Knowledge graph and internet of things based deep learning model as a service and composition method

Abstract

The deep learning model-as-a-service and composition method based on knowledge graphs and the Internet of Things includes: a service composition step and a recursive composition step. This method combines services by understanding the relationships between the inputs and outputs of the models corresponding to services in the knowledge graph. This allows multiple services to be combined into new services. Through continuous combination, all services in the model service knowledge graph can be arbitrarily combined to form many new services. Multiple services within the combined service can be provided by different users, enabling the same service to be provided collaboratively by different users. This achieves the function of multiple users providing services collaboratively, enriching the service offerings by combining different services.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a deep learning model-as-a-service and composition method based on knowledge graphs and the Internet of Things. Background Technology

[0002] In the process of realizing this invention, the inventors discovered that the prior art has at least the following problems: existing models, such as deep learning models and machine learning models, generally need to be deployed on the user side before they can be used, which is time-consuming and labor-intensive, inconvenient for users, and each model is independent of each other, forming an island. Different models are provided by different users or institutions, which also makes it impossible for them to cooperate to provide services.

[0003] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0004] Therefore, it is necessary to address the shortcomings or deficiencies of existing technologies by providing deep learning model-as-a-service and composition methods based on knowledge graphs and the Internet of Things, in order to solve the problem that existing models cannot be provided as cloud services and can provide services through the combination of multiple models.

[0005] In a first aspect, embodiments of the present invention provide a knowledge graph method, the method comprising:

[0006] Service composition steps: Obtain the model corresponding to the cloud service to be combined as the third model; obtain each model connected to the input of the third model through the model service knowledge graph as each fourth model; combine the data in the input data of the third model that is not connected to the fourth model with the input data of each fourth model to form the input data of the fifth model; use the output data of the third model as the output data of the fifth model; combine the service functions of the third model and the service functions of each fourth model as the service functions of the fifth model; add the fifth model to the model library; obtain the name of the fifth model created by the user as the name of the fifth cloud service; create the fifth cloud service with the name said in the cloud service platform; add the fifth model and the fifth cloud service as entities to the model service knowledge graph.

[0007] Recursive composition steps: Repeat the service composition steps continuously until it is impossible to combine and generate new models and new services.

[0008] Preferably, the method further includes:

[0009] Service model derivation steps: Obtain the name of the first cloud service requested by the second user. If the open-source attribute of the first cloud service is set to true, obtain the model source code with the name from the model source code library and send it to the second user. Obtain the executable code of the modified model by the second user and add it to the model library. Obtain the name of the model and use it as the name of the cloud service. Create a second cloud service with the name of the model on the cloud service platform. Use the second cloud service as a derived cloud service of the first cloud service. Add the modified model and the second cloud service as entities to the model service knowledge graph.

[0010] Preferably, the method further includes:

[0011] The Model-as-a-Service (MAS) steps are as follows: Obtain the name of the cloud service requested by the third user; obtain the input data format, output data format, function description, effect description, function type, and structure type of the cloud service with that name; if the input of the service points to a sensor digital twin through a data source relationship, then obtain the input data through the Internet of Things (IoT); otherwise, obtain the input data uploaded by the third user; determine whether the format of the uploaded input data is consistent with the input data format of the cloud service with that name; if consistent, input the input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if inconsistent, convert the data format of the input data to the input data format of the cloud service with that name; if the conversion is successful, input the converted input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if the conversion fails, remind the third user to re-upload the input data.

[0012] Preferably, the method further includes:

[0013] Service matching steps: Obtain the name of the cloud service requested by the fourth user; if the acquisition fails, prompt the user to input the input data format, output data format, and service function of the requested cloud service; obtain the input data format, output data format, and service function of the requested cloud service; match the input data format, output data format, and service function of the requested cloud service with the input data format, output data format, and service function of each cloud service in the model service knowledge graph; obtain the cloud service with the highest matching degree in the model service knowledge graph and recommend it as the preferred cloud service to the fourth user; obtain feedback on whether the fourth user accepts the preferred cloud service; if the feedback is yes, use the name of the preferred cloud service as the name of the cloud service requested by the fourth user; if the feedback is no, send the preferred cloud service, the input data format, output data format, and service function of the cloud service requested by the user to the engineer, accept the engineer's modification of the model of the preferred cloud service, and add the modified model to the model library; create the name of the model as the name of the cloud service, and create a fourth cloud service with the name said in the cloud service platform; add the model and the fourth cloud service as entities to the model service knowledge graph.

[0014] Secondly, embodiments of the present invention provide a knowledge graph system, the system comprising:

[0015] Service Composition Module: This module obtains the model corresponding to the cloud service to be combined as the third model; obtains each model connected to the input of the third model as a fourth model through the model service knowledge graph; combines the data from the input of the third model that is not connected to the fourth model with the input data of each fourth model to form the input data of the fifth model; uses the output data of the third model as the output data of the fifth model; combines the service functions of the third model and the service functions of each fourth model as the service functions of the fifth model; adds the fifth model to the model library; obtains the name of the fifth model created by the user as the name of the fifth cloud service; creates the fifth cloud service with the name mentioned above on the cloud service platform; and adds the fifth model and the fifth cloud service as entities to the model service knowledge graph.

[0016] Recursive composition module: The service composition module is repeated continuously until it is impossible to combine and generate new models and new services.

[0017] Preferably, the system further includes:

[0018] Service Model Derivation Module: Obtain the name of the first cloud service requested by the second user. If the open-source attribute of the first cloud service is set to true, obtain the model source code with the name from the model source code library and send it to the second user. Obtain the executable code of the modified model by the second user and add it to the model library. Obtain the name of the model and use it as the name of the cloud service. Create a second cloud service with the name of the model on the cloud service platform. Use the second cloud service as a derived cloud service of the first cloud service. Add the modified model and the second cloud service as entities to the model service knowledge graph.

[0019] Preferably, the system further includes:

[0020] Model-as-a-Service (MAS) module: This module obtains the name of the cloud service requested by the third user; it also obtains the input data format, output data format, functional description, effect description, functional type, and structural type of the cloud service with that name; if the input of the service points to a sensor digital twin through a data source relationship, it obtains the input data through the Internet of Things (IoT); otherwise, it obtains the input data uploaded by the third user; it determines whether the format of the uploaded input data is consistent with the input data format of the cloud service with that name; if consistent, it inputs the input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if inconsistent, it converts the data format of the input data to the input data format of the cloud service with that name; if the conversion is successful, it inputs the converted input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if the conversion fails, it prompts the third user to re-upload the input data.

[0021] Preferably, the system further includes:

[0022] Service Matching Module: This module retrieves the name of the cloud service requested by the fourth user. If retrieval fails, it prompts the user to input the input data format, output data format, and service function of the requested cloud service. It then retrieves the user-inputted input data format, output data format, and service function of the requested cloud service and matches them with the input data format, output data format, and service function of each cloud service in the model service knowledge graph. The module selects the cloud service with the highest matching degree and recommends it as the preferred cloud service to the fourth user. It then retrieves feedback on whether the fourth user accepts the preferred cloud service. If the feedback is yes, the name of the preferred cloud service is used as the name of the cloud service requested by the fourth user. If the feedback is no, the module sends the preferred cloud service, the user-inputted input data format, output data format, and service function to an engineer. The engineer modifies the model of the preferred cloud service, and the modified model is added to the model library. The module creates a name for the model, which is used as the name of the cloud service. A fourth cloud service with the name of this model is created on the cloud service platform. Finally, the model and the fourth cloud service are added as entities to the model service knowledge graph.

[0023] Thirdly, embodiments of the present invention provide an artificial intelligence device, wherein the system includes the device of any of the modules described in the second aspect of the embodiments.

[0024] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, characterized in that, when the program is executed by a processor, it implements the steps of the method described in any one of the embodiments of the first aspect.

[0025] Fifthly, embodiments of the present invention provide a robot system, including a memory, a processor, and an artificial intelligence robot program stored in the memory and executable on the processor, characterized in that the processor executes the program to implement the steps of any one of the methods described in the first aspect of the embodiments.

[0026] In a sixth aspect, embodiments of the present invention provide a metaverse, characterized in that the metaverse implements the steps of the method described in any one of the embodiments of the first aspect.

[0027] This embodiment provides a deep learning model-as-a-service and composition method based on knowledge graphs and the Internet of Things, including: a service composition step; and a recursive composition step. The method combines services by understanding the relationships between the inputs and outputs of the models corresponding to services in the knowledge graph. This allows multiple services to be combined into new services. Through continuous combination, all services in the model service knowledge graph can be arbitrarily combined to form many new services. Multiple services within the combined service can be provided by different users, enabling the same service to be provided collaboratively by different users. This achieves the function of multiple users providing services collaboratively, allowing different services to be combined to provide services to users, thus enriching the service offerings. Attached Figure Description

[0028] Figure 1 A module diagram of a knowledge graph system provided for embodiments of the present invention;

[0029] Figure 2 A module diagram of a knowledge graph system provided for embodiments of the present invention;

[0030] Figure 3 A module diagram of a knowledge graph system provided for embodiments of the present invention;

[0031] Figure 4 A module diagram of a knowledge graph system provided for embodiments of the present invention. Detailed Implementation

[0032] The technical solutions in the embodiments of the present invention will be described in detail below with reference to the embodiments of the present invention.

[0033] I. Basic Embodiments of the Invention

[0034] Firstly, embodiments of the present invention provide a knowledge graph method, the method comprising: a service composition step; and a recursive composition step. Technical effects: By leveraging the relationship between the inputs and outputs of the models corresponding to services in the knowledge graph, services are composed, enabling multiple services to be combined into new services. Through continuous combination, all services in the model service knowledge graph can be arbitrarily combined to form many new services. Multiple services within the combined services can be provided by different users, thus enabling the same service to be provided collaboratively by different users, realizing the function of multiple users providing services collaboratively, and allowing different services to be combined to provide services to users, making the services richer.

[0035] In a preferred embodiment, the method further includes a service model derivation step. Technical effect: By open-sourcing the models corresponding to services and allowing users to modify existing service models to create new models and services, the models and services become richer.

[0036] In a preferred embodiment, the method further includes a model-as-a-service step. Technical effect: By providing cloud services through models, models can accept user requests in the cloud and receive input and feedback outputs to users, making it more convenient for users to obtain model services. For example, in the context of the prevalence of deep learning models, model-as-a-service can greatly facilitate users' use of deep learning models.

[0037] In a preferred embodiment, the method further includes a service matching step. Technical effect: Because there are many services in the model service knowledge graph, users may not be able to browse them all to select a suitable service. Therefore, matching services based on the user's needs allows suitable services to be obtained from the knowledge graph. If the recommended service does not meet the user's needs, engineers can modify it and provide feedback to the user, thereby improving the user's service experience.

[0038] Secondly, embodiments of the present invention provide a knowledge graph system, such as... Figure 1 As shown, the system includes: a service composition module; and a recursive composition module.

[0039] In a preferred embodiment, such as Figure 2 As shown, the system also includes a service model derivation module.

[0040] In a preferred embodiment, such as Figure 3 As shown, the system also includes a model-as-a-service module.

[0041] In a preferred embodiment, such as Figure 4 As shown, the system also includes a service matching module.

[0042] Thirdly, embodiments of the present invention provide an artificial intelligence device, wherein the system includes the device of any of the modules described in the second aspect of the embodiments.

[0043] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, characterized in that, when the program is executed by a processor, it implements the steps of the method described in any one of the embodiments of the first aspect.

[0044] Fifthly, embodiments of the present invention provide a robot system, including a memory, a processor, and an artificial intelligence robot program stored in the memory and executable on the processor, characterized in that the processor executes the program to implement the steps of any one of the methods described in the first aspect of the embodiments.

[0045] In a sixth aspect, embodiments of the present invention provide a metaverse, characterized in that the metaverse implements the steps of the method described in any one of the embodiments of the first aspect.

[0046] II. Preferred Embodiments of the Invention

[0047] (I) Key Issues

[0048] How can we realize an open smart healthcare and elderly care service platform based on "model as a service" for third parties, encompassing home, community, and institutional care, to improve the personalization of multi-party healthcare and elderly care service provision across institutions, communities, and homes, and enhance the integration of healthcare and elderly care services such as early disease screening, rehabilitation of daily living abilities, and remote intervention management?

[0049] (II) Key Technologies

[0050] Through cloud computing and metaverse technology, we will realize an open smart healthcare and elderly care service platform for third parties, based on "model as a service" and accessible to third parties, covering home, community, and institutional settings.

[0051] (III) Key Technologies

[0052] This is an open smart healthcare service platform for third parties, based on the "model as a service" model, which can improve the personalization of the integrated healthcare, rehabilitation and elderly care services provided by institutions, communities and homes, and can also improve the integration of healthcare, rehabilitation and elderly care services such as early disease screening, rehabilitation of daily behavior ability and remote intervention management.

[0053] (iv) Technical Overview

[0054] A cloud service platform is established to accept registrations from institutions. Registered institutions can then register the services they wish to offer. The platform provides standard models for various services in institutional, community-based, and home-based elderly care. When registering a new service, institutions first select a suitable model from the platform's standard models. They then adjust the various parameters of the model based on the actual services they can provide. Following the model's intelligent customer service guidance, they connect the service providers (e.g., caregivers) or objects (e.g., smart devices) to the model via the Internet of Things (IoT) to form the institution's registered service model. If the model's intelligent customer service detects difficulties during the registration process, it will notify the platform for human assistance. If an institution cannot find a corresponding standard model for its desired service type on the platform, it can submit detailed service information. The platform will modify existing similar models to create a new standard model and add it to the platform. All service models on the platform are digital twin models. These models not only display service attributes but also reflect and interact with the actual services in real time, forming a comprehensive healthcare and elderly care ecosystem, serving as a third-party smart healthcare and elderly care service platform.

[0055] (V) Detailed Technical Plan

[0056] Part 1:

[0057] The steps for a user to upload a model as a service are as follows: Obtain the model uploaded by the first user (the model includes a model description and executable code), and add it to the model library; obtain the model's name, use it as the name of the cloud service, and create a first cloud service with the name stated in the cloud service platform; add the model and the cloud service as entities to the model service knowledge graph; the model and the cloud service are connected through model service relationships; the first user includes organizations and individuals.

[0058] Steps to obtain service model parameters: Obtain the model's input data format, output data format, functional description, effect description, model function type, and model structure type as the input data format, output data format, functional description, effect description, function type, and structure type of the first cloud service; Obtain the information of the first user as the provider information of the cloud service; the information of the first user includes the first user's name, qualifications, and identification number. The effect description includes accuracy, recall, false positive rate, and false negative rate; the functional description includes the tasks that can be completed; the model function type includes classification model, prediction model, and regression model; the model structure type includes deep learning model, expert system model, machine learning model, and data statistical model; Add the models uploaded by the organization to the model library.

[0059] Part 2:

[0060] Model connection steps: First, obtain the models to be connected as the first model and the second model. If the output data of the first model is the input data of the second model, then in the model service knowledge graph, the output of the first model points to the input of the second model through a full-full connection, where the attributes of the connection are the data format of the output data and the data format of the input data. If a portion of the output data of the first model is a portion of the input data of the second model, then in the model service knowledge graph, the output of the first model points to the input of the second model through a partial-partial connection, where the attributes of the connection are the data format of the portion of the output data and the data format of the portion of the input data. If a portion of the output data of the first model is the input data of the second model, then in the model service knowledge graph, the output of the first model points to the input of the second model through a partial-full connection, where the attributes of the connection are the data format of the portion of the output data and the data format of the input data. If the output data of the first model is a portion of the input data of the second model, then in the model service knowledge graph, the output of the first model points to the input of the second model through a full-partial connection, where the attributes of the connection are the data format of the output data and the data format of the portion of the input data. This step allows multiple model services to be combined to form a new model service. The first model and the second model have different providers or have the same provider.

[0061] Service composition steps: First, obtain the model corresponding to the cloud service to be combined as the third model. Second, obtain each model connected to the input of the third model through the model service knowledge graph as a fourth model. Third, combine the input data of the third model that is not connected to the fourth model with the input data of each fourth model to form the input data of the fifth model. Fourth, use the output data of the third model as the output data of the fifth model. Fifth, combine the service functions of the third model and the service functions of each fourth model as the service functions of the fifth model. Fifth, add the fifth model to the model library. Sixth, obtain the name of the user-created fifth model as the name of the fifth cloud service and create the fifth cloud service with the name stated on the cloud service platform. Seventh, add the fifth model and the fifth cloud service as entities to the model service knowledge graph. In the model service knowledge graph, the fifth model and the fifth cloud service are connected through model service relationships. The third model points to the fifth model through model composition relationships, and the fourth model points to the fifth model through model composition relationships. Fifth, using the fifth model as either the third or fourth model, can be used for new model and service combinations. Repeat the service composition steps until no new models or services can be generated. The third and fourth models may have different providers or the same provider.

[0062] Part 3:

[0063] Service model open source steps: prompt the first user to select whether the model of the first cloud service is open source; if the first user selects the model to be open source, obtain the model source code uploaded by the first user, add it to the model source code library, and set the open source attribute of the first cloud service to true;

[0064] Service model derivation steps: Obtain the name of the first cloud service requested by the second user. If the open-source attribute of the first cloud service is set to true, obtain the model source code with the name from the model source code library and send it to the second user. Obtain the executable code of the modified model by the second user and add it to the model library. Obtain the name of the model and use it as the name of the cloud service. Create a second cloud service with the name of the model on the cloud service platform. Use the second cloud service as a derived cloud service of the first cloud service. Add the modified model and the second cloud service as entities to the model service knowledge graph. The modified model and the second cloud service are connected through model service relationships. The first cloud service and the second cloud service are connected through derived service relationships.

[0065] Part 4:

[0066] IoT knowledge graph steps: Obtain the source of input data for each service in the IoT knowledge graph. If the source is data collected by sensors on the IoT, add the sensor digital twin from the metaverse to the model service knowledge graph. In the model service knowledge graph, point the input of each service to the sensor digital twin through the data source relationship.

[0067] The Model-as-a-Service (MAS) steps are as follows: Obtain the name of the cloud service requested by the third user; obtain the input data format, output data format, function description, effect description, function type, and structure type of the cloud service with that name; if the input of the service points to a sensor digital twin through a data source relationship, then obtain the input data through the Internet of Things (IoT); otherwise, obtain the input data uploaded by the third user; determine whether the format of the uploaded input data is consistent with the input data format of the cloud service with that name; if consistent, input the input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if inconsistent, convert the data format of the input data to the input data format of the cloud service with that name; if the conversion is successful, input the converted input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if the conversion fails, remind the third user to re-upload the input data.

[0068] Part 5:

[0069] Service matching steps: Obtain the name of the cloud service requested by the fourth user; if the acquisition fails, prompt the user to input the input data format, output data format, and service function of the requested cloud service; obtain the input data format, output data format, and service function of the requested cloud service; match the input data format, output data format, and service function of the requested cloud service with the input data format, output data format, and service function of each cloud service in the model service knowledge graph; obtain the cloud service with the highest matching degree in the model service knowledge graph and recommend it as the preferred cloud service to the fourth user; obtain feedback on whether the fourth user accepts the preferred cloud service. If the feedback is "accepted," the name of the preferred cloud service will be used as the name of the cloud service requested by the fourth user. If the feedback is "disapproved," the preferred cloud service, the input data format of the cloud service requested by the user, the output data format, and the service functions will be sent to the engineer. The engineer will modify the model of the preferred cloud service, and the modified model will be added to the model library. The name of the model will be created and used as the name of the cloud service. A fourth cloud service with the name of the model will be created on the cloud service platform. The model and the fourth cloud service will be added as entities to the model service knowledge graph. The model and the fourth cloud service will be connected through model service relationships. The fourth cloud service will be used as the name of the cloud service requested by the fourth user.

[0070] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A knowledge graph method, characterized in that, The method includes: Service composition steps: Obtain the model corresponding to the cloud service to be combined as the third model; obtain each model connected to the input of the third model through the model service knowledge graph as each fourth model; combine the data in the input data of the third model that is not connected to the fourth model with the input data of each fourth model to form the input data of the fifth model; use the output data of the third model as the output data of the fifth model; combine the service functions of the third model and the service functions of each fourth model as the service functions of the fifth model; add the fifth model to the model library; obtain the name of the fifth model created by the user as the name of the fifth cloud service; create the fifth cloud service with the name said in the cloud service platform; add the fifth model and the fifth cloud service as entities to the model service knowledge graph. Recursive composition steps: Repeat the service composition steps continuously until it is impossible to combine and generate new models and new services.

2. The knowledge graph method according to claim 1, characterized in that, The method further includes: Service model derivation steps: Obtain the name of the first cloud service requested by the second user. If the open-source attribute of the first cloud service is set to true, obtain the model source code with the name from the model source code library and send it to the second user. Obtain the executable code of the modified model by the second user and add it to the model library. Obtain the name of the model and use it as the name of the cloud service. Create a second cloud service with the name of the model on the cloud service platform. Use the second cloud service as a derived cloud service of the first cloud service. Add the modified model and the second cloud service as entities to the model service knowledge graph.

3. The knowledge graph method according to claim 1, characterized in that, The method further includes: The Model-as-a-Service (MAS) steps are as follows: Obtain the name of the cloud service requested by the third user; obtain the input data format, output data format, function description, effect description, function type, and structure type of the cloud service with that name; if the input of the service points to a sensor digital twin through a data source relationship, then obtain the input data through the Internet of Things (IoT); otherwise, obtain the input data uploaded by the third user; determine whether the format of the uploaded input data is consistent with the input data format of the cloud service with that name; if consistent, input the input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if inconsistent, convert the data format of the input data to the input data format of the cloud service with that name; if the conversion is successful, input the converted input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if the conversion fails, remind the third user to re-upload the input data.

4. The knowledge graph method according to claim 1, characterized in that, The method further includes: Service matching steps: Obtain the name of the cloud service requested by the fourth user; if the acquisition fails, prompt the user to input the input data format, output data format, and service function of the requested cloud service; obtain the input data format, output data format, and service function of the requested cloud service; match the input data format, output data format, and service function of the requested cloud service with the input data format, output data format, and service function of each cloud service in the model service knowledge graph; obtain the cloud service with the highest matching degree in the model service knowledge graph and recommend it as the preferred cloud service to the fourth user; obtain feedback on whether the fourth user accepts the preferred cloud service; if the feedback is yes, use the name of the preferred cloud service as the name of the cloud service requested by the fourth user; if the feedback is no, send the preferred cloud service, the input data format, output data format, and service function of the cloud service requested by the user to the engineer, accept the engineer's modification of the model of the preferred cloud service, and add the modified model to the model library; create the name of the model as the name of the cloud service, and create a fourth cloud service with the name said in the cloud service platform; add the model and the fourth cloud service as entities to the model service knowledge graph.

5. A knowledge graph system, characterized in that, The system includes: Service Composition Module: This module obtains the model corresponding to the cloud service to be combined as the third model; obtains each model connected to the input of the third model as a fourth model through the model service knowledge graph; combines the data from the input of the third model that is not connected to the fourth model with the input data of each fourth model to form the input data of the fifth model; uses the output data of the third model as the output data of the fifth model; combines the service functions of the third model and the service functions of each fourth model as the service functions of the fifth model; adds the fifth model to the model library; obtains the name of the fifth model created by the user as the name of the fifth cloud service; creates the fifth cloud service with the name mentioned above on the cloud service platform; and adds the fifth model and the fifth cloud service as entities to the model service knowledge graph. Recursive composition module: The service composition module is repeated continuously until it is impossible to combine and generate new models and new services.

6. The knowledge graph system according to claim 5, characterized in that, The system also includes: Service Model Derivation Module: Obtain the name of the first cloud service requested by the second user. If the open-source attribute of the first cloud service is set to true, obtain the model source code with the name from the model source code library and send it to the second user. Obtain the executable code of the modified model by the second user and add it to the model library. Obtain the name of the model and use it as the name of the cloud service. Create a second cloud service with the name of the model on the cloud service platform. Use the second cloud service as a derived cloud service of the first cloud service. Add the modified model and the second cloud service as entities to the model service knowledge graph.

7. The knowledge graph system according to claim 5, characterized in that, The system also includes: Model-as-a-Service (MAS) module: This module obtains the name of the cloud service requested by the third user; it also obtains the input data format, output data format, functional description, effect description, functional type, and structural type of the cloud service with that name; if the input of the service points to a sensor digital twin through a data source relationship, it obtains the input data through the Internet of Things (IoT); otherwise, it obtains the input data uploaded by the third user; it determines whether the format of the uploaded input data is consistent with the input data format of the cloud service with that name; if consistent, it inputs the input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if inconsistent, it converts the data format of the input data to the input data format of the cloud service with that name; if the conversion is successful, it inputs the converted input data into the model corresponding to the cloud service for calculation, and the output is fed back to the third user; if the conversion fails, it prompts the third user to re-upload the input data.

8. The knowledge graph system according to claim 5, characterized in that, The system also includes: Service Matching Module: This module retrieves the name of the cloud service requested by the fourth user. If retrieval fails, it prompts the user to input the input data format, output data format, and service function of the requested cloud service. It then retrieves the user-inputted input data format, output data format, and service function of the requested cloud service and matches them with the input data format, output data format, and service function of each cloud service in the model service knowledge graph. The module selects the cloud service with the highest matching degree and recommends it as the preferred cloud service to the fourth user. It then retrieves feedback on whether the fourth user accepts the preferred cloud service. If the feedback is yes, the name of the preferred cloud service is used as the name of the cloud service requested by the fourth user. If the feedback is no, the module sends the preferred cloud service, the user-inputted input data format, output data format, and service function to an engineer. The engineer modifies the model of the preferred cloud service, and the modified model is added to the model library. The module creates a name for the model, which is used as the name of the cloud service. A fourth cloud service with the name of this model is created on the cloud service platform. Finally, the model and the fourth cloud service are added as entities to the model service knowledge graph.

9. A robot system, characterized in that, The robot performs the steps of the method according to any one of claims 1-4.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-4.

Images