterized in that The step of generating a reporting identification code based on the identification code and the current time includes: The initial value of the generated serial number value is 0, and every time a preset time period passes, the serial number value is incremented by 1; Extracting the hash value of the identification code and the serial number value as the reporting identification code; The step of the server verifying the reporting identification code includes: Recording the time when the code of the AI workflow model and the identification code are sent to the client; Calculating a restored value of the current serial number value according to the current time and the preset time period; Extracting the hash value of the identification code and the restored value of the serial number value, comparing it with the reporting identification code, and determining that the reporting identification code passes the verification when the comparison is consistent.

3. The AI workflow model transaction platform confidentiality method according to claim 1, characterized in that The server stores the identifier of each client and records the interaction history with the client; After the server caches the setting statement for a preset cache duration, it deletes the setting statement; When the server receives an instruction to upload code, it verifies the reported identification code. If the verification fails, the following steps are executed: Obtain the client identifier and read the recorded interaction history according to the client identifier; Obtain the function usage rate of the user for the functions of the AI workflow model, the time concentration of the user using the AI workflow model, the total number of AI workflow models used by the user, and the parameter repetition degree of the user using the AI workflow model according to the interaction history; Obtain a trial score according to the function usage rate, time concentration, total number, and parameter repetition degree; When the trial score is greater than a preset threshold, reduce the cache duration corresponding to the setting statement.

4. The AI workflow model trading platform confidentiality method according to claim 1 or 2, characterized in that Before sending the code of the AI workflow model and the identification code to the client, the server adds an encryption module and a decryption module to the AI workflow model; When the encryption module runs on the client, it encrypts some code segments of the AI workflow model at a preset period; After the decryption module writes the decryption key pre-stored in the file into the memory and associates it with a preset variable, it deletes the file where the decryption key is located; When the client runs the encrypted code segment, the decryption module reads the decryption key from the memory and decrypts the code segment.

5. The AI workflow model trading platform confidentiality method according to claim 1 or 2, characterized in that Before sending the code of the AI workflow model and the identification code to the client, the server adds a sequence module to the AI workflow model. The sequence module is provided with a statement array, and the statements of the code of the AI workflow model are stored in the statement array after being shuffled. The sequence module is provided with a record file for recording the correct statement order; When the sequence module runs on the client, it reads the sorting table into the memory and then deletes the record file; When the client runs the AI workflow model, a copy statement array is newly created in the persistent storage space, and a copy sorting table is newly created in the memory. According to the sorting table, statements are taken out from the statement array one by one; Randomly store the taken-out statements in the positions in the copy statement array and record them in the copy sorting table; After all statements are taken out and executed, use the copy statement array and the copy sorting table to respectively overwrite the statement array and the sorting table.

6. The confidentiality system of the AI workflow model trading platform is characterized in that Including: A server and an AI server. The platform runs on the server. The server is connected to the AI server, and at least one AI service runs on the AI server. The server includes: A receiving module that receives the AI workflow model trial loading request sent by the client and generates an identification code for the client; A conversion module that converts the code in the AI workflow model that interacts with the AI server into a forwarding code forwarded by the server. The forwarding code includes an instruction upload code and a result acquisition code. A sending module, which adds an end reporting module to the AI workflow model and sends the code of the AI workflow model and the identification code to the client. When the end reporting module runs on the client, it periodically generates a reporting identification code according to the identification code and the current time and stores it in the memory. When the client executes the instruction to upload the code, it attaches the reporting identification code; A verification module, which verifies the reporting identification code when receiving the instruction to upload the code. When the verification is passed, it sends the instruction to upload the code to the AI server; A feedback module, which stores the feedback data received from the AI server at a storage address matching the result acquisition code. The result acquisition code can obtain the feedback data by reading the content from the corresponding storage address; The steps of sending the code of the AI workflow model and the identification code to the client include: Before sending the code of the AI workflow model and the identification code to the client, the server adds a program injection module to the AI workflow model and sets an injection code for the program injection module; The server generates a setting statement for writing the identification code into the memory and associating the corresponding memory location with a preset variable, caches the setting statement after associating it with the injection code; When the program injection module is executed on the client, it connects to the port specified by the server and provides the injection code; The server sends the corresponding setting statement to the client according to the injection code and clears the cached setting statement; The program injection module obtains the setting statement from the server, reads the setting statement into the memory for execution, and then deletes the setting statement.

7. An electronic device, characterized in that, It includes a processor and a memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method according to any one of claims 1-5; 8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program, when executed by the processor, implements the method according to any one of claims 1-5; 9. A computer program product, comprising a computer program, characterized in that, The computer program, when executed by the processor, implements the method according to any one of claims 1-5.

Images