Digital resource management method and apparatus, and cluster, medium and program product

Abstract

The embodiments of the present application relate to the technical field of artificial intelligence. Disclosed are a digital resource management method and apparatus, and a cluster, a medium and a program product, which ensure the availability of a model which has been subjected to training or inference. The method is applied to a management platform, wherein the management platform is used for managing at least one model training platform, the management platform is connected to a plurality of data spaces, a target data space is present among the plurality of data spaces, and digital resources of a data provider are stored in the target data space. The method comprises: acquiring target resource information, which is input by a data user, wherein the target resource information is used for indicating a target digital resource among at least one digital resource, and the target digital resource is provided with a target usage permission; acquiring training platform information, which is input by the data user; generating a data usage strategy on the basis of the target usage permission, wherein the data usage strategy is used for indicating a manner in which a target training platform uses the target digital resource during model training and / or model inference; and sending the data usage strategy to the target training platform.

Description

Digital resource management method, device, cluster, medium and program product

[0001] This application claims priority to the Chinese patent application filed with the State Intellectual Property Office on December 20, 2023, with application number 202311763489.3 and application name “A method and device for managing artificial intelligence generated products”, the entire contents of which are incorporated by reference into this application.

[0002] This application claims priority to the Chinese patent application filed with the State Intellectual Property Office on May 31, 2024, with application number 202410702755.X and application name “Digital Resource Management Method, Device, Cluster, Medium and Program Product”, the entire contents of which are incorporated by reference into this application. Technical Field

[0003] The present application relates to the field of artificial intelligence technology, and in particular to digital resource management methods, devices, clusters, media, and program products. Background Art

[0004] With the continuous maturity and widespread application of artificial intelligence (AI) technology, generative AI (AI generated content, AIGC) technology has emerged. Through the application of AIGC technology, mature digital resources can be used as training sets to train AIGC models. After users input the feature information of the digital resources they need to obtain into the AIGC model, the AIGC model can generate digital resources that meet their needs.

[0005] Since the digital resources used as training sets, for example, digital resources with certain usage permissions, are used for model training, the trained AIGC model will incorporate the relevant features of the digital resources with certain usage permissions. Currently, the training sets obtained when training the AIGC model can be digital resources crawled from the internet. Directly using digital resources obtained from the internet for AIGC model training enables the trained AIGC model to output digital resources that meet user needs.

[0006] In related technologies, since the digital resources used as training sets may have unclear usage permissions, after using the digital resources with unclear usage permissions to train the AIGC model, the digital resources output by the trained AIGC model may exceed the usage permissions that the data provider wants to grant to the digital resources, resulting in the AIGC model being unable to be used normally. Summary of the Invention

[0007] The embodiments of the present application provide a digital resource management method, device, cluster, medium and program product, which ensure the traceability control of digital resources and the unified management of the use rights of digital resources, so that the training platform can use digital resources within the use rights of digital resources to train or infer models, thereby ensuring the availability of the trained or inferred models.

[0008] In the first aspect, the present application provides a digital resource management method, which is applied to a management platform, the management platform is used to manage at least one model training platform, the management platform is connected to multiple data spaces, there is a target data space among the multiple data spaces, and the target data space stores at least one digital resource of the data provider. The method includes: obtaining target resource information input by the data user, the target resource information is used to indicate the target digital resource in at least one digital resource, wherein the target digital resource is set with a target usage permission; obtaining training platform information input by the data user, the training platform information is used to indicate the target training platform in at least one model training platform; generating a data usage policy based on the target usage permission, the data usage policy is used to indicate the way in which the target training platform uses the target digital resource when training the model and / or model inference; and sending the data usage policy to the target training platform.

[0009] It is understandable that by building a management platform that manages at least one model training platform and is connected to the data space of digital resources provided by multiple data providers, the management platform can determine the target digital resource and the target training platform after receiving the target resource information input by the data user and the training platform information input by the data user. Since the target digital resource is set with a target usage permission, the management platform can generate a data usage policy based on the target usage permission and send the data usage policy to the target training platform so that the target training platform can use the target digital resource to train or infer the model in the target training platform according to the data usage policy. Since the management platform can set clear usage permissions for the digital resources provided by each data provider, after receiving the target resource information indicating the target digital resource input by the data user, the management platform can generate a data usage policy based on the target usage permission of the target digital resource so that the target training platform can train or infer the model under the premise of complying with the target usage permission of the target digital resource, thereby ensuring the availability of the trained or inferred model on the target training platform.

[0010] In one possible implementation, if the target usage right is full usage right, generating a data usage policy based on the target usage right includes: generating a data usage policy based on the full usage right, and the data usage policy is used to instruct the target training platform to use the target digital resource as training set data multiple times when training the model.

[0011] It is understandable that when the usage permission of the target digital resource is set to full usage permission, since the full usage permission can grant the data user the right to use the target digital resource multiple times and the target digital resource can be used for model training, the data usage policy generated by the management platform can be that the target training platform supports multiple uses of the target digital resource as training set data when training the model, ensuring that the target digital resource is used to complete the training of the model when the usage permission permits.

[0012] In one possible implementation, if the target usage right is a restricted usage right, a data usage policy is generated based on the target usage right, including: generating a data usage policy based on the restricted usage right, the data usage policy is used to instruct the target training platform to use the target digital resource as the inference set data during model inference.

[0013] In one possible implementation, the multiple data spaces include an inference data space, and the data usage policy is further used to instruct the model to be transferred to the inference data space, and to use the target digital resource as the inference set data when performing model inference in the inference data space.

[0014] It is understandable that when the usage permission of the target digital resource is set to restricted usage permission, since the restricted usage permission can grant the data user one-time use of the target digital resource, it only supports the use of the target digital resource for model inference, and model inference needs to be performed in a non-completely open controlled environment. Therefore, the data usage policy generated by the management platform can be that the target training platform sends the model to the inference data space, and supports the single use of the target digital resource as the inference set data when inferring the model in the inference data space, ensuring that the target digital resource uses the target digital resource to complete the inference of the model when the usage permission permits.

[0015] In one possible implementation, before obtaining the target resource information input by the data user, it also includes: obtaining the target resource registration information input by the data provider, the target resource registration information includes target usage rights and digital resource information, and the digital resource information is used to indicate that the target digital resource is stored in the target data space.

[0016] It is understandable that the data provider can input the target resource registration information into the management platform, set the target usage rights of the target digital resources and the target data space for storing the target digital resources, so that the subsequent management platform can use the target digital resources to train or infer models within the scope of the target usage rights. At the same time, it can also determine the target data space for storing the target digital resources, so that the subsequent target training platform can obtain the target digital resources from the target data space.

[0017] In a possible implementation, the method further includes: if a permission termination operation of the data provider on the target digital resource is received, sending a deletion instruction to the target data space, wherein the deletion instruction is used to instruct the target data space to delete the target digital resource stored internally.

[0018] It is understandable that the data provider can terminate the permission operation to the management platform, issue a deletion instruction to the target data space, and notify the target data space to delete the target digital resources in the target data space. After the target usage permissions have been set for the target digital resources, the usage permissions granted to the target digital resources can be terminated to ensure the digital provider's control ability to cancel the usage permissions of the target digital resources, thereby enriching the management platform's control function over the usage permissions of the respective digital resources it manages.

[0019] In a possible implementation, the method further includes: receiving usage information of the target digital resource returned by the target data space, where the usage information of the target digital resource is used to indicate transmission and access status of the target digital resource; and storing the usage information of the target digital resource.

[0020] It is understandable that after the target digital resources stored in the target data space are accessed or transmitted, the usage information can be returned to the management platform, and the management platform will record the usage information in the form of a log, so that subsequent data users and / or data providers can query the log to determine the usage of the target digital resources, thereby improving the user experience of data users and / or data providers.

[0021] In a second aspect, an embodiment of the present application provides a digital resource management device, which is used to execute any one of the digital resource management methods provided in the first aspect.

[0022] In one possible implementation, the embodiment of the present application can divide the digital resource management device into functional modules according to the method provided in the first aspect above. For example, each functional module can be divided corresponding to each function, or two or more functions can be integrated into one processing module. Exemplarily, the embodiment of the present application can divide the digital resource management device into an acquisition module, a processing module, and a sending module, etc. according to the function. The description of the possible technical solutions and beneficial effects executed by each of the functional modules divided above can refer to the technical solutions provided by the first aspect above or its corresponding possible implementation method, and will not be repeated here.

[0023] In a third aspect, an embodiment of the present application provides a computing device, which includes a processor and a memory, wherein the processor is coupled to the memory; the memory is used to store computer instructions, which are loaded and executed by the processor to enable the computing device to implement the digital resource management method described in the above aspects.

[0024] In a fourth aspect, an embodiment of the present application provides a computing device cluster, which includes at least one computing device, each computing device including: a processor and a memory, the processor of at least one computing device being used to execute instructions stored in the memory of at least one computing device, so that the computing device cluster executes the digital resource management method provided in the various optional implementations of the above-mentioned first aspect.

[0025] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores at least one computer program instruction, and the computer program instruction is loaded and executed by a processor to implement the digital resource management method as described in the above aspects.

[0026] In a sixth aspect, embodiments of the present application provide a computer program product, comprising computer instructions stored in a computer-readable storage medium. A processor of a computing device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computing device cluster to perform the digital resource management method provided in various optional implementations of the first aspect.

[0027] For the specific descriptions of the second to sixth aspects and their various implementations in this application, reference can be made to the detailed descriptions in the first aspect and its various implementations; and for the beneficial effects of the second to sixth aspects and their various implementations, reference can be made to the analysis of the beneficial effects in the first aspect and its various implementations, which will not be repeated here.

[0028] These and other aspects of the present application will become more readily apparent from the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] FIG1 is a schematic diagram showing a digital resource management scenario according to an exemplary embodiment;

[0030] FIG2 is a schematic diagram showing an architecture of a management system in a cloud scenario according to an exemplary embodiment;

[0031] FIG3 is a flow chart showing a method for managing digital resources according to an exemplary embodiment;

[0032] FIG4 is a schematic diagram of a process of granting and withdrawing digital resource usage rights involved in the embodiment shown in FIG3 ;

[0033] FIG5 is a schematic diagram of a process of target digital resource management involved in the embodiment shown in FIG3 ;

[0034] FIG6 is a schematic diagram of a digital resource management process involved in the embodiment shown in FIG3 ;

[0035] FIG7 is a schematic structural diagram of a digital resource management device according to an exemplary embodiment;

[0036] FIG8 is a schematic diagram of a computing device according to an exemplary embodiment;

[0037] FIG9 is a schematic diagram showing a computing device cluster according to an exemplary embodiment;

[0038] FIG10 is a schematic diagram showing a connection mode between computing device clusters according to an exemplary embodiment. DETAILED DESCRIPTION

[0039] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0040] In this document, "plurality" refers to two or more. "And / or" describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. The character " / " generally indicates an "or" relationship between the associated objects.

[0041] Furthermore, in the description of this application, unless otherwise specified, "plurality" means two or more than two. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0042] In addition, in order to facilitate the clear description of the technical solutions of the embodiments of the present application, in the embodiments of the present application, words such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. Those skilled in the art will understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit differences. At the same time, in the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or design schemes. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a concrete way for easy understanding.

[0043] First, the nouns involved in the embodiments of this application are introduced.

[0044] Digital resources: These can be copyrighted digital resources such as books, audiovisual works, and works of art. These resources possess copyright value, indicating that they have a data provider (i.e., the owner), who has the authority to grant usage rights to the digital resource. The use of these digital resources may also involve compensation for the data provider's interests. Usage rights can include full or limited usage rights.

[0045] Among them, if the usage right of the target digital resource is full usage right, the data provider of the target digital resource grants multiple data users of the management platform the right to subscribe to and use the target digital resource multiple times, and feedback a compensation to the data provider; if the usage right of the target digital resource is limited usage right, the data provider of the target digital resource grants a single data user of the management platform the right to subscribe to and use the target digital resource once. After one subscription, the data provider needs to grant the management platform the right to subscribe again, and feedback a compensation to the data provider for each subscription to and use of the target digital resource.

[0046] Target Resource Information: Target resource information is used to indicate a target digital resource from at least one digital resource. Specifically, the target resource information may be input by a data user into the management platform and used to select one or more of the at least one digital resource managed by the management platform as the target digital resource.

[0047] Data usage policy: The data usage policy can be used to indicate how the target training platform uses the target digital resources when training models and / or reasoning with models. The data usage policy can be generated by the management platform based on the target usage rights of the target digital resources and sent to the target training platform so that the target training platform trains or reasons about the model according to the data usage policy.

[0048] Data space: The data space can be provided by a cloud service provider or a storage medium provided by a data provider. The data space only needs to meet the corresponding data specifications. The data specification can allow the data provider and the management space to access the data space according to the agreed identifier.

[0049] Then, the application scenarios of the embodiments of the present application are exemplarily introduced.

[0050] With the widespread use of AIGC-related applications, the AIGC models used in these applications often experience issues with unclear data usage permissions during pre-training, leading to actual usage exceeding the granted permissions. For example, in applications that automatically generate AI images, the AIGC models used in pre-training can use images obtained from the internet. Large numbers of images are obtained from the internet and accompanied by text descriptions, serving as the AIGC model training data. However, many of these images are works with permission. Without explicit authorization from the owners, using images from the internet for AIGC model training can lead to the issue of actual usage exceeding the granted permissions. This highlights the "opaqueness" of AIGC in pre-training datasets and the numerous issues of unclear usage permissions. Specifically, this includes the lack of transparency in the datasets used for pre-training AIGC models or AIGC applications. Users cannot clearly understand the authorization status of the datasets used in the pre-training process of the AIGC models they use, which leads to unclear usage rights due to unclear traceability. AIGC's deep learning algorithm is a "black box" model, and it is difficult to determine the use of usage rights during the traditional review process. Because the works generated by AIGC incorporate the characteristics of works with usage rights of other owners, the problem of tracing and managing the usage rights of works generated by AIGC is currently unresolved. During the AIGC model training process, technology may be used to grab undisclosed and unauthorized data, which exceeds the data usage rights.

[0051] In other words, the act of training an AIGC model is similar to the "thinking, absorbing, and re-creating" behavior of a natural person after reading a literary work or appreciating a work of art. This behavior doesn't correspond specifically to existing data usage rights, leading to uncertainty about whether the training data is authorized. Furthermore, the feasibility of authorizing training data is questionable, with issues such as excessive scale, unclear entities, and difficult mechanisms. AIGC model training involves a large number of works from diverse sources and ownership. Using prior authorization requires, on the one hand, precisely separating and extracting protected works from the massive amount of data; on the other hand, identifying the rights holders of each work with usage rights, negotiating authorization with them, and providing varying compensation. This process is lengthy, complex, and extremely difficult to implement. The significance of authorization remains to be evaluated, and it could potentially lead to negative effects such as "overfitting," a "chilling effect," and "model bias." In practice, any measure that limits the scale and availability of model training content could lead to unintended consequences, namely, increasing the probability that the model will simply output content that replicates the trained work.

[0052] In view of this, the embodiment of the present application constructs a management platform to manage at least one model training platform, and is connected to the data space of digital resources provided by multiple data providers. After receiving the target resource information input by the data user and the training platform information input by the data user, the management platform can determine the target digital resource and the target training platform. Since the target digital resource is set with a target usage permission, the management platform can generate a data usage policy based on the target usage permission and send the data usage policy to the target training platform so that the target training platform can use the target digital resource to train or infer the model in the target training platform according to the data usage policy. Since the management platform can set clear usage permissions for the digital resources provided by each data provider, after receiving the target resource information indicating the target digital resource input by the data user, the management platform can generate a data usage policy based on the target usage permission of the target digital resource, so that the target training platform can train or infer the model under the premise of complying with the target usage permission of the target digital resource, thereby ensuring the availability of the trained or inferred model.

[0053] For example, Figure 1 shows a schematic diagram of a digital resource management scenario provided by an embodiment of the present application. As shown in Figure 1, a digital resource management system 30 may include a management platform 31, a terminal on the data provider side, a target data space 34, a computing device running a target training platform 35, and a terminal on the data user side.

[0054] The terminal on the data provider side runs a first client 21, which is a client run by the management platform 31 on the data provider side. The terminal on the data user side runs a second client 22, which is a client run by the management platform 31 on the data user side.

[0055] The target data space 34 may be a connector space agreed upon by the terminal on the data provider side and the management platform 31. The target data space 34 is used to store the target digital resources uploaded through the first client 21. The first client 21 may also upload the authorization status of the received usage rights to the management platform 31 for recording. The terminal on the data user side may query the authorization status of the usage rights of each digital resource stored in the management platform 31 through the second client 22, and the data user may subscribe to the authorized digital resources as needed. The subscription operation for the target digital resources may be that the data user inputs the target resource information to the management platform. The management platform 31 may notify the target data space 34 based on the received subscription operation and send the target digital resources stored therein to the target training platform 35. The target training platform 35 may use the target digital resources to perform model training or model inference on the AIGC model.

[0056] As shown in Figure 2, the cloud management platform 20, that is, the management platform is used to manage the infrastructure 1. The infrastructure 1 includes cloud data center clusters set up in multiple regions. The exemplary multiple regions include region 10, region 11, and region 12. Each region is provided with a cloud data center cluster (not shown in the figure). Each cloud data center includes multiple cloud data centers. The exemplary cloud data center in region 10 includes cloud data center 101, cloud data center 102, and cloud data center 103. The cloud data center cluster located in region 11 includes cloud data center 111, cloud data center 112, and cloud data center 113. The cloud data center cluster located in region 13 includes cloud data center 121, cloud data center 122, and cloud data center 123. Each cloud data center includes multiple servers. The exemplary data center 101 includes server 1011, server 1012...

[0057] Continuing with Figure 2, a cloud management platform 20 is used to manage network infrastructure 1. A data provider, as a tenant, connects to the internet via a first client 21 and logs in to the cloud management platform 20 using an account pre-registered with the data provider. A data user connects to the internet via a second client 22 and logs in to the cloud management platform 20 using an account pre-registered with the data user. The cloud management platform 20 provides a configuration interface. The data provider accesses the configuration interface via the first client 21 and enters a resource request 1 on the configuration interface based on a resource change template provided by the cloud management platform. The data user accesses the configuration interface via the second client 22 and enters a resource request 2 on the configuration interface based on a resource change template provided by the cloud management platform. The cloud management platform 20 retrieves the resource request 1 from the configuration interface and orchestrates the corresponding target cloud resources based on the resource request 1. The data provider and the data user are users of the cloud management platform 20.

[0058] For example, in a public cloud scenario, each cloud data center can be identified by an availability zone (AZ). Different availability zones correspond to different cloud data centers, and cloud data centers can be located in one or more computer rooms. In a private cloud scenario, the infrastructure is managed by the user and includes at least one cloud data center. Each cloud data center includes at least one server, and cloud resources run on the at least one server. In a hybrid cloud scenario, the public cloud infrastructure is managed by a cloud management platform provided by the cloud resource provider, while the private cloud infrastructure is managed by the user.

[0059] Exemplarily, the client 21 and the client 22 may be terminal devices such as mobile phones with Internet access capabilities, personal computers, personal digital assistants, thin clients, vehicle-mounted hosts, or other terminal devices with Internet access capabilities.

[0060] It is worth noting that the public cloud application scenario is only one application scenario of the embodiment of the present application. The embodiment of the present application can also be applied to various scenarios such as hybrid cloud and private cloud, and the embodiment of the present application can also be implemented without limitation.

[0061] It should be noted that the application scenarios and system architectures described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Ordinary technicians in this field can know that with the evolution of the system architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0062] For ease of understanding, the digital resource management method provided in this application is exemplarily introduced below with reference to the accompanying drawings. The digital resource management method is applicable to the management platform shown in FIG1 .

[0063] FIG3 shows a flow chart of a digital resource management method provided by an exemplary embodiment of the present application. The digital resource management method can be executed by a management platform, which can be run on a computing device, such as the management platform 31 shown in FIG1 . The digital resource management method includes the following steps:

[0064] S101: The management platform obtains target resource information input by the data user.

[0065] The target resource information may be used to indicate a target digital resource in at least one digital resource, and the target digital resource is provided with a target usage permission.

[0066] That is, the target resource information input by the data user and received by the management platform can be used to enable the management platform to determine the target digital resource from various digital resources under management according to the target resource information.

[0067] For example, if the target resource information includes characteristic information, after receiving the target resource information, the management platform searches for a digital resource that matches the characteristic information included in the target resource information from various digital resources as the target digital resource. If the target resource information includes a subscription operation, the digital resource indicated by the subscription operation is used as the target digital resource.

[0068] Exemplarily, the resource management system may include a background server of a management platform, a first client of the management platform, a second client of the management platform, a target data space, and other data spaces. The first client is a client of the management platform running on a computing device used by a data provider of a digital resource. The first client is used to obtain target resource registration information of the digital resource provided by the data provider of the digital resource. The target resource registration information may include target usage rights and digital resource information. The digital resource information is used to indicate that the target digital resource is stored in the target data space. The first client may also allow the data provider of the digital resource to upload digital resources for which usage rights are provided to the corresponding data space. The target digital resources uploaded by the first client and for which usage rights are provided by the data provider can be uploaded to the target data space through the first client. The target data space may be a connector space for storing target digital resources for which usage rights are granted.

[0069] Among them, the first client is the client of the resource management system corresponding to the data provider of the target digital resource, the authorization operation can be used to instruct the data provider of the target digital resource to grant the resource management system the right to use the target digital resource, and the target data space can be a connector space for storing the target digital resource.

[0070] In a possible implementation, before obtaining the target resource information input by the data user, the management platform may obtain the target resource registration information input by the data provider.

[0071] The target resource registration information includes target usage rights and digital resource information. The digital resource information can be used to indicate that the target digital resource is stored in the target data space.

[0072] That is to say, in the process of registering the target digital resources, the data provider provides the target usage rights and target data space to the management platform, so that the management platform can subsequently clarify the target usage rights of the target digital resources, and the management platform can store the target digital resources in the target data space.

[0073] In a possible implementation, in response to receiving, through the first client, an authorization operation of the data provider for the target digital resource, the management platform may determine a target data space in which the target digital resource needs to be stored.

[0074] That is to say, when the resource management system receives an authorization operation for the target digital resource through the first client on the data provider side, the first client can send instructions corresponding to the authorization operation to the background server of the management platform. After the background server receives the instructions corresponding to the authorization operation, it can determine the target data space and upload the target digital resource received through the first client to the target data space.

[0075] In addition, target digital resources can include copyrighted books, audiovisual works, works of art, and other digital resources with copyright value. Copyrighted digital resources can indicate that the digital resource has a data provider, who has the authority to grant usage rights to the digital resource. The use of the digital resource may also involve compensation for the data provider's interests. In other words, the user can only use the digital resource to train or infer generative AI models after the data provider grants usage rights to the user.

[0076] In a possible implementation, the authorization operation on the target digital resource may include at least one of a full authorization operation on the target digital resource or a limited authorization operation on the target digital resource. The authorization operation may be an operation performed by the data provider when setting the target usage rights of the target digital resource.

[0077] Among them, the full authorization operation for the target digital resource can be used to instruct the data provider to grant multiple users of the management platform the right to subscribe to and use the target digital resource multiple times, and to provide compensation to the data provider. The limited authorization operation for the target digital resource is used to instruct the data provider to grant a data user of the management platform the right to subscribe to and use the target digital resource once. After the subscription is completed, the data provider needs to grant the management platform the right to subscribe to and use the target digital resource once again, and compensation will be provided to the data provider for each subscription to and use of the target digital resource.

[0078] That is to say, after the full authorization operation on the target digital resource, the target usage right of the target digital resource is the full usage right. The data user can determine the authorization operation on the target digital resource through the first client to grant multiple data users of the management platform the right to subscribe to and use the target digital resource multiple times, and each user who subscribes to the target digital resource can be compensated by the management platform.

[0079] For example, a fully authorized operation on a target digital resource can mean that the copyright data provider only needs to authorize it once, and then it can be used multiple times by all data users, that is, subscribers. The compensation brought to the data provider by the use of the target digital resource can be settled according to a pre-agreed fixed compensation, which is unrelated to the number of times the target digital resource is used by the subscriber or the effectiveness of the use. The target digital resource for which a fully authorized operation is performed can be pre-agreed in advance to grant compensation to the data provider indefinitely, or it can be granted compensation to the data provider in a one-time unit based on time. For example, after granting a fully authorized operation on the target digital resource, the data provider can obtain a single pre-agreed compensation, which is unrelated to the number of data users who subscribe to the target digital resource or the effectiveness of the use; the data provider can also obtain a pre-agreed compensation per unit time after granting a fully authorized operation on the target digital resource. If there are data users who subscribe to the target digital resource within a unit time, the management platform can allocate the pre-agreed compensation corresponding to the unit time to the data provider. Similarly, this compensation is unrelated to the number of data users who subscribe to the target digital resource or the effectiveness of the use.

[0080] Alternatively, after the restricted authorization operation on the target digital resource, the target usage right of the target digital resource is a restricted usage right. The data user may determine through the first client that the authorization operation on the target digital resource is to grant a single user of the management platform the right to subscribe to and use the target digital resource once. After one subscription, the data provider needs to grant the management platform the right to subscribe again, and each time the management platform determines to subscribe to and use the target digital resource once, it can provide compensation to the data provider.

[0081] For example, a restricted authorization operation on a target digital resource may refer to a copyright data provider authorizing the target digital resource once each time, which can be used by a single data user, that is, a single subscriber, and the compensation for the use of the target digital resource to the data provider can be settled according to the pre-agreed compensation for this time. A restricted authorization operation on a target digital resource may be compensation to the data provider in accordance with a single agreement for each authorization. For example, a restricted authorization operation on a target digital resource may be a restricted authorization for a certain subscription application, and the restricted authorization may be pre-agreed with a subscription usage period, that is, the restricted authorization for a certain subscription application may be valid for the subscription usage period, and when the management platform receives other subscription applications or subscribes again, it is necessary to apply for restricted authorization from the data provider of the target digital resource again.

[0082] Generally speaking, the target digital resources for which a data provider applies for restricted authorization for a single subscription application include relatively sensitive digital resources. To protect the use rights of the target digital resources as much as possible, the target digital resources for which restricted authorization is subsequently granted may not participate in the pre-training process of the generative AI model. In other words, the target digital resources for which restricted authorization is granted may not be integrated into the pre-training or fine-tuning process of the model. The target digital resources for which restricted authorization is granted may be allowed to participate in model reasoning in the form of context.

[0083] For example, after logging into the first client, data provider 1 can upload the target digital resource to the management platform through the first client, and data provider 1 can trigger the operation by the authorization operation selection control displayed on the first client, and select the required authorization operation, that is, full authorization operation or limited authorization operation. The first client will notify the backend server of the management platform of the selected authorization operation and store the authorization status of the target digital resource in the backend server for subsequent operations such as searching, modifying, and deleting the authorization status of the target digital resource.

[0084] In a possible implementation manner, the first client applies for a target data space corresponding to the first client, wherein the target data space has a unique identity.

[0085] Among them, the target data space can be a connector space provided by the publisher of the target digital resources. The connector space provided by the publisher of the target digital resources can be a connector applied for by the publisher of the target digital resources through the digital asset space. The connector can be an authentication and encryption mechanism pre-agreed between the publisher of the target digital resources and the backend server of the management platform, or a data space environment that meets specific specifications, and each connector has a unique identity to ensure the security of the digital resources in the connector.

[0086] Specifically, the target data space may be a connector space that complies with the IDS technical specifications published by the International Data Spaces Association (IDSA).

[0087] For example, FIG4 is a flowchart of the granting and withdrawal of digital resource usage rights involved in an embodiment of the present application. As shown in FIG4 , the management platform may include two modules: a usage rights registration center and a data directory market. After logging in through the first client, the data provider can choose to grant a usage rights license through the usage rights registration center module. The choice of granting a usage rights license can be based on the value or sensitivity of the digital resource. Different licensing methods can be selected. Alternatively, the data provider can choose to withdraw from or cancel the usage rights license for the authorized digital resource, thereby removing the published data from the shelves. In addition, the data provider or data user can also query whether the digital resource is being used and the usage or compensation status through the data directory market module. The data provider of the digital resource can upload the digital resource to different databases of the target data space for storage according to the selected authorization situation. For example, a fully authorized digital resource can be converted into a vector and uploaded to the database of authorized data, a digital resource with limited authorization can be converted into a vector and uploaded to the database of sensitive data, and a digital resource with open permissions can be converted into a vector and uploaded to the database of open data for storage.

[0088] Among them, the relevant information of digital resources recorded in the usage rights registration center may include the title of the digital resource, the name of the author, the description of the work, the content of the work, etc.; it may also include the information of the data provider, the ownership certificate of usage rights, the statement of granting usage rights, etc.; it may also include the watermark information of the digital resource, the identification information of the allocated data space, the identification information of the digital resource, and the current authorization status information. The data catalog market module can record the statistical information of the subscription status of the digital resource and the subscription application record of the digital resource. In other words, the data provider of the usage rights can actively claim, grant and cancel the usage rights through the management platform. The management platform can check the information submitted by the owner of the usage rights and register it. The data user can find the required digital resources as a data set through the data catalog market and subscribe for further use.

[0089] In a possible implementation, after obtaining the target resource information input by the data user, the management platform may determine the target digital resource according to the target resource information.

[0090] For example, if the target resource information is feature information, after obtaining the feature information, the management platform determines the digital resource matching the feature information as the target digital resource.

[0091] In a possible implementation manner, the target digital resource uploaded by the first client is received through the target data space.

[0092] In an embodiment of the present application, after receiving the target digital resource uploaded by the data provider, the first client sends the target digital resource to the target data space of the resource management system. The target data space can be a connector space provided by the first client or a connector space already existing in the resource management system.

[0093] In one possible implementation, after receiving the target digital resource, the target data space can process the target digital resource, perform relevant classification, cleanup, etc. on the data, to ensure that when the target digital resource is subsequently used, the resource management system can accurately find and obtain the target digital resource stored in the target data space.

[0094] In one possible implementation, after the target digital resources are received in the target data space, they can be classified and stored according to the authorization status corresponding to the target digital resources. That is, the target data space may include storage space corresponding to open digital resources, storage space corresponding to fully authorized digital resources, or storage space corresponding to restricted authorized digital resources.

[0095] For example, if the target digital resource is a fully authorized digital resource, the target digital resource can be stored in the storage space corresponding to the fully authorized digital resource; if the target digital resource is a digital resource with limited authorization, the target digital resource can be stored in the storage space corresponding to the digital resource with limited authorization; if the target digital resource is a public digital resource with no copyright interest, the target digital resource can be stored in the storage space corresponding to the open digital resource.

[0096] In one possible implementation, after receiving the target digital resource, the first client may convert the target digital resource into vector data and send the converted vector data to the target data space. Specifically, for more sensitive digital resources, such as digital resources subject to restricted authorization operations, the digital resources subject to restricted authorization operations may be converted into embedded vector data, and the converted embedded vector data is then sent to the target data space for storage.

[0097] In a possible implementation, the resource management system may display relevant information of the target digital resource through the second client.

[0098] The relevant information of the target digital resource can be used to prompt the data user whether to subscribe to the target digital resource; the relevant information can include at least one of the following: identification information of the digital resource, the rights granted to the digital resource, and the subscription status of the digital resource. The second client is the data user, that is, the client of the resource management system running on the subscriber side.

[0099] That is to say, after receiving the uploaded target digital resources, the target data space can send the relevant information of the target digital resources to the background server of the resource management system. The background server integrates the relevant information of the target digital resources so that the subscriber can query and obtain the relevant information of the target digital resources from the background server by logging into the second client.

[0100] For example, if data provider 1 performs a restricted authorization operation on digital resource A through the first client, the digital resource A can be uploaded to the connector space agreed upon by the first client and the back-end server through the first client. After the connector space receives the digital resource A, the first client can send the relevant information of digital resource A to the back-end server of the resource management system, which will be counted by the back-end server, so that user 1 can query the back-end server of the resource management system through the second client, thereby obtaining relevant information including digital resource A managed by the resource management system.

[0101] In a possible implementation, in response to receiving a suspend permission operation on the target digital resource through the first client, the relevant information of the target digital resource displayed through the second client is deleted.

[0102] Among them, the suspend permission operation can be a trigger operation on the authorization suspend control displayed by the first client. After completing the authorization operation on the target digital resource through the first client, the first client can upload the relevant information of the target digital resource to the background server, so that the user can query the background server through the second client to obtain the relevant information of the target digital resource, so that the user can select a suitable target digital resource to subscribe. In addition to performing authorization operations on target digital resources, the data provider can also suspend permission operations on authorized target digital resources. Specifically, the authorization suspend control of the authorized target digital resource can be displayed through the first client. After receiving the trigger operation of the authorization suspend control of the target digital resource, the first client can notify the background server of the resource management system to delete or remove the relevant information of the target digital resource, so that the user cannot query the relevant information of the target digital resource through the second client, thereby avoiding the situation where the user subscribes to the target digital resource for which authorization has been suspended.

[0103] S102: The management platform obtains the training platform information input by the data user.

[0104] In an embodiment of the present application, a data user inputs training platform information to the management platform, so that the management platform determines a target training platform.

[0105] The training platform information is used to indicate a target training platform in at least one model training platform.

[0106] For example, since the management platform can manage multiple model training platforms, the models to be trained or inferred in each model training platform may be different, and the environment for training or inferring the models may also be different. Therefore, different data users may need to use different platforms for training models using target digital resources. The training platform information input by the data user may include the characteristics of the model training platform that the data user needs to select. The management platform determines the target training platform that meets the needs of the data user according to the training platform information.

[0107] S103: The management platform generates a data usage policy based on the target usage rights.

[0108] The data usage policy is used to indicate how the target training platform uses target digital resources when training models and / or model inference.

[0109] In one possible implementation, if the target usage permission is full usage permission, the data usage policy supports multiple uses of the target digital resource as training set data when the target training platform trains a model. If the target usage permission is limited usage permission, the data usage policy supports single use of the target digital resource as inference set data when the target training platform performs model inference.

[0110] That is, in one case, the management platform can generate a data usage policy based on full usage permissions. In this case, the data usage policy is used to instruct the target training platform to use the target digital resource as training set data multiple times when training the model. In another case, the management platform can generate a data usage policy based on limited usage permissions. In this case, the data usage policy is used to instruct the target training platform to transfer the model to the inference data space and use the target digital resource as inference set data once when performing model inference in the inference data space. The inference data space is a data space in a controlled environment.

[0111] For example, the management platform may pre-store the correspondence between usage rights and data usage policies. After determining the target usage rights of the target digital resource, the management platform may determine the data usage policy corresponding to the target usage rights according to the correspondence.

[0112] S104: The management platform sends the data usage policy to the target training platform.

[0113] In an embodiment of the present application, the management platform can send the data usage policy corresponding to the target digital resource to the target training platform, so that the target training platform can subsequently use the target digital resource in accordance with the data usage policy, thereby ensuring that the use of the target digital resource does not exceed the scope specified by the target usage authority.

[0114] In a possible implementation, in response to receiving a subscription operation for the target digital resource through the second client, the resource management system sends the target digital resource to the target training platform through the target data space.

[0115] In an embodiment of the present application, a user can query digital resources with various authorization statuses through a second client. If a subscription operation for a target digital resource is received through the second client, the second client can send a subscription notification to the backend server. The backend server sends a corresponding subscription instruction to the corresponding target data space according to the subscription notification. The target data space can send the target digital resource to a target training platform according to the subscription instruction, so that the target training platform can use the target digital resource to perform model training on the neural network model that the subscriber needs to train.

[0116] The second client may be a client of a resource management system corresponding to the user, and the target training platform may be a platform for updating a neural network model using digital resources.

[0117] In a possible implementation, if the authorization operation on the target digital resource includes a full authorization operation, in response to receiving a subscription operation on the target digital resource through the second client, the target digital resource is sent to the target training platform.

[0118] Among them, the target training platform can be used to train the neural network model using the target digital resources as training set data.

[0119] That is to say, if the authorization status of the target digital resource is fully authorized, after the second client receives the subscription operation for the target digital resource, the second client can send a subscription notification to the backend server, which includes at least the identification information of the target digital resource. The backend server determines the corresponding target data space according to the identification information of the target digital resource, and then sends a subscription instruction to the target data space. The subscription instruction includes the authorization status of the target digital resource, which is used to enable the target data space to send the target digital resource to the target training platform according to the authorization status, so that the target training platform can use the subscribed target digital resource as a training set to perform model training on the neural network model (for example, AIGC model). In this case, the target training platform can be an open training platform.

[0120] In addition, the backend server can record relevant information about subscription operations for target digital resources, including but not limited to the number of subscriptions to the digital resource, the time of each subscription, subscriber information, etc. The relevant information about the subscription operation can be queried by the data provider, thereby ensuring that the data provider has a clear understanding of the usage of all its digital resources.

[0121] In one possible implementation, if the authorization operation on the target digital resource includes a restricted authorization operation, in response to receiving a subscription operation on the target digital resource through the second client, an inference data space is determined, and the target digital resource is sent to the inference data space.

[0122] The inference data space may be a connector space for performing contextual reasoning on a neural network model using target digital resources as inference set data.

[0123] Since the target digital resource for restricted authorization is generally a more sensitive digital resource, if the authorization status of the target digital resource is restricted authorization, after the second client receives the subscription operation for the target digital resource, the second client can send a subscription notification to the background server, and the subscription notification includes at least the identification information of the target digital resource. The background server determines the corresponding target data space according to the identification information of the target digital resource, and then sends a subscription instruction to the target data space, and the subscription instruction includes the authorization status of the target digital resource, which is used to make the target data space send the target digital resource to the inference data space. The background server can send a data usage policy to the target training platform, so that the target training platform sends a neural network model to the inference data space according to the data usage policy, so that the inference data space uses the subscribed target digital resource as an inference set to perform model context reasoning on the neural network model (for example, AIGC model). In this case, the inference data space can be in a controlled environment, and the controlled environment can refer to that the inference data space is a connector space that is not completely open.

[0124] For example, FIG5 is a flow chart of a target digital resource management process involved in an embodiment of the present application. As shown in FIG5 , a data provider can input target resource registration information into a management platform (S21). The target resource registration information includes the target data space where the target digital resource is stored and the target usage rights of the target digital resource. The data provider uploads the target digital resource and stores it in the target data space. The management platform can record the target data space where the data provider's target digital resource is stored and the target usage rights based on the target resource registration information (S22). A data user can input target resource information into the management platform (S23). The target resource information can indicate the digital resource that the data user needs to select. The management platform can determine the target digital resource selected by the data user and the target usage rights corresponding to the target digital resource based on the target resource information (S24). The data user can also input training platform information into the management platform (S25). The training platform information can be used to indicate the target training platform selected by the data user. The management platform can determine the target training platform selected by the data user based on the training platform information (S26). The management platform generates a data usage policy based on the determined target usage rights (S27). The management platform can then send the data usage policy to the target training platform (S28). At the same time, the management platform can also send a digital resource transmission instruction to the target data space (S29). The digital resource transmission instruction can be divided into two types according to different data usage policies. If the data usage policy is a data usage policy corresponding to full usage authority, the digital resource transmission instruction is used to instruct the target data space to send the target digital resource to the target training platform; if the data usage policy is a data usage policy corresponding to limited usage authority, the digital resource transmission instruction is used to instruct the target data space to send the target digital resource to the inference data space. If the data usage policy is a data usage policy corresponding to full usage authority, the target training platform can use the target digital resource as training set data for multiple training models (S210). If the data usage policy is a data usage policy corresponding to limited usage authority, the neural network model is sent to the inference data space (S211), and the inference data space uses the target digital resource as inference set data for model inference once (S212).

[0125] In addition, after receiving a subscription operation for a target digital resource with a restricted authorization status and sending the target digital resource to the subscriber for use, the backend server may delete or remove the relevant information of the target digital resource from the shelf to prevent the subscriber from subscribing to the target digital resource again. The backend server may also update the compensation information for the data provider of the target digital resource, which the data provider can query through the first client.

[0126] In one possible implementation, a subscription option corresponding to the target digital resource is displayed through the second client, and in response to receiving a subscription operation for the subscription option corresponding to the target digital resource through the second client, the target digital resource is sent to the target training platform.

[0127] The subscription option may be used to indicate that the corresponding digital resource supports being subscribed to by the user.

[0128] That is, the authorized target digital resource may display a corresponding subscription option in the second client, and the subscription operation may be a triggering operation on a control corresponding to the subscription option.

[0129] In a possible implementation, if a data provider's permission termination operation for a target digital resource is received, a deletion instruction is sent to the target data space, where the deletion instruction is used to instruct the target data space to delete the target digital resource stored therein.

[0130] Among them, the suspend permission operation can be a trigger operation on the authorization suspend control displayed by the first client. After the authorization operation on the target digital resource is completed through the first client, the first client can upload the relevant information of the target digital resource to the background server, so that the user can query the background server through the second client to obtain the relevant information of the target digital resource, so that the user can choose a suitable target digital resource to subscribe. In addition to performing authorization operations on target digital resources, data providers can also suspend permission operations on authorized target digital resources. Specifically, the authorization suspend control of the authorized target digital resource can be displayed through the first client. After receiving the trigger operation of the authorization suspend control of the target digital resource, the first client can notify the background server of the resource management system to delete or remove the relevant information of the target digital resource, and can also delete the target digital resource whose authorization has been suspended from the first data control space. To a certain extent, the storage pressure on digital resources can be reduced.

[0131] In a possible implementation, the management platform may receive usage information of the target digital resource returned by the target data space, where the usage information of the target digital resource is used to indicate the transmission and access status of the target digital resource; and store the usage information of the target digital resource.

[0132] That is to say, the management platform can receive the usage information of the target digital resources returned by the target data space, and the usage information of the target digital resources is used to indicate the transmission and access status of the target digital resources; the usage information of the target digital resources is stored in the form of logs; wherein, the usage information of the target digital resources stored in the form of logs supports queries by data users and / or data providers.

[0133] For example, Figure 6 is a schematic diagram of a digital resource management process involved in an embodiment of the present application. As shown in Figure 6, the registration center is the module where digital resource data providers or digital resource providers complete dataset registration and rights or copyright authentication. Digital resource data providers or data providers upload digital resources to the provider's connector space, i.e., the target data space, or provide an existing connector space to complete registration. Digital resource providers can perform relevant classification and cleanup on the data. Data sets are listed and published on the data catalog marketplace, where digital resources can then be searched and subscribed to. Data set users can subscribe to data usage rights through the data catalog marketplace and receive relevant data usage agreements. The data usage control center can issue usage policy files corresponding to subscription requests to authorize access to the data files. Data files can be transferred using agreed secure transmission protocols, channels, and tools. Subscribed data can be loaded onto a certified large model pre-training platform to complete large model pre-training or incremental model training. After training and tuning a large model based on a specific dataset, a model file is output, which contains compressed and stored knowledge specific to the dataset. The trained, specified version of the model file can be deployed to the model's inference environment and made available externally via an API. Model-based APIs or AI applications provide users with model reasoning services. The model's reasoning process utilizes unique sensitive data, and sensitive data is provided within the context of reasoning. The model's reasoning service is not fully open, but rather controlled: for specific users and specific purposes. Data files' end-to-end process operations, access operations, usage, and other related traceability and usage information are stored in logs, and verification services are provided through specific interfaces. Connectors can regularly record data access details and report them to the audit and tracing center, which can then conduct compensation confirmations and conduct audits and backtracking of exceptions.

[0134] In summary, by constructing a management platform that manages at least one model training platform and is connected to the data space storing digital resources provided by multiple data providers, the management platform can determine the target digital resource and the target training platform after receiving the target resource information input by the data user and the training platform information input by the data user. Since the target digital resource is set with a target usage permission, the management platform can generate a data usage policy based on the target usage permission and send the data usage policy to the target training platform so that the target training platform can use the target digital resource to train or infer the model in the target training platform according to the data usage policy. Since the management platform can set clear usage permissions for the digital resources provided by each data provider, after receiving the target resource information indicating the target digital resource input by the data user, the management platform can generate a data usage policy based on the target usage permission of the target digital resource, so that the target training platform can train or infer the model under the premise of complying with the target usage permission of the target digital resource, thereby ensuring the availability of the trained or inferred model.

[0135] The above mainly introduces the scheme of the embodiment of the present application from the perspective of method. It can be understood that in order to realize the above functions, the digital resource management device includes at least one of the hardware structure and software modules corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0136] The embodiment of the present application can divide the digital resource management device into functional units according to the above method example. For example, each functional unit can be divided corresponding to each function, or two or more functions can be integrated into one processing unit. The above integrated unit can be implemented in the form of hardware or in the form of software functional units. It should be noted that the division of units in the embodiment of the present application is schematic and is only a logical functional division. There may be other division methods in actual implementation.

[0137] For example, FIG7 shows a schematic diagram of the structure of a digital resource management device 500 provided by an exemplary embodiment of the present application. The digital resource management device is applied to a management platform, which is used to manage at least one model training platform. The management platform is connected to multiple data spaces, among which there is a target data space, which stores at least one digital resource of a data provider. The digital resource management device 500 includes:

[0138] An acquisition module 510 is configured to acquire target resource information input by a data user, wherein the target resource information is used to indicate a target digital resource among the at least one digital resource, wherein the target digital resource is provided with a target usage permission;

[0139] The acquisition module 510 is further configured to acquire training platform information input by the data user, wherein the training platform information is configured to indicate a target training platform in the at least one model training platform;

[0140] A processing module 520 is configured to generate a data usage policy based on the target usage permission, wherein the data usage policy is used to indicate how the target training platform uses the target digital resource when training a model and / or performing model inference.

[0141] The sending module 530 is configured to send the data usage policy to the target training platform.

[0142] For example, in conjunction with FIG3 , the acquisition module 510 may be used to execute S101 and S102 as shown in FIG3 , the processing module 520 may be used to execute S103 as shown in FIG3 , and the sending module 530 may be used to execute S104 as shown in FIG3 .

[0143] In one possible implementation, if the target usage right is full usage right, the processing module 520 is also used to generate the data usage policy based on the full usage right, and the data usage policy is used to instruct the target training platform to use the target digital resource as training set data multiple times when training the model.

[0144] In one possible implementation, if the target usage right is a restricted usage right, the processing module 520 is also used to generate the data usage policy based on the restricted usage right, and the data usage policy is used to instruct the target training platform to use the target digital resource as the inference set data during model inference.

[0145] In one possible implementation, the multiple data spaces include an inference data space, and the data usage policy is further used to instruct the model to be transferred to the inference data space, and to use the target digital resource as inference set data when performing model inference in the inference data space.

[0146] In one possible implementation, the acquisition module 510 is also used to obtain the target resource registration information input by the data provider before obtaining the target resource information input by the data user. The target resource registration information includes target usage rights and digital resource information. The digital resource information is used to indicate that the target digital resource is stored in the target data space.

[0147] In a possible implementation, the apparatus further includes:

[0148] The deletion module is configured to send a deletion instruction to the target data space upon receiving a permission termination operation from the data provider on the target digital resource, wherein the deletion instruction is used to instruct the target data space to delete the target digital resource stored therein.

[0149] In a possible implementation, the apparatus further includes:

[0150] The storage module is used to receive the usage information of the target digital resource returned by the target data space, wherein the usage information of the target digital resource is used to indicate the transmission and access status of the target digital resource; and store the usage information of the target digital resource.

[0151] For the detailed description of the above optional methods, please refer to the above method embodiments, which will not be repeated here. In addition, the explanation and beneficial effects of any of the above digital resource management devices can be referred to the above corresponding method embodiments, which will not be repeated here.

[0152] The acquisition module 510, processing module 520, and sending module 530 can all be implemented in software or hardware. For example, the implementation of acquisition module 510 will be described below using acquisition module 510 as an example. Similarly, the implementation of processing module 520 and sending module 530 can refer to the implementation of acquisition module 510.

[0153] As an example of a software functional unit, the acquisition module 510 may include code running on a computing instance. The computing instance may include at least one of a physical host (computing device), a virtual machine, and a container. Furthermore, the computing instance may be one or more. For example, the acquisition module 510 may include code running on multiple hosts / virtual machines / containers. It should be noted that the multiple hosts / virtual machines / containers used to run the code may be distributed in the same region or in different regions. Furthermore, the multiple hosts / virtual machines / containers used to run the code may be distributed in the same availability zone (AZ) or in different AZs, each AZ including one data center or multiple geographically close data centers. Typically, a region may include multiple AZs.

[0154] Similarly, multiple hosts / virtual machines / containers running the code can be distributed within the same virtual private cloud (VPC) or across multiple VPCs. Typically, a VPC is set up within a region. Cross-region communication between two VPCs within the same region, or between VPCs in different regions, requires a communication gateway within each VPC to interconnect the VPCs.

[0155] As an example of a hardware functional unit, the acquisition module 510 may include at least one computing device, such as a server. Alternatively, the acquisition module 510 may be implemented using an application-specific integrated circuit (ASIC) or a programmable logic device (PLD). The PLD may be a complex programmable logical device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0156] The multiple computing devices included in acquisition module 510 can be distributed in the same region or in different regions. The multiple computing devices included in acquisition module 510 can be distributed in the same AZ or in different AZs. Similarly, the multiple computing devices included in acquisition module 510 can be distributed in the same VPC or in multiple VPCs. The multiple computing devices can be any combination of servers, ASICs, PLDs, CPLDs, FPGAs, GALs, and other computing devices.

[0157] It should be noted that, in other embodiments, the acquisition module 510 can be used to execute any step in the digital resource management method, the processing module 520 can be used to execute any step in the digital resource management method, and the sending module 530 can be used to execute any step in the digital resource management method. The steps that the acquisition module 510, the processing module 520, and the sending module 530 are responsible for implementing can be specified as needed. By having the acquisition module 510, the processing module 520, and the sending module 530 respectively implement different steps in the digital resource management method, the full functionality of the digital resource management device is achieved. This application also provides a computing device 100. As shown in Figure 8, the computing device 100 includes: a bus 102, a processor 104, a memory 106, and a communication interface 108. The processor 104, the memory 106, and the communication interface 108 communicate with each other via the bus 102. The computing device 100 can be a server or a terminal device. It should be understood that this application does not limit the number of processors and memories in the computing device 100.

[0158] Bus 102 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, among others. Buses may be classified as address buses, data buses, control buses, and the like. For ease of illustration, FIG8 illustrates a single bus line, but this does not imply a single bus or type of bus. Bus 102 may include a path for transmitting information between various components of computing device 100 (e.g., memory 106, processor 104, and communication interface 108).

[0159] The processor 104 may include any one or more processors such as a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).

[0160] The memory 106 may include volatile memory, such as random access memory (RAM). The processor 104 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0161] The memory 106 stores executable program code, and the processor 104 executes the executable program code to respectively implement the functions of the aforementioned acquisition module 510, processing module 520, and sending module 530, thereby implementing the digital resource management method. In other words, the memory 106 stores instructions for executing the digital resource management method.

[0162] Alternatively, the memory 106 stores executable codes, and the processor 104 executes the executable codes to respectively implement the functions of the aforementioned digital resource management apparatus, thereby implementing the digital resource management method.

[0163] The communication interface 103 uses a transceiver module such as, but not limited to, a network interface card or a transceiver to implement communication between the computing device 100 and other devices or a communication network.

[0164] Embodiments of the present application also provide a computing device cluster. The computing device cluster includes at least one computing device. The computing device can be a server, such as a central server, an edge server, or a local server in a local data center. In some embodiments, the computing device can also be a terminal device such as a desktop computer, a laptop computer, or a smartphone.

[0165] As shown in Figure 9, the computing device cluster includes at least one computing device 100. The memory 106 in one or more computing devices 100 in the computing device cluster may store the same instructions for executing the digital asset management method.

[0166] In some possible implementations, the memory 106 of one or more computing devices 100 in the computing device cluster may also store some instructions for executing the digital asset management method. In other words, the combination of one or more computing devices 100 can jointly execute the instructions for executing the digital asset management method.

[0167] It should be noted that the memory 106 in different computing devices 100 in the computing device cluster can store different instructions, each for executing a portion of the functions of the digital resource management apparatus. In other words, the instructions stored in the memory 106 in different computing devices 100 can implement the functions of one or more of the acquisition module 510, the processing module 520, and the sending module 530.

[0168] In some possible implementations, one or more computing devices in a computing device cluster may be connected via a network. The network may be a wide area network or a local area network, etc. FIG10 shows a possible implementation. As shown in FIG10 , two computing devices 100A and 100B are connected via a network. Specifically, the connection to the network is made via a communication interface in each computing device. In this type of possible implementation, the memory 106 in the computing device 100A stores instructions for executing the functions of the acquisition module 510. At the same time, the memory 106 in the computing device 100B stores instructions for executing the functions of the processing module 520 and the sending module 530.

[0169] The connection method between the computing device clusters shown in Figure 10 can be considered to be that the digital resource management method provided in this application requires a large amount of data storage and calculation data, so it is considered to entrust the functions implemented by the processing module 520 and the sending module 530 to the computing device 100B for execution.

[0170] It should be understood that the functions of the computing device 100A shown in FIG10 may also be completed by multiple computing devices 100. Similarly, the functions of the computing device 100B may also be completed by multiple computing devices 100.

[0171] The present application also provides another computing device cluster. The connection relationship between the computing devices in this computing device cluster can be similar to the connection method of the computing device cluster described in Figures 9 and 10. However, the memory 106 of one or more computing devices 100 in this computing device cluster can store the same instructions for executing the digital resource management method.

[0172] In some possible implementations, the memory 106 of one or more computing devices 100 in the computing device cluster may also store some instructions for executing the digital asset management method. In other words, the combination of one or more computing devices 100 can jointly execute the instructions for executing the digital asset management method.

[0173] It should be noted that the memory 106 in different computing devices 100 in the computing device cluster can store different instructions for executing part of the functions of the data processing system. In other words, the instructions stored in the memory 106 in different computing devices 100 can implement the functions of one or more devices in the digital resource management device.

[0174] The present application also provides a computer program product comprising instructions. The computer program product may be software or a program product comprising instructions that can be run on a computing device or stored in any available medium. When the computer program product is run on at least one computing device, the at least one computing device executes the digital resource management method.

[0175] The present application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that can be stored by a computing device, or a data storage device such as a data center that contains one or more available media. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive). The computer-readable storage medium includes instructions that instruct a computing device to execute a digital resource management method, or instructs a computing device to execute a digital resource management method.

[0176] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the protection scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A digital resource management method, characterized in that: The method is applied to a management platform, the management platform is used to manage at least one model training platform, the management platform is connected to multiple data spaces, there is a target data space among the multiple data spaces, and the target data space stores at least one digital resource of a data provider, the method includes: Acquire target resource information input by a data user, wherein the target resource information is used to indicate a target digital resource in the at least one digital resource, wherein the target digital resource is set with a target usage permission; Acquire training platform information input by the data user, where the training platform information is used to indicate a target training platform in the at least one model training platform; Generate a data usage policy according to the target usage permission, wherein the data usage policy is used to indicate the way in which the target training platform uses the target digital resource when training a model and / or performing model inference; The data usage policy is sent to the target training platform.

2. The method according to claim 1, characterized in that If the target usage permission is a one-time usage permission, generating a data usage policy according to the target usage permission includes: The data usage policy is generated according to the full usage rights, and the data usage policy is used to instruct the target training platform to use the target digital resource as training set data when training a model.

3. The method according to claim 1, characterized in that: If the target usage permission is a restricted usage permission, generating a data usage policy according to the target usage permission includes: The data usage policy is generated according to the restricted usage rights, and the data usage policy is used to instruct the target training platform to use the target digital resource as the inference set data during model inference.

4. The method according to claim 3, characterized in that: The multiple data spaces include an inference data space, and the data usage policy is further used to instruct the model to be transferred to the inference data space, and the target digital resource is used as the inference set data when performing model inference in the inference data space.

5. The method according to any one of claims 1 to 4, characterized in that: Before obtaining the target resource information input by the data user, the method further includes: The target resource registration information input by the data provider is obtained, wherein the target resource registration information includes the target usage authority and digital resource information, and the digital resource information is used to indicate that the target digital resource is stored in the target data space.

6. The method according to any one of claims 1 to 5, characterized in that: The method further comprises: If a permission termination operation of the data provider on the target digital resource is received, a deletion instruction is sent to the target data space, wherein the deletion instruction is used to instruct the target data space to delete the target digital resource stored internally.

7. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: Receiving usage information of the target digital resource returned by the target data space, wherein the usage information of the target digital resource is used to indicate the transmission and access status of the target digital resource; The usage information of the target digital resource is stored.

8. A digital resource management device, characterized in that: The device is applied to a management platform, the management platform is used to manage at least one model training platform, the management platform is connected to multiple data spaces, there is a target data space among the multiple data spaces, and the target data space stores at least one digital resource of a data provider, and the device includes: An acquisition module, used for acquiring target resource information input by a data user, wherein the target resource information is used for indicating a target digital resource in the at least one digital resource, wherein the target digital resource is provided with a target usage permission; The acquisition module is further used to acquire the training platform information input by the data user, wherein the training platform information is used to indicate a target training platform in the at least one model training platform; A processing module, configured to generate a data usage policy according to the target usage permission, wherein the data usage policy is used to indicate a method for using the target digital resource when the target training platform trains a model and / or performs model inference; A sending module is used to send the data usage policy to the target training platform.

9. The device according to claim 8, characterized in that If the target usage right is full usage right, the processing module is further used to generate the data usage policy according to the full usage right, and the data usage policy is used to instruct the target training platform to use the target digital resource as training set data multiple times when training the model.

10. The device according to claim 8, characterized in that If the target usage right is a restricted usage right, the processing module is further used to generate the data usage policy based on the restricted usage right, and the data usage policy is used to instruct the target training platform to use the target digital resource as the inference set data during model inference.

11. The device according to claim 10, characterized in that The multiple data spaces include an inference data space, and the data usage policy is further used to instruct the model to be transferred to the inference data space, and the target digital resource is used as the inference set data when performing model inference in the inference data space.

12. The device according to any one of claims 8 to 11, characterized in that The acquisition module is also used to acquire the target resource registration information input by the data provider before acquiring the target resource information input by the data user. The target resource registration information includes the target usage rights and digital resource information. The digital resource information is used to indicate that the target digital resource is stored in the target data space.

13. The device according to any one of claims 8 to 12, characterized in that The device also includes: The deletion module is used to send a deletion instruction to the target data space if receiving a termination authority operation of the data provider on the target digital resource, wherein the deletion instruction is used to instruct the target data space to delete the target digital resource stored internally.

14. The device according to any one of claims 8 to 13, characterized in that The device also includes: A storage module is used to receive usage information of the target digital resource returned by the target data space, wherein the usage information of the target digital resource is used to indicate the transmission and access status of the target digital resource; and to store the usage information of the target digital resource in the form of a log; wherein the usage information of the target digital resource stored in the form of a log supports inquiries by the data user and / or the data provider.

15. A computing device cluster, characterized in that: It includes at least one computing device, each computing device includes: a processor and a memory, the processor of the at least one computing device is used to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster executes the digital resource management method as described in any one of claims 1 to 7.

16. A computer-readable storage medium, characterized in that: The method comprises computer program instructions. When the computer program instructions are executed by a computing device cluster, the computing device cluster executes the digital asset management method according to any one of claims 1 to 7.

17. A computer program product, characterized in that The computer program product comprises instructions, and when the instructions are executed by a computing device cluster, the computing device cluster executes the digital asset management method according to any one of claims 1 to 7.

Images