Response information generation method and apparatus, device, storage medium, and product

Abstract

Embodiments of the present application disclose a response information generation method, device, equipment, storage medium and product. The method comprises: based on received task information sent by a model protocol client, determining calling interface information from pre-stored interface configuration information of at least one service provider; wherein the model protocol client is a client connected to a model protocol server in a model context protocol architecture; based on the calling interface information, accessing a calling service provider corresponding to the calling interface information to obtain associated service data related to the task information from service data of the calling service provider; sending the associated service data to the model protocol client to enable the model protocol client to input the associated service data and the task information into a pre-trained task execution model, and determining response information corresponding to the task information based on a first output result of the task execution model. The technical solution of the embodiments of the present application can reduce the modification cost.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of artificial intelligence, and in particular to a response information generation method and device, equipment, storage medium and product. BACKGROUND

[0002] With the continuous expansion of artificial intelligence application scenarios, in order to be able to call external tools more, a model context protocol (MCP) is proposed, which establishes a secure two-way connection between a task execution model and a data source through the model context protocol.

[0003] In the prior art, a task execution model needs to execute a task based on business data and obtain an execution result corresponding to the task, so a large number of different business parties need to be configured as MCP servers, so that an MCP client can access the MCP servers to obtain business data for use by the task execution model. However, in the process of implementing the present application, it is found that the prior art at least has the following technical problems: configuring business parties as MCP servers has high modification cost and difficulty; and the more business parties participate, the more MCP servers are modified, which is not convenient for maintenance. SUMMARY

[0004] Embodiments of the present application provide a response information generation method, device, equipment, storage medium and product to achieve the purpose of reducing workload, reducing modification cost and facilitating server maintenance.

[0005] According to an aspect of the present application, a response information generation method is provided, applied to a model protocol server, comprising:

[0006] Based on the received task information sent by the model protocol client, the calling interface information is determined from the pre-stored interface configuration information of at least one business party; wherein the model protocol client is a client connected to the model protocol server in the model context protocol architecture;

[0007] Based on the calling interface information, the calling business party corresponding to the calling interface information is accessed to obtain associated business data related to the task information from the business data of the calling business party;

[0008] The associated business data is sent to the model protocol client, so that the model protocol client inputs the associated business data and the task information into a pre-trained task execution model, and determines the response information corresponding to the task information based on the first output result of the task execution model.

[0009] According to another aspect of the present application, there is provided a response information generation method applied to a model protocol client, the method comprising:

[0010] generating task information, sending the task information to a model protocol server, so that the model protocol server determines calling interface information from pre-stored interface configuration information of at least one service party, accesses a calling service party corresponding to the calling interface information based on the calling interface information, and acquires associated service data related to the task information from service data of the calling service party; wherein the model protocol server is a server connected to the model protocol client in a model context protocol architecture;

[0011] receiving the associated service data sent by the model protocol server, inputting the associated service data and the task information into a pre-trained task execution model, and determining response information corresponding to the task information based on a first output result of the task execution model.

[0012] According to another aspect of the present application, there is provided a response information generation device configured in a model protocol server, the device comprising:

[0013] an information determination module configured to determine calling interface information from pre-stored interface configuration information of at least one service party based on received task information sent by a model protocol client; wherein the model protocol client is a client connected to the model protocol server in a model context protocol architecture;

[0014] a data acquisition module configured to access a calling service party corresponding to the calling interface information based on the calling interface information, and to acquire associated service data related to the task information from service data of the calling service party;

[0015] a data sending module configured to send the associated service data to the model protocol client, so that the model protocol client inputs the associated service data and the task information into a pre-trained task execution model, and determines response information corresponding to the task information based on a first output result of the task execution model.

[0016] According to another aspect of the present application, there is provided a response information generation device configured in a model protocol client, the device comprising:

[0017] The information generation module is configured to generate task information, send the task information to a model protocol server, and cause the model protocol server to determine calling interface information from pre-stored interface configuration information of at least one service provider, access a calling service provider corresponding to the calling interface information based on the calling interface information, and obtain associated service data related to the task information from service data of the calling service provider; the model protocol server is a server connected to a model protocol client in a model context protocol architecture.

[0018] The data receiving module is configured to receive the associated service data sent by the model protocol server, input the associated service data and the task information into a pre-trained task execution model, and determine response information corresponding to the task information based on a first output result of the task execution model.

[0019] According to another aspect of the present application, a model protocol server is provided, comprising:

[0020] at least one processor; and

[0021] a memory in communication connection with the at least one processor; wherein

[0022] the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the response information generation method applied to the model protocol server according to any one of the embodiments of the present application.

[0023] According to another aspect of the present application, a model protocol client is provided, comprising:

[0024] at least one processor; and

[0025] a memory in communication connection with the at least one processor; wherein

[0026] the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the response information generation method applied to the model protocol client according to any one of the embodiments of the present application.

[0027] According to another aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for causing a processor to execute the response information generation method according to any one of the embodiments of the present application.

[0028] According to another aspect of the present application, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the response information generation method according to any of the embodiments of the present application.

[0029] The technical scheme of the embodiment of the present application determines the calling interface information from the pre-stored interface configuration information of at least one service party based on the task information sent by the model protocol client, wherein the model protocol client is a client connected with the model protocol server in the model context protocol architecture; the calling service party corresponding to the calling interface information is accessed through the calling interface information, so that the calling service party can be accessed without being configured as the server corresponding to the model context protocol, and the associated business data can be obtained from the calling service party; then, the associated business data is sent to the model protocol client, so that the model protocol client inputs the associated business data and the task information into the pre-trained task execution model, and determines the response information corresponding to the task information based on the first output result of the task execution model. The scheme accesses the calling service party through the pre-stored interface configuration information, so that the associated business data can be obtained without modifying the service party, the workload is reduced, the modification cost is reduced, and the number of MCP servers can be effectively reduced, which is convenient for maintenance and management.

[0030] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0032] Figure 1 is a flow chart of a response information generation method according to an embodiment of the present application;

[0033] Figure 2 is a schematic diagram of a model context protocol use case structure according to an embodiment of the present application;

[0034] Figure 3 is a flow chart of another response information generation method according to an embodiment of the present application;

[0035] Figure 4 is a design timing diagram of an MCP mechanism according to an embodiment of the present application;

[0036] Figure 5 is a flow chart of another response information generation method according to an embodiment of the present application;

[0037] Figure 6 is a schematic diagram of an MPC architecture interface configuration process according to an embodiment of the present application;

[0038] Figure 7 is a flow chart of an MCP core process according to an embodiment of the present application;

[0039] Figure 8 is a flow chart of another response information generation method according to an embodiment of the present application;

[0040] Figure 9 is a structural schematic diagram of a response information generation apparatus according to an embodiment of the present application;

[0041] Figure 10 is a structural schematic diagram of another response information generation apparatus according to an embodiment of the present application;

[0042] Figure 11 is a structural schematic diagram of a model protocol server for implementing an embodiment of the present application;

[0043] Figure 12 is a structural schematic diagram of a model protocol client for implementing an embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to make the personnel in the art better understand the present application scheme, the technical scheme in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by the person skilled in the art without making creative efforts should belong to the scope of protection of the present application.

[0045] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "etc." and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0046] It should be noted that in the technical solutions of the present disclosure, the collection, collection, updating, analysis, processing, use, transmission, storage and other aspects of user personal information comply with relevant laws and regulations, are used for legal purposes, and do not violate public order and good customs. Necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain user personal information security and network security.

[0047] Before introducing the technical solutions, the application scenario can be exemplarily described. In the prior art, when answering user questions, the task execution model needs to combine the business data of different business parties to give accurate answers. Since different business parties need to be built as MCP servers according to the MCP architecture and pre-bound to the corresponding MCP client, the business data corresponding to the business party can be obtained.

[0048] For example, an accountant wants to count the company's expenses through a task execution model, which needs to obtain employee salaries through a financial business party and obtain expenses generated by company construction through an administrative business party. To obtain data in these two aspects, the financial business party and the administrative business party need to be configured as MCP servers to effectively obtain business data, resulting in cumbersome operation, high cost, and inconvenience for maintenance.

[0049] Moreover, the business party usually has multiple applications, and each application is equipped with a certain amount of business interface. Due to the variety of business types, the number of business interfaces is huge. If the business party is configured as an MCP server, the amount of reconstruction work is huge, it is difficult to promote and land, and it cannot be used in a wide range of business. In order to support artificial intelligence to determine response information, the business data source needs to be provided to the task execution model for use, resulting in data security problems. If the MCP server is bound to the MCP client in advance, the flexibility in use is poor.

[0050] The technical solution can be applied to a scene of combining business data to execute a to-be-executed task to obtain an execution result corresponding to the to-be-executed task. Exemplarily, the to-be-executed task can be a question raised by a user, and the execution result is an answer corresponding to the question. Through the technical solution, each business party does not need to be configured as an MCP server, and the business party is called through pre-stored interface configuration information, and then business data is obtained, thereby reducing cost investment and facilitating MCP server maintenance. It should be noted that the task execution model provided in the technical solution is a large model for combining business data to determine an execution result corresponding to a to-be-executed task. The large model is also called a large language model or a base model, which refers to a machine learning model with a large parameter size and a complex calculation structure, such as a GPT model, a Gemini model, and a Yanyi large model. By inputting task information and business data into the task execution model, response information matched with the question can be obtained.

[0051] Figure 1 A flowchart of a response information generation method according to an embodiment of the application is shown in FIG. 1. The embodiment can be applied to a case of combining business data to determine response information corresponding to task information. The method can be executed by a response information generation apparatus, which can be implemented in the form of hardware and / or software.

[0052] As shown in FIG. 2, the method of the embodiment can be applied to a model protocol server and can specifically include the following steps. Figure 1

[0053] S110, based on the task information sent by the received model protocol client, determining the calling interface information from the pre-stored interface configuration information of at least one business party.

[0054] Exemplarily, when the task information is question information, the corresponding response information is question answer information, and the corresponding task execution model can be a question and answer model. When the task information is instruction information, the corresponding response information is execution result information corresponding to the instruction. The model protocol client is a client connected to the model protocol server in a model context protocol architecture. It should be noted that the response information generation method provided in the embodiment is applied to a model protocol server.

[0055] To facilitate a more clear description of the response information generation method, Figure 2 A schematic diagram of a model context protocol use case structure according to an embodiment of the application is shown in FIG. 3. As shown in FIG. 3, the model context protocol use case structure can include the following steps. Figure 2 ​As shown, the model protocol client, the model protocol server and the business interface are connected in the manner of a connection pipeline. The model protocol client is an upper container, is not bound to a specific business and the number thereof can be N. The business interface is a lower container and the number thereof can be N. The model protocol server is a connection pipeline and the number thereof can be 1. N is a positive integer greater than 1. The function of the model protocol client is to call a task execution model to obtain response information. The function of the model protocol server is to provide a unified gateway service, query interface information and access a business interface. The function of the business interface is to obtain business data from a business data source. The business interface includes an RPC (Remote Procedure Call) interface and / or an Http (Hypertext Transfer Protocol) interface. The model protocol server, the business interface and the model protocol client can implement the above functions through a through, include mechanism.

[0056] In this embodiment, the model protocol client can be embedded into an artificial intelligence server. The artificial intelligence server can also be deployed with an interface and a master control. Through the interface, task information sent by an artificial intelligence (AI) client can be obtained. The task information can be a question input on an interface corresponding to the artificial intelligence client. It can be understood that the question is a question that a user needs to obtain an answer through a task execution model. The task information is transmitted to the master control through the interface for format verification and security verification, and after verification, the task information is sent to the model protocol client. The artificial intelligence client can be a terminal device embedded with an artificial intelligence module, for example, the artificial intelligence client includes at least one of a mobile phone, a computer, a wearable device and a smart home device.

[0057] The model protocol client and the model protocol server in this embodiment are not bound to a specific business, so that the model protocol client can be dynamically created, the flexibility is stronger and the operation is more convenient. In addition, the model protocol client, the model protocol server and the business interface are connected in the manner of a connection pipeline, only one model protocol server is needed, which greatly improves the maintainability of the model protocol server and reduces the transformation cost.

[0058] In the embodiment, the interface configuration information of at least one service interface corresponding to each service party can be pre-stored in the database. The interface configuration information is used to reflect the interaction specification between the service party and the model protocol client. The interface configuration information can include interface-level metadata and field-level metadata. The interface-level metadata can include interface address, interface name, interface description and the like. The field-level metadata can include parameter field name, parameter field description, field type, whether to be transmitted, field length and the like. Through the interface configuration information, the service data can be provided for the task execution model. Exemplarily, the service interface can include an RPC interface and / or an Http interface, so that the service parties with different interface types can be covered, and the service data can be comprehensively and effectively obtained.

[0059] In a specific implementation, after the task information is received by the model protocol client, the service data needs to be obtained, so that the task execution model can determine the response information corresponding to the task information based on the service data. In order to accurately provide the service data while improving the efficiency and reducing the processing time as much as possible, the interface information can be determined based on the number of interface configuration information. Specifically, when the number of interface configuration information is less than the preset interface number, all the interface configuration information can be used as the calling interface information, so that the service data can be comprehensively provided, and the task execution model can be facilitated to process. When the number of interface configuration information is greater than or equal to the preset interface number, the task information can be compared with the historical tasks in the historical task record, the historical task with a similarity higher than a first preset threshold to the task information can be used as a reference task, and the calling interface information corresponding to the reference task can be used as the calling interface information corresponding to the task information, so that the historical task record is combined, and the calling interface information of the task information can be quickly and accurately determined.

[0060] S120, based on the calling interface information, accessing the calling service party corresponding to the calling interface information to obtain the associated service data related to the task information from the service data of the calling service party.

[0061] In the embodiment, the calling interface information can be an interface address, and the model protocol server can determine the interface input parameter corresponding to the interface address. Exemplarily, the interface input parameter can be randomly determined in an enumeration manner, or the interface input parameter used in the historical access can be used as the interface input parameter required for the current access based on the historical access record. Through the interface input parameter, the data acquisition request can be generated. Further, the data acquisition request corresponding to the interface address can be sent to the calling service party through the interface address, so that the calling service party can obtain the service data corresponding to the data acquisition request as the associated service data related to the task information.

[0062] S130, send the associated business data to the model protocol client, so that the model protocol client inputs the associated business data and the task information into the pre-trained task execution model, determines the response information corresponding to the task information based on the first output result of the task execution model.

[0063] In this embodiment, after obtaining the associated business data, in order to ensure the learning accuracy of the task execution model, the associated business data corresponding to the calling business party can be verified in data format based on the business type of the calling business party. For example, the data format verification can include at least one of data type verification, data value range verification, and logic rule verification. For ease of understanding, a scenario of counting attendance duration is taken as an example for description. In this scenario, the associated business data can be data for reflecting attendance date, therefore, the data type of the associated business data can be time date type, and then the associated business data conforming to the time date type can be retained and the associated business data not conforming to the time date type can be deleted.

[0064] Further, the associated business data that passes the data format verification can be sent to the model protocol client. After receiving the associated business data, the model protocol client inputs each associated business data and the task information into the task execution model, and outputs the response information corresponding to the task information as the first output result of the task execution model after combining the associated business data. By analyzing the first output result, the response information corresponding to the task information can be obtained.

[0065] In actual application, the task information has diversity. If the response information is displayed in a unified format for different task information, it is not easy to reflect the difference between the response information corresponding to different task information. In order to facilitate the user to view the response information, different response formats can be used to generate the response information for different task information.

[0066] Specifically, the associated business data is sent to the model protocol client, so that the model protocol client inputs the associated business data and the task information into the pre-trained task execution model, and determines the response information corresponding to the task information based on the first output result of the task execution model. The method comprises the following steps: sending the associated business data to the model protocol client, so that the model protocol client sends the associated business data, the task information and the pre-set format requirement information to the task execution model, obtains the first output result corresponding to the task information, and takes the first output result as the response information corresponding to the task information.

[0067] The format requirement information records the correspondence between the task type and the response format. For example, when the task information is question information, the task type can include technical implementation type questions, concept explanation type questions, recommendation evaluation type questions, and troubleshooting type questions, etc. The technical implementation type question can be "how to write code for implementing function A?", and the corresponding response format is the format corresponding to the code snippet. The concept explanation type question can be "what is MCP?", and the response format can be "definition + analogy + case" format, so as to comprehensively explain MCP. The recommendation evaluation type question can be "which one is better, tool A or tool B?", and the corresponding response format is a comparison table form. The troubleshooting type question can be "disk space is insufficient", and the corresponding answer can be in the form of showing the diagnosis process and the fault reason.

[0068] In this embodiment, the associated business data, the task information, and the format requirement information can be sent to the task execution model. The task execution model determines the answer corresponding to the task information according to the associated business data and the task information, and takes the response format corresponding to the task type of the task information as the target format. The determined answer is arranged according to the target format, and the arrangement result is taken as the first output result, which is taken as the response information corresponding to the task information.

[0069] In this embodiment, the format requirement information is sent to the task execution model, so that the task execution model obtains the response information corresponding to the task information according to the format requirement matched with the task information, realizes the function of flexibly setting the format of the response information based on the difference of the task information, and is beneficial to intuitively and clearly display the response information and facilitate the user to view.

[0070] Further, the format requirement information can further include a general format, which can be used for the answer of any task information. For example, when the task information corresponding to the task type is not recorded in the format requirement information, the general format can be used to generate the first output result corresponding to the task information, and then the response information of the general format is obtained. By setting the general format, it is ensured that the answer corresponding to the task information can be arranged to obtain the response information. Moreover, the response information can be orderly and clearly displayed, and the convenience of the user to view is improved.

[0071] The technical scheme of the embodiment of the application is based on the received task information sent by the model protocol client, determines the calling interface information from the pre-stored interface configuration information of at least one service party, wherein the model protocol client is a client connected with the model protocol server in the model context protocol architecture; the calling service party corresponding to the calling interface information is accessed through the calling interface information, so that the calling service party can be accessed without being configured as the server corresponding to the model context protocol, and the associated business data can be obtained from the calling service party; then, the associated business data is sent to the model protocol client, so that the model protocol client inputs the associated business data and the task information into the pre-trained task execution model, determines the response information corresponding to the task information based on the first output result of the task execution model. The scheme accesses the calling service party through the pre-stored interface configuration information, so that the associated business data can be obtained without modifying the service party, the workload is reduced, the modification cost is reduced, and the number of MCP servers can be effectively reduced, which is convenient for maintenance and management.

[0072] Figure 3 is a flowchart of another response information generation method provided according to the embodiment of the application. Based on the above-mentioned embodiment, the method can optionally include the following steps. Figure 3

[0073] S210, based on the received task information sent by the model protocol client and the interface configuration information, determining the information prompt word of the interface configuration information associated with the task information.

[0074] The information prompt word can be the description information corresponding to the interface configuration information associated with the task information, and the description information is used to reflect the interface function and the use mode.

[0075] It should be noted that, in order to more accurately obtain the business data associated with the task information and avoid inputting a large amount of business data into the task execution model, the calling interface can be determined based on the information prompt word of the interface configuration information. For the convenience of clearly understanding the process, please refer to Figure 4 . As shown in Figure 4 ​As shown, the user inputs the task information to the AI client, and the AI client submits the task information to the interface deployed in the AI server, and calls the model protocol client through the interface. As an example, the number of model protocol clients can be multiple, and one model protocol client can be started based on the remaining resources of each model protocol client, to send the task information to the model protocol server to determine the information prompt word. For example, the model protocol client with the most remaining resources can be called to transmit the task information to the model protocol server through the model protocol client.

[0076] In this embodiment, the historical task with a similarity greater than a second preset threshold to the task information can be determined as an associated task based on the historical tasks in the historical task record, and the calling interface information corresponding to the associated task in the historical task record can be determined as the interface configuration information associated with the task information.

[0077] Further, the information prompt word corresponding to each interface configuration information can be pre-stored, the model protocol server determines the similarity between the interface configuration information and the task information, determines the interface configuration information associated with the task information based on the similarity between the interface configuration information and the task information, and determines the information prompt word of the interface configuration information associated with the task information based on the stored correspondence between the interface configuration information and the information prompt word.

[0078] In addition, the information prompt word can be obtained by transforming the interface configuration information. As an example, the type of the information prompt word can include at least one of a vector type, an image type, a structured table, and a pure text type. The type of the information prompt word that can be recognized by the model protocol client can be taken as a target type. The interface configuration information associated with the task information is converted into target information of the target type, and the target information is taken as the information prompt word corresponding to the interface configuration information.

[0079] The embodiments provide different ways to determine the information prompt word, which facilitates flexible and effective determination of the information prompt word of the interface configuration information.

[0080] Optionally, the interface configuration information includes an interface configuration vector, and the interface configuration vector is stored in a vector library. Based on the received task information sent by the model protocol client and the interface configuration information, the information prompt word of the interface configuration information associated with the task information is determined, including: sending the received task information sent by the model protocol client to the vector library, so that the vector library determines the interface configuration vector associated with the task information based on the similarity between at least one interface configuration vector and the task information, determines the information prompt word of the interface configuration vector associated with the task information, and receives the information prompt word fed back by the vector library.

[0081] The vector library is a database system for storing, retrieving and managing vector data. Through the vector library, complex tasks such as semantic search, similarity recommendation and transmembrane state alignment can be performed. By storing the interface configuration vector corresponding to each business interface in the vector library, the interface configuration vector can be searched and similarity recommended.

[0082] In a specific implementation, the task information can be sent to the vector library. The similarity between the interface configuration vector and the task information is determined by the vector library. Specifically, the task information is in natural language format, which can be converted into a task vector in vector form by a pre-trained model. The pre-trained model can be a BERT (Bidirectional Encoder Representations from Transformers) model. The similarity between each interface configuration vector and the task vector can be determined by cosine similarity and / or Euclidean distance, as the similarity between the interface configuration vector and the task information. The higher the similarity, the higher the matching degree between the interface configuration vector and the task information; otherwise, the matching degree is lower.

[0083] In this embodiment, the interface configuration vector with a similarity greater than a first preset degree threshold can be used as the interface configuration vector associated with the task information; or a preset number of interface configuration vectors with the maximum similarity can be used as the interface configuration vectors associated with the task information. It should be noted that the first preset degree threshold and the preset number can be set according to actual application by those skilled in the art, and this embodiment is not limited.

[0084] Further, the information prompt word corresponding to the interface configuration vector can be determined based on the interface configuration vector associated with the task information. For example, the interface configuration vector can be directly used as the information prompt word; or the interface configuration vector can be converted into natural language information, and the information can be used as the information prompt word of the interface configuration vector. The determined information prompt word associated with the task information can be sent to the model protocol server from the vector library, and the received information prompt word can be sent to the model protocol client by the model protocol server. It should be noted that the format of the information prompt word can be determined based on the type of information that can be recognized by the model protocol client.

[0085] In this embodiment, the interface configuration vector is used to determine the information prompt word associated with the task information, so that the unstructured text is converted into a mathematical representation in a high-dimensional semantic space. Even if the same semantics but different vocabulary expressions are encountered, the association between the information prompt word and the task information can still be accurately established, which is beneficial to accurately and effectively determining the information prompt word associated with the task information.

[0086] S220, send the information prompt word to the model protocol client, so that the model protocol client inputs the information prompt word into the task execution model, determines the calling interface information corresponding to the task information based on the second output result output by the task execution model; receive the calling interface information fed back by the model protocol client.

[0087] As Figure 4 shown, after the model protocol server determines the information prompt word, the determined task information and the information prompt word associated with the task information are fed back to the model protocol client. The obtained information prompt word and task information are input into the task execution model through the model protocol client. The task execution model can obtain a second output result based on the task information and the information prompt word. The second output result includes calling interface information, which includes a calling interface address and / or interface input parameters corresponding to the calling interface address. By analyzing the second output result, the calling interface information can be determined, and the model protocol client can send the calling interface information to the model protocol server.

[0088] Exemplarily, the interface configuration information includes an interface address, the calling interface address can be an interface address corresponding to each input information prompt word; or, the calling interface address can be an interface address corresponding to an information prompt word that satisfies a preset matching condition after the information prompt word is matched with the task information through the task execution model. The preset matching condition can be that the similarity between the information prompt word and the task information is greater than a second preset degree threshold.

[0089] S230, based on the calling interface information, access the calling business party corresponding to the calling interface information to obtain associated business data related to the task information from the business data of the calling business party.

[0090] It should be noted that, since the amount of business data corresponding to the business party is large, using enumeration to determine the interface input parameters, or using the interface input parameters used in the historical access as the interface input parameters required for the current access, has certain randomness, i.e., the association between the obtained associated business data and the task information is weak, which causes the task execution model to be unable to accurately obtain the response information based on the associated business data. Therefore, in order to improve the processing efficiency and accuracy of the task execution model, the business data corresponding to the calling business party can be further filtered to filter out business data with strong association with the task information as the associated business data.

[0091] Optionally, the calling interface information corresponding to the task information includes a calling interface address and interface input parameters corresponding to the calling interface address. That is, through the second output result, the interface input parameters for requesting the associated business data of the calling business party can be directly determined.

[0092] In this embodiment, based on the calling interface information, an access is made to a calling business party corresponding to the calling interface information to obtain the specific implementation manner of the associated business data related to the task information from the business data of the calling business party, which includes: generating a data acquisition request corresponding to the task information based on the interface input parameter corresponding to the calling interface address; sending the data acquisition request to the calling interface corresponding to the calling interface address, so that the calling business party to which the calling interface belongs determines the associated business data based on the received data acquisition request, and receives the associated business data sent by the calling business party.

[0093] Specifically, through the interface input parameter, a data acquisition request corresponding to the task information can be generated. For example, the task information is "how to apply for cross-departmental cooperation budget?", and the interface input parameter can include the name of the application department, the name of the department that needs to be combined, and the project plan, etc.

[0094] As shown in Figure 4 , the model protocol client can generate a data acquisition request based on the interface input parameter fed back by the task execution model, and send the data acquisition request to the calling interface corresponding to the calling interface address, so that the calling business party corresponding to the calling interface obtains the business data corresponding to the data acquisition request in the business database through the data acquisition request as the associated business data. The calling business party returns the associated business data to the model protocol server, and the model protocol server returns the associated business data to the model protocol client, so as to input the associated business data and the task information into the task execution model through the model protocol client.

[0095] In this embodiment, the interface input parameter output by the task execution model generates a data acquisition request to obtain the associated business data, so as to accurately and effectively obtain the business data associated with the task information, thereby improving the accuracy of the response information.

[0096] S240, the associated business data is sent to the model protocol client, so that the model protocol client inputs the associated business data and the task information into the pre-trained task execution model, and determines the response information corresponding to the task information based on the first output result of the task execution model.

[0097] It should be noted that, in this embodiment, the response information is determined by combining the business data of different large business parties and the task execution model, and the RAG (Retrieval-Augmented Generation) method is adopted, which can retrieve information from a wide range of data sources, provide rich context for the task execution model, and better understand and handle complex problems.

[0098] As shown in Figure 4As shown, the task execution model obtains a first output result based on the received associated business data and the task information, and the model protocol client obtains response information corresponding to the task information by analyzing the first output result.

[0099] In this embodiment, the second output result output by the task execution model is combined to determine the calling interface information, the calling interface information with strong association with the task information is determined, and the associated business data associated with the task information is accurately obtained; and the associated business data associated with the task information is provided to the task execution model in the RAG mode, that is, the minimum business data is provided to the task execution model, and the full amount of business data does not need to be provided to the task execution model, so that the demand of the task execution model can be ensured while the security of the business data is improved.

[0100] Figure 5 is a flowchart of another response information generation method according to an embodiment of the application. Based on the above-mentioned embodiments, before determining the calling interface information, the embodiment further includes: generating interface registration information in response to a registration operation of a business interface of at least one business party; determining interface configuration information corresponding to the registered interface based on the interface registration information, and storing the interface configuration information. Wherein, the same or corresponding terms as in the above-mentioned embodiments are not repeated here. For example, Figure 5 As shown, the method comprises:

[0101] S310, in response to a registration operation of a business interface of at least one business party, generating interface registration information; determining interface configuration information corresponding to the registered interface based on the interface registration information, and storing the interface configuration information.

[0102] The interface configuration information includes the interface address, the interface name and the field information of the registered interface; the field information includes at least one of the parameter field name, the parameter field description, the interface type and whether it is mandatory.

[0103] In order to obtain the calling interface information in time and quickly when determining the response information, it is necessary to generate and store the interface configuration information in advance. Specifically, the registration operation of the business interface of at least one business party can be responded, wherein the registration operation includes the content filling operation of the information input box on the user interface, and the interface registration information can be determined based on the content filled in the input box. The interface registration information can include the interface address, the interface name and the field information; the field information includes at least one of the parameter field name, the parameter field description, the interface type and whether it is mandatory. The interface registration information is stored as the interface configuration information corresponding to the registered interface.

[0104] Figure 6is a schematic diagram of an MPC architecture interface configuration process provided by an embodiment of the present application, as shown in Figure 6 As shown in the figure, under the MPC architecture, configuration of two types of interfaces, RPC interface and Http interface, is supported. For the RPC interface, the interface registration information can include the RPC interface address, interface name, address of the RPC consumer SDK (Software Development Kit), and field information. For the Http interface, the interface registration information can include the Http interface address and field information, and the configuration mode can be the include mode. In order to improve the convenience of obtaining the interface registration information, the RPC interface name and field information can be automatically generated by parsing the SDK code. The interface name and field information of the Http interface can be obtained by manual filling. The field information can include description information and whether it is mandatory, and the description information can be the parameter field name and parameter field description.

[0105] The embodiment pre-registers the interface registration information, thereby obtaining the interface configuration information based on the interface registration information, and storing the interface configuration information, which facilitates subsequent calling of the business interface based on the stored interface registration information to obtain the business data.

[0106] Optionally, based on the interface registration information, the interface configuration information corresponding to the registered interface is determined, including: performing an audit operation on the interface registration information to determine an audit result of the interface registration information; in the case where the audit result is an audit pass, taking the business interface corresponding to the interface registration information as the registered interface, and taking the interface registration information as the interface configuration information of the registered interface.

[0107] In order to ensure the accuracy and compliance of the interface configuration information, and to obtain safe business data when performing a task, the interface registration information can be audited. For details, please refer to Figure 6As shown, after obtaining the interface registration information, the interface registration information can be sent to an auditing terminal, and an administrator verifies the compliance of the interface registration information and generates an audit result corresponding to the interface registration information. For example, the administrator approves to determine whether the interface registration information is permitted to use artificial intelligence technology in business operation, that is, whether the interface registration information is allowed to use artificial intelligence technology. The audit result includes audit pass and audit fail. In the case where the received audit result is audit pass, the business interface corresponding to the interface registration information can be stored as a registered interface, the interface registration information can be stored as interface configuration information, and the interface configuration information table can be obtained in the form of a table. At the same time, the interface configuration information can be converted into a vector form, and the corresponding interface configuration information table is recorded. The interface configuration information can include interface address, interface name, address of RPC consumer SDK, parameter field name, parameter field description, interface type, and whether it is mandatory. The interface configuration information table is used for information echoing; for example, when a user registers interface information next time, the user is prompted to input interface registration information whether it is registered by displaying the stored interface configuration information, so as to avoid the case of repeated registration.

[0108] In the case where the audit result is audit fail, the interface registration information can be deleted. And a prompt information is generated and fed back to the registration terminal sending the interface registration information, so that the registration terminal verifies whether the interface registration information is accurate before registration.

[0109] In the embodiment, by auditing the interface registration information, the compliance and accuracy of the interface registration information can be ensured, and when the business interface corresponding to the interface registration information is called to obtain business data, the security of the obtained business data can be ensured.

[0110] S320, based on the received task information sent by the model protocol client, determine the calling interface information from the pre-stored interface configuration information of at least one business party.

[0111] Optionally, the response information generation method provided in the embodiment is applied to a target model protocol server, the target model protocol server is a model protocol server in a distributed server system, and the domain names of each model protocol server in the distributed server system are the same.

[0112] It should be noted that, in order to reasonably utilize resources, the distributed server system can be used to implement the response information generation method. The distributed server system includes a plurality of model protocol servers, and each model protocol server can use the same domain name, thereby constructing a high-availability and scalable architecture.

[0113] In a specific implementation, after receiving the task information, the artificial intelligence server can determine a target model protocol server based on the remaining resources of each model protocol server in the distributed server system, and send the task information to the target model protocol server. For example, the model protocol server with the most remaining resources in the distributed server system can be determined as the target model protocol server.

[0114] The remaining resources can be CPU resources, memory resources, network bandwidth, and the like of the model protocol server.

[0115] In this embodiment, the target model protocol server is determined in the distributed server system based on the remaining resources, which can ensure load balancing between the model protocol servers and improve the processing efficiency of the task information.

[0116] S330, based on the calling interface information, access the calling business party corresponding to the calling interface information, to obtain the associated business data related to the task information from the business data of the calling business party.

[0117] S340, send the associated business data to the model protocol client, so that the model protocol client inputs the associated business data and the task information into the pre-trained task execution model, and determines the response information corresponding to the task information based on the first output result of the task execution model.

[0118] The embodiments of the response information generation method are described in detail above. In order to make the technical personnel in the art further clear the technical scheme of the method, the specific application scenarios are given below.

[0119] Figure 7 is a MCP core flowchart provided by an embodiment of the application, as shown in Figure 7 The user performs MCP support configuration, including registering an RPC interface or an HTTP interface. After registration, the administrator needs to approve, and after the audit is passed, it is saved to the interface configuration vector table and the interface configuration information table. The interface configuration vector stored in the interface configuration vector table and the configuration information stored in the interface configuration information table in a natural language manner are used for information echo when the interface is registered again. The user inputs task information through the AI client and submits the task information to the AI server. The AI server accepts the task information, and the task execution process is performed through the model protocol client. For example, the model protocol client includes multiple MCP Clients without specific business, such as MCP Client1, MCP Client2, …, MCP ClientN. One of the multiple MCP Clients can be determined to receive the task information. Wherein, N is a positive integer greater than 1.

[0120] Specifically, the model protocol client sends task information to the model protocol server. The model server queries the interface configuration vector table and determines the information prompt words associated with the task information, then sends these prompt words back to the model protocol client. Upon receiving the information prompt words, the model protocol client sends the task information and the prompt words to the task execution model. Based on the output of the task execution model, the model protocol client determines the interface call information and interface input parameters. The model protocol client then sends the interface call information and interface input parameters to the model protocol server. The model protocol server generates a data retrieval request based on the interface input parameters and queries the business interface through the interface call information, sending the data retrieval request to the corresponding business party to obtain the associated business data. For example, the business parties include Business Party 1, Business Party 2, ..., Business Party N. Each business party may include multiple business interfaces. Here, N is a positive integer greater than 1.

[0121] It should be noted that there may be multiple model protocol servers, each with the same domain name and the same server responsibilities, and each model protocol server is deployed in a distributed manner.

[0122] Furthermore, the model protocol server obtains the associated business data and sends it to the model protocol client. The model protocol client then inputs the associated business data, task information, and format requirements into the task execution model and returns the first output result of the task execution model. Based on the first output result, the response information corresponding to the task information is determined.

[0123] In this embodiment, multiple unbound model protocol clients are created using the MCP protocol, achieving business-level de-identification. This reduces deployment costs, facilitates maintenance and expansion, improves the reusability of the model protocol clients, avoids resource waste, and saves significant costs. Furthermore, by creating a model protocol server using the MCP protocol, which is decoupled from specific business processes, zero intrusion into the business logic is achieved. This allows business parties to obtain business data without any development effort, thereby shortening the time required for various businesses to integrate with artificial intelligence technologies and reducing human resource requirements.

[0124] Figure 8 This is a flowchart of another response information generation method provided by an embodiment of the present invention. This embodiment is applicable to situations where response information corresponding to task information is determined by combining business data. This method is applied to a model protocol client. This method belongs to the same inventive concept as the response information generation methods in the above embodiments. Details not described in detail in the embodiments of the response information generation method can be referred to the above embodiments. The method includes:

[0125] S410, generate task information, and send the task information to the model protocol server, so that the model protocol server determines the calling interface information from the pre-stored interface configuration information of at least one service party, accesses the calling service party corresponding to the calling interface information based on the calling interface information, and obtains the associated service data related to the task information from the service data of the calling service party.

[0126] The model protocol server is a server connected to the model protocol client in the model context protocol architecture.

[0127] Optionally, the task information is sent to the model protocol server, so that the model protocol server determines the calling interface information from the pre-stored interface configuration information of at least one service party. The specific implementation is that the task information is sent to the model protocol server, so that the model protocol server determines the information prompt word of the interface configuration information associated with the task information based on the received task information and the interface configuration information, sends the information prompt word to the model protocol client, receives the information prompt word, inputs the information prompt word into the task execution model, determines the calling interface information corresponding to the task information based on the second output result of the task execution model, and sends the calling interface information to the model protocol server.

[0128] In this embodiment, the calling interface information associated with the task information is determined based on the second output result of the task execution model, so that the calling interface information with strong association with the task information can be determined.

[0129] S420, receive the associated service data sent by the model protocol server, input the associated service data and the task information into the pre-trained task execution model, and determine the response information corresponding to the task information based on the first output result of the task execution model.

[0130] The technical scheme of the embodiment of the application generates task information, sends the task information to a model protocol server, so that the model protocol server determines calling interface information from pre-stored interface configuration information of at least one service party, accesses a calling service party corresponding to the calling interface information based on the calling interface information, and obtains associated service data related to the task information from service data of the calling service party, thereby being capable of obtaining the associated service data from the calling service party without configuring the calling service party as a server corresponding to a model context protocol, and further inputting the associated service data and the task information into a pre-trained task execution model by receiving the associated service data, determining response information corresponding to the task information based on a first output result of the task execution model. The scheme accesses the calling service party through the pre-stored interface configuration information, thereby obtaining the associated service data without modifying the service party, reducing workload and modification cost, and effectively reducing the number of MCP servers, facilitating maintenance and management.

[0131] Figure 9 A structural schematic diagram of a response information generation device according to an embodiment of the application is shown in the figure, which is used for executing the response information generation method applied to the model protocol server provided by any of the above embodiments. The device and the response information generation method of each of the above embodiments belong to the same inventive concept, and details not described in the embodiment of the response information generation device can be referred to the embodiment of the response information generation method. As shown in the figure, the device comprises: Figure 9

[0132] An information determination module 10 is configured to determine calling interface information from pre-stored interface configuration information of at least one service party based on task information sent by a model protocol client;

[0133] A data acquisition module 11 is configured to access a calling service party corresponding to the calling interface information based on the calling interface information, so as to obtain associated service data related to the task information from service data of the calling service party;

[0134] A data sending module 12 is configured to send the associated service data to the model protocol client, so that the model protocol client inputs the associated service data and the task information into a pre-trained task execution model, and determines response information corresponding to the task information based on a first output result of the task execution model.

[0135] Optionally, the information determination module 10 comprises:

[0136] ​The information prompt word determination submodule is configured to determine an information prompt word of the interface configuration information associated with the task information based on the received task information sent by the model protocol client and the interface configuration information.

[0137] The prompt word sending submodule is configured to send the information prompt word to the model protocol client, so that the model protocol client inputs the information prompt word into the task execution model, and determines the calling interface information corresponding to the task information based on a second output result output by the task execution model.

[0138] The first information interface submodule is configured to receive the calling interface information fed back by the model protocol client.

[0139] Optionally, in any of the optional technical solutions in the embodiments of the present application, the interface configuration information comprises an interface configuration vector, and the interface configuration vector is stored in a vector library.

[0140] The information prompt word determination submodule comprises:

[0141] The task sending unit is configured to send the received task information sent by the model protocol client to the vector library, so that the vector library determines the interface configuration vector associated with the task information based on the similarity between at least one interface configuration vector and the task information, and determines the information prompt word of the interface configuration vector associated with the task information.

[0142] The prompt word interface unit is configured to receive the information prompt word fed back by the vector library.

[0143] Optionally, in any of the optional technical solutions in the embodiments of the present application, the calling interface information corresponding to the task information comprises a calling interface address and an interface input parameter corresponding to the calling interface address.

[0144] The data acquisition module 11 comprises:

[0145] The request generation unit is configured to generate a data acquisition request corresponding to the task information based on the interface input parameter corresponding to the calling interface address.

[0146] The request sending unit is configured to send the data acquisition request to the calling interface corresponding to the calling interface address, so that a calling business party to which the calling interface belongs determines associated business data based on the received data acquisition request, and receives the associated business data sent by the calling business party.

[0147] Optionally, in any of the optional technical solutions in the embodiments of the present application, the data sending module 12 comprises:

[0148] The response information determination submodule is configured to send the associated business data to the model protocol client, so that the model protocol client sends the associated business data, the task information, and preset format requirement information to the task execution model to obtain a first output result corresponding to the task information, and takes the first output result as the response information corresponding to the task information.

[0149] The format requirement information records a corresponding relationship between the task type and the response format.

[0150] Optionally, the application is applied to a target model protocol server, and the target model protocol server is a model protocol server in a distributed server system.

[0151] Optionally, the application further comprises:

[0152] The registration interface information generation module is configured to, before determining the calling interface information based on the received task information sent by the model protocol client and the pre-stored interface configuration information of at least one business party, generate interface registration information in response to a registration operation on the business interface of the at least one business party.

[0153] The information storage module is configured to determine the interface configuration information corresponding to the registered interface based on the interface registration information, and store the interface configuration information.

[0154] The interface configuration information comprises an interface address, an interface name, and field information of the registered interface.

[0155] Optionally, the information storage module comprises:

[0156] The audit submodule is configured to perform an audit operation on the interface registration information, and determine an audit result of the interface registration information.

[0157] The configuration information determination submodule is configured to, in a case where the audit result is an audit pass, take the business interface corresponding to the interface registration information as the registered interface, and take the interface registration information as the interface configuration information of the registered interface.

[0158] The response information generation device provided in the embodiments of the application can execute the response information generation method provided in any of the embodiments of the application, and has the corresponding functional modules and beneficial effects of the execution method.

[0159] It is worth noting that the above-mentioned embodiments of the response information generation device include various units and modules only according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be realized; in addition, the specific name of each functional unit is only for easy mutual differentiation, and does not limit the protection scope of the present application.

[0160] Figure 10 Another structural schematic diagram of a response information generation device according to an embodiment of the present application is provided; the device is used to execute the response information generation method applied to the model protocol client provided by any of the above-mentioned embodiments. The device and the response information generation method of each of the above-mentioned embodiments belong to the same inventive concept, and the details not described in the embodiment of the response information generation device can be referred to the embodiment of the response information generation method. As shown in the above-mentioned embodiment, the device comprises: Figure 10

[0161] The information generation module 20 is configured to generate task information and send the task information to the model protocol server, so that the model protocol server determines the calling interface information from the pre-stored interface configuration information of at least one service provider, accesses the calling service provider corresponding to the calling interface information based on the calling interface information, and obtains the associated service data related to the task information from the service data of the calling service provider; wherein the model protocol server is a server connected to the model protocol client in the model context protocol architecture;

[0162] The data receiving module 21 is configured to receive the associated service data sent by the model protocol server, input the associated service data and the task information into the pre-trained task execution model, and determine the response information corresponding to the task information based on the first output result of the task execution model.

[0163] In any optional technical solution in the embodiments of the present application, the information generation module 20 can optionally comprise:

[0164] The task sending sub-module is configured to send the task information to the model protocol server, so that the model protocol server determines the information prompt word of the interface configuration information associated with the task information based on the received task information and the interface configuration information, and sends the information prompt word to the model protocol client;

[0165] The prompt word receiving sub-module is configured to receive the information prompt word, input the information prompt word into the task execution model, determine the calling interface information corresponding to the task information based on the second output result output by the task execution model, and send the calling interface information to the model protocol server.

[0166] ​The response information generation apparatus provided by the embodiments of the present application can execute the response information generation method provided by any of the embodiments of the present application, and has the function modules and beneficial effects corresponding to the execution method.

[0167] Figure 11 is a structural schematic diagram of a model protocol server for implementing the embodiments of the present application. As shown in Figure 11 The model protocol server 30 includes at least one first processor 31, and a first memory, such as a first read-only memory 32, a first random access memory 33, etc., which is communicatively connected to the at least one first processor 31, wherein the first memory stores a computer program executable by the at least one processor, and the first processor 31 can execute various appropriate actions and processes according to the computer program stored in the first read-only memory 32 or loaded into the first random access memory 33 from the first storage unit 38. In the first random access memory 33, various programs and data required for the operation of the model protocol server 30 can also be stored. The first processor 31, the first read-only memory 32, and the first random access memory 33 are connected to each other through a first bus 34. A first input / output interface 35 is also connected to the first bus 34.

[0168] A plurality of components in the model protocol server 30 are connected to the first input / output interface 35, including: a first input unit 36, such as a keyboard, a mouse, etc.; a first output unit 37, such as various types of displays, speakers, etc.; a first storage unit 38, such as a magnetic disk, an optical disk, etc.; and a first communication unit 39, such as a network card, a modem, a wireless communication transceiver, etc. The first communication unit 39 allows the model protocol server 30 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0169] The first processor 31 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the first processor 31 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The first processor 31 executes various methods and processes described above, such as the response information generation method.

[0170] In some embodiments, the response information generation method can be implemented as a computer program tangibly embodied in a computer readable storage medium, e.g., the first storage unit 38. In some embodiments, portions or all of the computer program can be loaded onto and / or installed on the model protocol server 30 via the first read only memory 32 and / or the first communication unit 39. One or more steps of the response information generation method described above can be performed when the computer program is loaded into the first random access memory 33 and executed by the first processor 31. Alternatively, in other embodiments, the first processor 31 can be configured, by any suitable means (e.g., by way of firmware), to perform the response information generation method.

[0171] Figure 12 is a structural schematic diagram of a model protocol client implementing embodiments of the present application. As shown in Figure 12 The model protocol client 40 includes at least one second processor 41, and a second memory, e.g., a second read only memory 42, a second random access memory 43, etc., communicatively connected to the at least one second processor 41, where the second memory stores a computer program executable by the at least one processor, and the second processor 41 can perform various suitable actions and processes according to the computer program stored in the second read only memory 42 or loaded into the second random access memory 43 from the second storage unit 48. Various programs and data required for operation of the model protocol client 40 can also be stored in the second random access memory 43. The second processor 41, the second read only memory 42, and the second random access memory 43 are connected to each other through a second bus 44. A second input / output interface 45 is also connected to the second bus 44.

[0172] Various components in the model protocol client 40 are connected to the second input / output interface 45, including: a second input unit 46, e.g., a keyboard, a mouse, etc.; a second output unit 47, e.g., various types of displays, speakers, etc.; a second storage unit 48, e.g., a magnetic disk, an optical disk, etc.; and a second communication unit 49, e.g., a network card, a modem, a wireless communication transceiver, etc. The second communication unit 49 allows the model protocol client 40 to exchange information / data with other devices through a computer network, e.g., the Internet, and / or various telecommunication networks.

[0173] The second processor 41 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the second processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, and the like. The second processor 41 performs various methods and processes described above, such as the response information generation method.

[0174] In some embodiments, the response information generation method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the second storage unit 48. In some embodiments, part or all of the computer program can be loaded and / or installed onto the model protocol client 40 via the second read only memory 42 and / or the second communication unit 49. When the computer program is loaded onto the second random access memory 43 and executed by the second processor 41, one or more steps of the response information generation method described above can be performed. Alternatively, in other embodiments, the second processor 41 can be configured, by any other appropriate means (e.g., by means of firmware), to perform the response information generation method.

[0175] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a complex programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0176] Computer programs used to implement the methods of the application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program, when executed by the processor of the machine, implements the functions / acts specified in the flowcharts and / or block diagrams. The computer program can be executed entirely on a machine, partially on a machine, partially on a machine as a stand-alone software package, or entirely on a remote machine or server.

[0177] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0178] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0179] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. A server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0180] In particular, the processes described above with reference to the flowcharts can be implemented as a computer software program according to embodiments of the present application. For example, embodiments of the present application include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for executing the methods illustrated by the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network by a communication unit, or installed from a storage unit, or installed from a ROM. When the computer program is executed by a processor, the above-described functions defined in the methods of the embodiments of the present application are executed.

[0181] The computer program product can be implemented in one or more computer programs comprising computer program code executing on or across the processing unit(s) of the computer. The computer program code can comprise one or both of object-oriented and conventional programmatic code implementing the methods as discussed above and / or below. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0182] It should be understood that the various forms of flow shown above can be re-ordered, added to, or have steps deleted, using the steps of the flow. For example, the steps recited in the present application can be performed in parallel, in series, or in a different order, without limitation herein, as long as the desired results of the technical solutions of the present application are achieved.

[0183] The specific embodiments described above are not intended to limit the scope of the present application. Those skilled in the art will understand that various modifications, combinations, sub-combinations, and alternatives of the embodiments described above can be made within the spirit and scope of the present application. Any modifications, equivalent substitutions, improvements, and the like, which are made without departing from the spirit and principles of the present application, are intended to be included in the scope of the present application.

Claims

1. A response information generation method characterized by comprising: Applied to a model protocol server, the method comprises: Based on the received task information sent by the model protocol client, determine the calling interface information from the pre-stored interface configuration information of at least one service provider; wherein the model protocol client is a client connected to the model protocol server in the model context protocol architecture; the interface configuration information includes an interface configuration vector, which is stored in a vector library; Based on the calling interface information, access the calling service provider corresponding to the calling interface information to obtain the associated business data related to the task information from the business data of the calling service provider; Based on the business type of the calling service provider, the associated business data corresponding to the calling service provider is subjected to data format verification; The associated business data subjected to data format verification is sent to the model protocol client, so that the model protocol client inputs the associated business data subjected to data format verification and the task information into the pre-trained task execution model, and determines the response information corresponding to the task information based on the first output result of the task execution model; Wherein, the calling interface information is determined from the pre-stored interface configuration information of at least one service provider based on the received task information sent by the model protocol client, comprising: Send the received task information sent by the model protocol client to the vector library, so that the vector library determines the interface configuration vector associated with the task information based on the similarity between at least one interface configuration vector and the task information, and determines the information prompt word of the interface configuration vector associated with the task information; wherein the type of the information prompt word includes at least one of vector type, image type, structured table and pure text type; Receive the information prompt word fed back by the vector library; Send the information prompt word to the model protocol client, so that the model protocol client inputs the information prompt word into the task execution model, and determines the calling interface information corresponding to the task information based on the second output result output by the task execution model; Receive the calling interface information fed back by the model protocol client.

2. The method of claim 1, wherein, The calling interface information corresponding to the task information includes a calling interface address and an interface input parameter corresponding to the calling interface address; Based on the calling interface information, access the calling service provider corresponding to the calling interface information to obtain the associated business data related to the task information from the business data of the calling service provider, comprising: Based on the interface input parameter corresponding to the calling interface address, generate a data acquisition request corresponding to the task information; Send the data acquisition request to the calling interface corresponding to the calling interface address, so that the calling service provider to which the calling interface belongs determines the associated business data based on the received data acquisition request, and receives the associated business data sent by the calling service provider.

3. The method of claim 1, wherein, The data format-verified associated business data is sent to the model protocol client, so that the model protocol client inputs the data format-verified associated business data and the task information into a pre-trained task execution model, determines response information corresponding to the task information based on a first output result of the task execution model, and includes: The data format-verified associated business data is sent to the model protocol client, so that the model protocol client inputs the data format-verified associated business data, the task information and preset format requirement information into the task execution model, obtains a first output result corresponding to the task information, and takes the first output result as response information corresponding to the task information; The format requirement information records the corresponding relationship between the task type and the response format.

4. The method of claim 1, wherein, The application is applied to a target model protocol server, which is a model protocol server in a distributed server system, and the domain names of each model protocol server in the distributed server system are the same.

5. The method of claim 1, wherein, Before the calling interface information is determined from the pre-stored interface configuration information of at least one business party based on the received task information sent by the model protocol client, the method further includes: In response to a registration operation of a business interface of at least one business party, interface registration information is generated; Based on the interface registration information, interface configuration information corresponding to the registered interface is determined, and the interface configuration information is stored; The interface configuration information includes the interface address, interface name and field information of the registered interface; the field information includes at least one of the parameter field name, parameter field description, interface type and whether it is mandatory.

6. The method of claim 5, wherein, The method includes: An audit operation is performed on the interface registration information to determine an audit result of the interface registration information; In the case where the audit result is an audit pass, the business interface corresponding to the interface registration information is taken as a registered interface, and the interface registration information is taken as the interface configuration information of the registered interface.

7. A response information generation method characterized by comprising: The application is applied to a model protocol client, and the method includes: Task information is generated, and the task information is sent to a model protocol server, so that the model protocol server determines calling interface information from pre-stored interface configuration information of at least one business party, accesses a calling business party corresponding to the calling interface information based on the calling interface information, obtains associated business data related to the task information from business data of the calling business party, and performs data format verification on the associated business data corresponding to the calling business party based on the business type of the calling business party; the model protocol server is a server connected to the model protocol client in a model context protocol architecture. receive the associated business data validated by the model protocol server, input the associated business data validated by the data format and the task information into the pre-trained task execution model, determine the response information corresponding to the task information based on the first output result of the task execution model; The task information is sent to the model protocol server, so that the model protocol server determines the calling interface information from the pre-stored interface configuration information of at least one business party. The task information is sent to the model protocol server, so that the model protocol server determines the information prompt word of the interface configuration information associated with the task information based on the received task information and the interface configuration information, and sends the information prompt word to the model protocol client; wherein the interface configuration information includes an interface configuration vector, and the interface configuration vector is stored in a vector library; Receive the information prompt word, input the information prompt word into the task execution model, determine the calling interface information corresponding to the task information based on the second output result of the task execution model, and send the calling interface information to the model protocol server. The information prompt word of the interface configuration information associated with the task information is determined based on the received task information and the interface configuration information. The task information sent by the model protocol client is sent to the vector library, so that the vector library determines the interface configuration vector associated with the task information based on the similarity between at least one interface configuration vector and the task information, and determines the information prompt word of the interface configuration vector associated with the task information; wherein the type of the information prompt word includes at least one of vector type, image type, structured table and pure text type; Receive the information prompt word fed back by the vector library.

8. A response information generating apparatus characterized by comprising: The device is configured in the model protocol server, and the device comprises: An information determination module is configured to determine calling interface information from pre-stored interface configuration information of at least one business party based on task information sent by a model protocol client; wherein the model protocol client is a client connected to the model protocol server in the model context protocol architecture; A data acquisition module is configured to access a calling business party corresponding to the calling interface information based on the calling interface information, to obtain associated business data related to the task information from business data of the calling business party, and to perform data format verification on the associated business data corresponding to the calling business party based on the business type of the calling business party. A data sending module is configured to send the associated business data validated by the data format to the model protocol client, so that the model protocol client inputs the associated business data validated by the data format and the task information into a pre-trained task execution model, and determines response information corresponding to the task information based on a first output result of the task execution model. The information determination module comprises: An information prompt word determination submodule is configured to determine an information prompt word of interface configuration information associated with the task information based on the received task information sent by the model protocol client and the interface configuration information; the interface configuration information comprises an interface configuration vector, and the interface configuration vector is stored in a vector library; An prompt word sending submodule is configured to send the information prompt word to the model protocol client, so that the model protocol client inputs the information prompt word into the task execution model, and determines the calling interface information corresponding to the task information based on a second output result output by the task execution model; A first information interface submodule is configured to receive the calling interface information fed back by the model protocol client; The information prompt word determination submodule comprises: A task sending unit is configured to send the received task information sent by the model protocol client to the vector library, so that the vector library determines an interface configuration vector associated with the task information based on a similarity degree between at least one interface configuration vector and the task information, and determines an information prompt word of the interface configuration vector associated with the task information; An prompt word interface unit is configured to receive the information prompt word fed back by the vector library.

9. A response information generating apparatus characterized by comprising: The device is configured in a model protocol client, and the device comprises: An information generation module is configured to generate task information, and send the task information to a model protocol server, so that the model protocol server determines calling interface information from pre-stored interface configuration information of at least one business party, accesses a calling business party corresponding to the calling interface information based on the calling interface information, acquires associated business data related to the task information from business data of the calling business party, and performs data format verification on the associated business data corresponding to the calling business party based on a business type of the calling business party; the model protocol server is a server connected to the model protocol client in a model context protocol architecture; A data receiving module is configured to receive the associated business data that has passed the data format verification and sent by the model protocol server, input the associated business data that has passed the data format verification and the task information into a pre-trained task execution model, and determine response information corresponding to the task information based on a first output result of the task execution model; The information generation module comprises: A task sending submodule is configured to send the task information to a model protocol server, so that the model protocol server determines an information prompt word of interface configuration information associated with the task information based on the received task information and the interface configuration information, and sends the information prompt word to the model protocol client; the interface configuration information comprises an interface configuration vector, and the interface configuration vector is stored in a vector library; The prompt word receiving submodule is configured to receive the information prompt word, input the information prompt word into the task execution model, determine the calling interface information corresponding to the task information based on a second output result output by the task execution model, and send the calling interface information to the model protocol server. The information prompt word of the interface configuration information associated with the task information is determined based on the received task information and the interface configuration information, and includes: The task information sent by the model protocol client is sent to the vector library, so that the vector library determines the interface configuration vector associated with the task information based on the similarity between at least one interface configuration vector and the task information, and determines the information prompt word of the interface configuration vector associated with the task information. The type of the information prompt word includes at least one of a vector type, an image type, a structured table, and a pure text type. The information prompt word fed back by the vector library is received.

10. A model protocol server, characterized by It includes: At least one processor; And The memory is in communication connection with the at least one processor; wherein The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the response information generation method in any one of claims 1-6.

11. A model protocol client, characterized by, It includes: At least one processor; And The memory is in communication connection with the at least one processor; wherein The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the response information generation method in any one of claims 1-6.

12. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for enabling the processor to execute the response information generation method in any one of claims 1-7 when executed.

13. A computer program product comprising a computer program which, when executed by a processor, implements the response information generation method according to any one of claims 1-7.

13. A computer program product comprising a computer program which, when executed by a processor, implements the response information generation method according to any one of claims 1-7.

Images