Active AI agent configuration method and system based on function call

Abstract

The present invention provides a method and system for configuring an active AI agent based on function call. Based on a message initiated by a user, it determines whether the user specifies a hosting AI and a binding monitoring AI, so as to respond to the user and associate an intelligent agent. It also uses an intelligent agent to call a large model tool to identify new messages subsequently initiated by the user, and further respond to the user. It also identifies topic information from new messages initiated by the user, so as to generate a corresponding questioning strategy through a large language model, actively ask questions to the user, and push matching information data to the user during the interactive communication with the user. By introducing the function call calling logic, it allows the active intelligent agent to actively initiate interaction and collect information when necessary, effectively predict user needs and intervene in interaction in advance, and realize highly integrated and automated interaction directly with the user.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and in particular to a method and system for configuring an active AI agent based on function call. Background Art

[0002] With the development of artificial intelligence technology, the application of AI agents in many fields has become more and more extensive. Especially in customer service, online consultation, automated management, etc., AI agents significantly improve operational efficiency and user satisfaction by providing functions such as question-and-answer services, information retrieval, and task execution. At present, most AI agents rely on passive answering mode, that is, users ask questions or requests, and AI agents respond according to preset rules or learned knowledge bases. The above mode performs well in simple query and instruction execution tasks, but there are obvious limitations in applications that require active interaction and complex scenarios. Although more complex dialogue management can be achieved by calling technologies such as ChatGPT through the langchain framework and related packages, the above technologies are mainly designed for responsive interaction, that is, the agent can only give feedback after receiving user input. This passive agent lacks the ability to obtain information, predict user needs, and intervene in interactions in advance, which limits its application potential in certain scenarios and makes it difficult to directly interact with users in a highly integrated and automated manner. Summary of the invention

[0003] In response to the defects of the prior art, the present invention provides an active AI agent configuration method and system based on function call, which determines whether the user specifies the hosting AI and the binding monitoring AI based on the message initiated by the user, so as to respond to the user and associate the intelligent agent; it also uses the intelligent agent to call the large model tool to identify the new message subsequently initiated by the user, and further respond to the user; it also identifies the topic information from the new message initiated by the user, and generates the corresponding questioning strategy through the large language model, actively asks questions to the user, and pushes matching information data to the user during the interactive communication with the user. By introducing the function call calling logic, it allows the active intelligent agent to actively initiate interaction and collect information when necessary, effectively predict user needs and intervene in interaction in advance, and realize highly integrated and automated interaction with the user.

[0004] The present invention provides an active AI agent configuration method based on function call, comprising the following steps:

[0005] Step S1, analyzing the message initiated by the user to determine whether the user specifies the hosting AI; if the user specifies the hosting AI, the message is identified using a large language model, and the corresponding large model tool is called to return a corresponding response to the user;

[0006] Step S2, when the user does not specify a hosting AI, determining whether the user is bound to a monitoring AI; when the user is bound to a monitoring AI, using the monitoring AI to perform text recognition on the message to determine whether the message is associated with an intelligent agent;

[0007] Step S3, when the message is associated with an agent, the agent calls the corresponding large model tool to identify the new message subsequently initiated by the user, so as to return a corresponding reply to the user; when the message is not associated with an agent, the message subsequently initiated by the user is identified to obtain the corresponding topic information;

[0008] Step S4, based on the topic information, generate a corresponding questioning strategy through the large language model, and based on the questioning strategy, actively ask questions to the user; based on the user's reply message to the questioning operation, interact with the user, and push matching information data to the user during the interactive communication process.

[0009] In one embodiment disclosed in the present application, in step S1, the message initiated by the user is analyzed to determine whether the user specifies the hosting AI; when the user specifies the hosting AI, the message is identified using a large language model, and the corresponding large model tool is called to return a corresponding response to the user, including:

[0010] Performing text analysis on the message initiated by the user to obtain the vocabulary and symbol content information contained in the message; performing semantic fuzzy comparison on the vocabulary and symbol content information and the preset trusteeship AI specified instruction to determine whether the vocabulary and symbol content information contains the preset trusteeship AI specified instruction; if so, determining that the user specifies the trusteeship AI; if not, determining that the user does not specify the trusteeship AI;

[0011] When the user specifies a hosted AI, the large language model is used to perform semantic recognition on the message, and based on the result of the semantic recognition, a function call operation is performed to call the corresponding large model tool, and then the large model tool is used to return a corresponding response to the user based on the historical interaction data of the SQL database.

[0012] In one embodiment disclosed in the present application, in step S2, when the user does not specify a hosting AI, it is determined whether the user is bound to a monitoring AI; when the user is bound to a monitoring AI, the monitoring AI is used to perform text recognition on the message to determine whether the message is associated with an intelligent agent, including:

[0013] When the user does not specify a hosting AI, the user's identity information is compared with the monitoring AI binding directory. If the identity information exists in the monitoring AI binding directory, it is determined that the user has been bound to the monitoring AI; otherwise, it is determined that the user has not been bound to the monitoring AI;

[0014] When the user has bound the monitoring AI, the monitoring AI is used to perform text recognition on the message to obtain the text content of the message; based on the text keywords contained in the text content, it is determined whether the message is associated with the corresponding type of intelligent agent.

[0015] In one embodiment disclosed in the present application, in step S3, when the message is associated with an agent, the agent calls a corresponding large model tool to identify new messages subsequently initiated by the user, thereby returning a corresponding response to the user; when the message is not associated with an agent, the message subsequently initiated by the user is identified to obtain corresponding topic information, including:

[0016] When the message is associated with the agent, all new messages initiated by the user in the subsequent preset time period are obtained, and the agent is used to perform a function call operation to call the corresponding large model tool to identify all new messages, so as to return a corresponding response to the user based on the historical interaction data of the SQL database;

[0017] When the message is not associated with an intelligent agent, all new messages initiated by the user within a subsequent preset time period are obtained, and all new messages are formatted and saved in a SQL database as historical interaction data; and language logic recognition is performed on all new messages to obtain corresponding topic information.

[0018] In one embodiment disclosed in the present application, in the step S4, based on the topic information, a corresponding questioning strategy is generated by the large language model, and based on the questioning strategy, a questioning operation is actively performed to the user; based on the user's reply message to the questioning operation, an interactive communication is performed with the user, and matching information data is pushed to the user during the interactive communication process, including:

[0019] Based on the topic information, a corresponding questioning strategy is generated through the large language model; wherein the questioning strategy includes a plurality of questions; the questioning strategy is compared with the SQL database, and the questions that have been asked in the questioning strategy are eliminated; and based on the logical relationship between all the questions included in the questioning strategy, a questioning operation is actively performed to the user;

[0020] Semantic recognition is performed on the reply message of the user to the question operation, and the question association expansion is performed based on the result of the semantic recognition, so as to interact with the user and push matching information data to the user during the idle time of the interactive communication process.

[0021] The present invention also provides an active AI agent configuration system based on function call, including:

[0022] The AI ​​hosting status judgment module is used to analyze the message initiated by the user and determine whether the user specifies the hosting AI;

[0023] The first response processing module is used to use the large language model to identify the message when the user specifies the hosting AI, and call the corresponding large model tool to return a corresponding response to the user;

[0024] A monitoring AI binding status judgment module, used to judge whether the user is bound to the monitoring AI when the user has not specified the hosting AI;

[0025] An agent association status judgment module is used to use the monitoring AI to perform text recognition on the message when the user is bound to the monitoring AI to determine whether the message is associated with the agent;

[0026] The second response processing module is used for, when the message is associated with an agent, using the agent to call the corresponding large model tool to identify new messages subsequently initiated by the user, so as to return a corresponding response to the user;

[0027] A subject information identification module, used to identify the message subsequently initiated by the user and obtain the corresponding subject information when the message is not associated with an agent;

[0028] The active questioning and data push execution module is used to generate a corresponding questioning strategy based on the topic information through the large language model, and actively ask questions to the user based on the questioning strategy; based on the user's reply message to the questioning operation, interact with the user and push matching information data to the user during the interactive communication process.

[0029] In one embodiment disclosed in the present application, the AI ​​hosting status judgment module is used to analyze the message initiated by the user to determine whether the user specifies the hosting AI, including:

[0030] Performing text analysis on the message initiated by the user to obtain the vocabulary and symbol content information contained in the message; performing semantic fuzzy comparison on the vocabulary and symbol content information and the preset trusteeship AI specified instruction to determine whether the vocabulary and symbol content information contains the preset trusteeship AI specified instruction; if so, determining that the user specifies the trusteeship AI; if not, determining that the user does not specify the trusteeship AI;

[0031] The first response processing module is used to use the large language model to identify the message when the user specifies the hosting AI, and call the corresponding large model tool to return a corresponding response to the user, including:

[0032] When the user specifies a hosted AI, the large language model is used to perform semantic recognition on the message, and based on the result of the semantic recognition, a function call operation is performed to call the corresponding large model tool, and then the large model tool is used to return a corresponding response to the user based on the historical interaction data of the SQL database.

[0033] In one embodiment disclosed in the present application, the monitoring AI binding status judgment module is used to judge whether the user is bound to the monitoring AI when the user does not specify the hosting AI, including:

[0034] When the user does not specify a hosting AI, the user's identity information is compared with the monitoring AI binding directory. If the identity information exists in the monitoring AI binding directory, it is determined that the user has been bound to the monitoring AI; otherwise, it is determined that the user has not been bound to the monitoring AI;

[0035] The agent association status judgment module is used to use the monitoring AI to perform text recognition on the message when the user is bound to the monitoring AI, and to judge whether the message is associated with the agent, including:

[0036] When the user has bound the monitoring AI, the monitoring AI is used to perform text recognition on the message to obtain the text content of the message; based on the text keywords contained in the text content, it is determined whether the message is associated with the corresponding type of intelligent agent.

[0037] In one embodiment disclosed in the present application, the second response processing module is used to use the agent to call the corresponding large model tool to identify the new message subsequently initiated by the user when the message is associated with the agent, so as to return a corresponding response to the user, including:

[0038] When the message is associated with the agent, all new messages initiated by the user in the subsequent preset time period are obtained, and the agent is used to perform a function call operation to call the corresponding large model tool to identify all new messages, so as to return a corresponding response to the user based on the historical interaction data of the SQL database;

[0039] The subject information identification module is used to identify the message subsequently initiated by the user when the message is not associated with the agent, and obtain the corresponding subject information, including:

[0040] When the message is not associated with an intelligent agent, all new messages initiated by the user within a subsequent preset time period are obtained, and all new messages are formatted and saved in a SQL database as historical interaction data; and language logic recognition is performed on all new messages to obtain corresponding topic information.

[0041] In one embodiment disclosed in the present application, the active questioning and data push execution module is used to generate a corresponding questioning strategy based on the topic information through the large language model, and actively ask questions to the user based on the questioning strategy; based on the user's reply message to the questioning operation, interact with the user, and push matching information data to the user during the interactive communication process, including:

[0042] Based on the topic information, a corresponding questioning strategy is generated through the large language model; wherein the questioning strategy includes a plurality of questions; the questioning strategy is compared with the SQL database, and the questions that have been asked in the questioning strategy are eliminated; and based on the logical relationship between all the questions included in the questioning strategy, a questioning operation is actively performed to the user;

[0043] Semantic recognition is performed on the reply message of the user to the question operation, and the question association expansion is performed based on the result of the semantic recognition, so as to interact with the user and push matching information data to the user during the idle time of the interactive communication process.

[0044] Compared with the existing technology, the active AI agent configuration method and system based on function call determines whether the user specifies the hosting AI and the binding monitoring AI based on the message initiated by the user, so as to respond to the user and associate the intelligent agent; it also uses the intelligent agent to call the large model tool to identify the new messages subsequently initiated by the user, and further respond to the user; it also identifies the topic information from the new message initiated by the user, and generates the corresponding questioning strategy through the large language model, actively asks questions to the user, and pushes matching information data to the user during the interactive communication with the user. By introducing the function call calling logic, it allows the active intelligent agent to actively initiate interaction and collect information when necessary, effectively predict user needs and intervene in interaction in advance, and realize highly integrated and automated interaction directly with the user.

[0045] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0046] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 A flowchart of the active AI agent configuration method based on function call provided by the present invention.

[0049] Figure 2 A schematic diagram of the framework of the active AI agent configuration system based on function call provided by the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] See also Figure 1 , is a flow chart of a method for configuring an active AI agent based on function call provided in an embodiment of the present invention. The method for configuring an active AI agent based on function call includes:

[0052] Step S1, analyzing the message initiated by the user to determine whether the user specifies the hosting AI; if the user specifies the hosting AI, the message is recognized by using the large language model, and the corresponding large model tool is called to return the corresponding response to the user;

[0053] Step S2: when the user does not specify a hosting AI, determine whether the user is bound to a monitoring AI; when the user is bound to a monitoring AI, use the monitoring AI to perform text recognition on the message to determine whether the message is associated with an intelligent agent;

[0054] Step S3, when the message is associated with an agent, the agent calls the corresponding large model tool to identify new messages subsequently initiated by the user, so as to return a corresponding response to the user; when the message is not associated with an agent, the message subsequently initiated by the user is identified to obtain corresponding topic information;

[0055] Step S4, based on the topic information, a corresponding questioning strategy is generated through the large language model, and based on the questioning strategy, a questioning operation is actively performed to the user; based on the user's reply message to the questioning operation, an interactive communication is performed with the user, and matching information data is pushed to the user during the interactive communication process.

[0056] The beneficial effects of the above technical solution are as follows: the active AI agent configuration method based on function call determines whether the user specifies the hosting AI and the binding monitoring AI based on the message initiated by the user, so as to respond to the user and associate the intelligent agent; it also uses the intelligent agent to call the large model tool to identify the new messages subsequently initiated by the user, and further respond to the user; it also identifies the topic information from the new message initiated by the user, and generates the corresponding questioning strategy through the large language model, actively asks questions to the user, and pushes matching information data to the user during the interactive communication with the user. By introducing the function call calling logic, it allows the active intelligent agent to actively initiate interaction and collect information when necessary, effectively predict user needs and intervene in the interaction in advance, and realize highly integrated and automated interaction directly with the user.

[0057] Preferably, in step S1, the message initiated by the user is analyzed to determine whether the user specifies the hosting AI; when the user specifies the hosting AI, the message is identified using the large language model, and the corresponding large model tool is called to return a corresponding response to the user, including:

[0058] Performing text analysis on the message initiated by the user to obtain the vocabulary and symbol content information contained in the message; performing semantic fuzzy comparison on the vocabulary and symbol content information and the preset trusteeship AI specified instruction to determine whether the vocabulary and symbol content information contains the preset trusteeship AI specified instruction; if so, determining that the user specifies the trusteeship AI; if not, determining that the user does not specify the trusteeship AI;

[0059] When the user specifies a hosted AI, the large language model is used to perform semantic recognition on the message, and based on the result of the semantic recognition, a function call operation is performed to call the corresponding large model tool, which is then used to return a corresponding response to the user based on the historical interaction data in the SQL database.

[0060] The beneficial effects of the above technical solution are as follows: in the actual user-machine interaction process, the user first initiates a message (such as a text message or a voice message) to trigger the interaction with the machine. At this time, the message initiated by the user is subjected to text analysis to obtain the vocabulary and symbol content information contained in the message, so that the message can be fully fuzzy identified, and the vocabulary and symbol content information are semantically fuzzy compared with the preset managed AI specified instructions to determine whether the vocabulary and symbol content information contains the preset managed AI specified instructions, so that whether the user specifies the managed AI can be accurately identified, providing a basis for whether to use the managed AI for subsequent interaction. Also, when the user specifies the managed AI, the large language model is used to perform semantic recognition of the message, and based on the result of the semantic recognition, the function call operation is performed to call the corresponding large model tool, so that the corresponding response is returned to the user based on the historical interaction data of the SQL database, thereby improving the accuracy and real-time nature of the response to the user.

[0061] Preferably, in step S2, when the user does not specify a hosting AI, it is determined whether the user is bound to a monitoring AI; when the user is bound to a monitoring AI, the monitoring AI is used to perform text recognition on the message to determine whether the message is associated with an intelligent agent, including:

[0062] When the user does not specify a hosting AI, the user's identity information is compared with the monitoring AI binding directory. If the identity information exists in the monitoring AI binding directory, it is determined that the user has been bound to the monitoring AI; otherwise, it is determined that the user has not been bound to the monitoring AI;

[0063] When the user has bound the monitoring AI, the monitoring AI is used to perform text recognition on the message to obtain the text content of the message; based on the text keywords contained in the text content, it is determined whether the message is associated with the corresponding type of intelligent agent.

[0064] The beneficial effect of the above technical solution is: when the user has not specified a hosting AI, the user's identity information is compared with the monitoring AI binding directory to determine whether the user is bound to the monitoring AI, ensuring that the user has the corresponding AI for interactive docking under any circumstances. When the user has bound the monitoring AI, the monitoring AI is used to perform text recognition on the message to obtain the text content of the message, and based on the text keywords contained in the text content, it is determined whether the message is associated with the corresponding type of intelligent agent, so that the intelligent agent can be used to provide users with suitable interactive services.

[0065] Preferably, in step S3, when the message is associated with an agent, the agent calls a corresponding large model tool to identify new messages subsequently initiated by the user, thereby returning a corresponding response to the user; when the message is not associated with an agent, the message subsequently initiated by the user is identified to obtain corresponding topic information, including:

[0066] When the message is associated with the agent, all new messages initiated by the user in the subsequent preset time period are obtained, and the agent is used to perform function call operations to call the corresponding large model tool to identify all new messages, so as to return the corresponding response to the user based on the historical interaction data of the SQL database;

[0067] When the message is not associated with an intelligent agent, all new messages initiated by the user within a subsequent preset time period are obtained, and all new messages are converted into a format and saved in a SQL database as historical interaction data; and language logic recognition is performed on all new messages to obtain corresponding topic information.

[0068] The beneficial effects of the above technical solution are: when the message is associated with an intelligent agent, all new messages initiated by the user in the subsequent preset time period are obtained, and the intelligent agent is used to perform a function call operation to call the corresponding large model tool to identify all new messages, so that the new messages initiated by the user in the subsequent interaction process can be processed in a timely manner. When the message is not associated with an intelligent agent, all new messages initiated by the user in the subsequent preset time period are obtained, and all new messages are formatted and saved as historical interaction data in the SQL database to enrich the data content of the SQL database, providing reliable data support for accurate responses to users in different scenarios in the future, and language logic recognition is performed on all new messages to obtain corresponding topic information, so that the topic content of interest to the user can be accurately identified, ensuring that effective chat interaction topics are provided to users.

[0069] Preferably, in step S4, based on the topic information, a corresponding questioning strategy is generated by the large language model, and based on the questioning strategy, a questioning operation is actively performed to the user; based on the user's reply message to the questioning operation, an interactive communication is performed with the user, and matching information data is pushed to the user during the interactive communication process, including:

[0070] Based on the topic information, a corresponding questioning strategy is generated through the large language model; wherein the questioning strategy includes a number of questions; the questioning strategy is compared with the SQL database, and the questions that have been asked in the questioning strategy are eliminated; and based on the logical relationship between all the questions included in the questioning strategy, a questioning operation is actively performed on the user;

[0071] The user's reply message to the question operation is semantically recognized, and the question association expansion is performed based on the result of the semantic recognition, so as to interact with the user and push matching information data to the user during the idle time of the interactive communication process.

[0072] The beneficial effects of the above technical solution are: based on the subject information, the corresponding questioning strategy is generated through the large language model, and the question design can be performed for the user to actively ask questions, ensuring that useful information is obtained in the process of actively asking questions to the user. The questioning strategy is also compared with the SQL database, and the questions that have been asked in the questioning strategy are eliminated. Then, based on the logical relationship between all the questions contained in the questioning strategy, the user is actively asked questions, thereby improving the effectiveness and accuracy of the user's active questions. In addition, the user's reply message to the question operation is semantically recognized, and the question association is expanded based on the result of the semantic recognition, so as to interact with the user, and push matching information data to the user during the idle time of the interactive communication process, effectively predicting user needs and the ability to intervene in the interaction in advance, and realizing highly integrated and automated interaction directly with the user.

[0073] See also Figure 2 , is a schematic diagram of a framework of an active AI agent configuration system based on function call provided by an embodiment of the present invention. The active AI agent configuration system based on function call includes:

[0074] The AI ​​hosting status judgment module is used to analyze the message initiated by the user and determine whether the user specifies the hosting AI;

[0075] The first response processing module is used to use the large language model to identify the message when the user specifies the hosting AI, and call the corresponding large model tool to return a corresponding response to the user;

[0076] The monitoring AI binding status judgment module is used to judge whether the user is bound to the monitoring AI when the user has not specified the hosting AI;

[0077] The intelligent agent association status judgment module is used to use the monitoring AI to perform text recognition on the message when the user is bound to the monitoring AI to determine whether the message is associated with the intelligent agent;

[0078] The second response processing module is used to use the agent to call the corresponding large model tool to identify the new message subsequently initiated by the user when the message is associated with the agent, so as to return the corresponding response to the user;

[0079] The subject information identification module is used to identify the subsequent messages initiated by the user when the message is not associated with the intelligent agent, and obtain the corresponding subject information;

[0080] The active questioning and data push execution module is used to generate a corresponding questioning strategy based on the topic information through the large language model, and actively ask questions to the user based on the questioning strategy; based on the user's reply message to the questioning operation, interact with the user, and push matching information data to the user during the interactive communication process.

[0081] The beneficial effects of the above technical solution are: the active AI agent configuration system based on function call determines whether the user specifies the hosting AI and the binding monitoring AI based on the message initiated by the user, so as to respond to the user and associate the intelligent agent; it also uses the intelligent agent to call the large model tool to identify the new messages subsequently initiated by the user, and further respond to the user; it also identifies the topic information from the new message initiated by the user, and generates the corresponding questioning strategy through the large language model, actively asks questions to the user, and pushes matching information data to the user during the interactive communication process. By introducing the function call calling logic, it allows the active intelligent agent to actively initiate interaction and collect information when necessary, effectively predict user needs and intervene in interaction in advance, and realize highly integrated and automated interaction directly with the user.

[0082] Preferably, the AI ​​hosting status judgment module is used to analyze the message initiated by the user to determine whether the user specifies the hosting AI, including:

[0083] Performing text analysis on the message initiated by the user to obtain the vocabulary and symbol content information contained in the message; performing semantic fuzzy comparison on the vocabulary and symbol content information and the preset trusteeship AI specified instruction to determine whether the vocabulary and symbol content information contains the preset trusteeship AI specified instruction; if so, determining that the user specifies the trusteeship AI; if not, determining that the user does not specify the trusteeship AI;

[0084] The first response processing module is used to use the large language model to identify the message when the user specifies the hosting AI, and call the corresponding large model tool to return a corresponding response to the user, including:

[0085] When the user specifies a hosted AI, the large language model is used to perform semantic recognition on the message, and based on the result of the semantic recognition, a function call operation is performed to call the corresponding large model tool, which is then used to return a corresponding response to the user based on the historical interaction data in the SQL database.

[0086] The beneficial effects of the above technical solution are as follows: in the actual user-machine interaction process, the user first initiates a message (such as a text message or a voice message) to trigger the interaction with the machine. At this time, the message initiated by the user is subjected to text analysis to obtain the vocabulary and symbol content information contained in the message, so that the message can be fully fuzzy identified, and the vocabulary and symbol content information are semantically fuzzy compared with the preset managed AI specified instructions to determine whether the vocabulary and symbol content information contains the preset managed AI specified instructions, so that whether the user specifies the managed AI can be accurately identified, providing a basis for whether to use the managed AI for subsequent interaction. Also, when the user specifies the managed AI, the large language model is used to perform semantic recognition of the message, and based on the result of the semantic recognition, the function call operation is performed to call the corresponding large model tool, so that the corresponding response is returned to the user based on the historical interaction data of the SQL database, thereby improving the accuracy and real-time nature of the response to the user.

[0087] Preferably, the monitoring AI binding status judgment module is used to judge whether the user is bound to the monitoring AI when the user does not specify the hosting AI, including:

[0088] When the user does not specify a hosting AI, the user's identity information is compared with the monitoring AI binding directory. If the identity information exists in the monitoring AI binding directory, it is determined that the user has been bound to the monitoring AI; otherwise, it is determined that the user has not been bound to the monitoring AI;

[0089] The agent association status judgment module is used to use the monitoring AI to perform text recognition on the message when the user is bound to the monitoring AI to determine whether the message is associated with the agent, including:

[0090] When the user has bound the monitoring AI, the monitoring AI is used to perform text recognition on the message to obtain the text content of the message; based on the text keywords contained in the text content, it is determined whether the message is associated with the corresponding type of intelligent agent.

[0091] The beneficial effect of the above technical solution is: when the user has not specified a hosting AI, the user's identity information is compared with the monitoring AI binding directory to determine whether the user is bound to the monitoring AI, ensuring that the user has the corresponding AI for interactive docking under any circumstances. When the user has bound the monitoring AI, the monitoring AI is used to perform text recognition on the message to obtain the text content of the message, and based on the text keywords contained in the text content, it is determined whether the message is associated with the corresponding type of intelligent agent, so that the intelligent agent can be used to provide users with suitable interactive services.

[0092] Preferably, the second response processing module is used to use the agent to call the corresponding large model tool to identify the new message subsequently initiated by the user when the message is associated with the agent, so as to return the corresponding response reply to the user, including:

[0093] When the message is associated with the agent, all new messages initiated by the user in the subsequent preset time period are obtained, and the agent is used to perform function call operations to call the corresponding large model tool to identify all new messages, so as to return the corresponding response to the user based on the historical interaction data of the SQL database;

[0094] The topic information identification module is used to identify the subsequent messages initiated by the user when the message is not associated with the agent, and obtain the corresponding topic information, including:

[0095] When the message is not associated with an intelligent agent, all new messages initiated by the user within a subsequent preset time period are obtained, and all new messages are converted into a format and saved in a SQL database as historical interaction data; and language logic recognition is performed on all new messages to obtain corresponding topic information.

[0096] The beneficial effects of the above technical solution are: when the message is associated with an intelligent agent, all new messages initiated by the user in the subsequent preset time period are obtained, and the intelligent agent is used to perform a function call operation to call the corresponding large model tool to identify all new messages, so that the new messages initiated by the user in the subsequent interaction process can be processed in a timely manner. When the message is not associated with an intelligent agent, all new messages initiated by the user in the subsequent preset time period are obtained, and all new messages are formatted and saved as historical interaction data in the SQL database to enrich the data content of the SQL database, providing reliable data support for accurate responses to users in different scenarios in the future, and language logic recognition is performed on all new messages to obtain corresponding topic information, so that the topic content of interest to the user can be accurately identified, ensuring that effective chat interaction topics are provided to users.

[0097] Preferably, the active questioning and data push execution module is used to generate a corresponding questioning strategy based on the topic information through the large language model, and actively ask questions to the user based on the questioning strategy; based on the user's reply message to the questioning operation, interact with the user, and push matching information data to the user during the interactive communication process, including:

[0098] Based on the topic information, a corresponding questioning strategy is generated through the large language model; wherein the questioning strategy includes a number of questions; the questioning strategy is compared with the SQL database, and the questions that have been asked in the questioning strategy are eliminated; and based on the logical relationship between all the questions included in the questioning strategy, a questioning operation is actively performed on the user;

[0099] The user's reply message to the question operation is semantically recognized, and the question association expansion is performed based on the result of the semantic recognition, so as to interact with the user and push matching information data to the user during the idle time of the interactive communication process.

[0100] The beneficial effects of the above technical solution are: based on the subject information, the corresponding questioning strategy is generated through the large language model, and the question design can be performed for the user to actively ask questions, ensuring that useful information is obtained in the process of actively asking questions to the user. The questioning strategy is also compared with the SQL database, and the questions that have been asked in the questioning strategy are eliminated. Then, based on the logical relationship between all the questions contained in the questioning strategy, the user is actively asked questions, thereby improving the effectiveness and accuracy of the user's active questions. In addition, the user's reply message to the question operation is semantically recognized, and the question association is expanded based on the result of the semantic recognition, so as to interact with the user, and push matching information data to the user during the idle time of the interactive communication process, effectively predicting user needs and the ability to intervene in the interaction in advance, and realizing highly integrated and automated interaction directly with the user.

[0101] It can be seen from the contents of the above embodiments that the active AI agent configuration method and system based on function call determines whether the user specifies the hosting AI and the binding monitoring AI based on the message initiated by the user, so as to respond to the user and associate the intelligent agent; it also uses the intelligent agent to call the large model tool to identify the new messages subsequently initiated by the user, and further respond to the user; it also identifies the topic information from the new message initiated by the user, and generates the corresponding questioning strategy through the large language model, actively asks questions to the user, and pushes matching information data to the user during the interactive communication with the user. By introducing the function call calling logic, it allows the active intelligent agent to actively initiate interaction and collect information when necessary, effectively predict user needs and intervene in the interaction in advance, and realize highly integrated and automated interaction directly with the user.

[0102] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. The active AI agent configuration method based on function call is characterized by: It includes the following steps: Step S1, analyzing the message initiated by the user to determine whether the user specifies the hosting AI; if the user specifies the hosting AI, the message is identified using a large language model, and the corresponding large model tool is called to return a corresponding response to the user; Step S2, when the user does not specify a hosting AI, determining whether the user is bound to a monitoring AI; when the user is bound to a monitoring AI, using the monitoring AI to perform text recognition on the message to determine whether the message is associated with an intelligent agent; Step S3, when the message is associated with an agent, the agent calls the corresponding large model tool to identify the new message subsequently initiated by the user, so as to return a corresponding reply to the user; when the message is not associated with an agent, the message subsequently initiated by the user is identified to obtain the corresponding topic information; Step S4, based on the topic information, generate a corresponding questioning strategy through the large language model, and based on the questioning strategy, actively ask questions to the user; based on the user's reply message to the questioning operation, interact with the user, and push matching information data to the user during the interactive communication process.

2. The active AI agent configuration method based on function call according to claim 1, characterized in that: In step S1, the message initiated by the user is analyzed to determine whether the user specifies the hosting AI; when the user specifies the hosting AI, the message is identified using a large language model, and the corresponding large model tool is called to return a corresponding response to the user, including: Performing text analysis on the message initiated by the user to obtain the vocabulary and symbol content information contained in the message; performing semantic fuzzy comparison on the vocabulary and symbol content information and the preset trusteeship AI specified instruction to determine whether the vocabulary and symbol content information contains the preset trusteeship AI specified instruction; if so, determining that the user specifies the trusteeship AI; if not, determining that the user does not specify the trusteeship AI; When the user specifies a hosted AI, the large language model is used to perform semantic recognition on the message, and based on the result of the semantic recognition, a function call operation is performed to call the corresponding large model tool, and then the large model tool is used to return a corresponding response to the user based on the historical interaction data of the SQL database.

3. The active AI agent configuration method based on function call according to claim 1, characterized in that: In step S2, when the user does not specify a hosting AI, it is determined whether the user is bound to a monitoring AI; when the user is bound to a monitoring AI, the monitoring AI is used to perform text recognition on the message to determine whether the message is associated with an intelligent agent, including: When the user does not specify a hosting AI, the user's identity information is compared with the monitoring AI binding directory. If the identity information exists in the monitoring AI binding directory, it is determined that the user has been bound to the monitoring AI; otherwise, it is determined that the user has not been bound to the monitoring AI; When the user has bound the monitoring AI, the monitoring AI is used to perform text recognition on the message to obtain the text content of the message; based on the text keywords contained in the text content, it is determined whether the message is associated with the corresponding type of intelligent agent.

4. The active AI agent configuration method based on function call according to claim 1, characterized in that: In step S3, when the message is associated with an agent, the agent calls a corresponding large model tool to identify new messages subsequently initiated by the user, thereby returning a corresponding response to the user; When the message is not associated with an agent, the subsequent message initiated by the user is identified to obtain corresponding subject information, including: When the message is associated with the agent, all new messages initiated by the user in the subsequent preset time period are obtained, and the agent is used to perform a function call operation to call the corresponding large model tool to identify all new messages, so as to return a corresponding response to the user based on the historical interaction data of the SQL database; When the message is not associated with an intelligent agent, all new messages initiated by the user within a subsequent preset time period are obtained, and all new messages are formatted and saved in a SQL database as historical interaction data; and language logic recognition is performed on all new messages to obtain corresponding topic information.

5. The active AI agent configuration method based on function call according to claim 1, characterized in that: In the step S4, based on the topic information, a corresponding questioning strategy is generated by the large language model, and based on the questioning strategy, a questioning operation is actively performed to the user; based on the user's reply message to the questioning operation, an interactive communication is performed with the user, and matching information data is pushed to the user during the interactive communication process, including: Based on the topic information, a corresponding questioning strategy is generated through the large language model; wherein the questioning strategy includes a plurality of questions; the questioning strategy is compared with the SQL database, and the questions that have been asked in the questioning strategy are eliminated; and based on the logical relationship between all the questions included in the questioning strategy, a questioning operation is actively performed to the user; Semantic recognition is performed on the reply message of the user to the question operation, and the question association expansion is performed based on the result of the semantic recognition, so as to interact with the user and push matching information data to the user during the idle time of the interactive communication process.

6. Active AI agent configuration system based on function call, characterized in that: include: The AI ​​hosting status judgment module is used to analyze the message initiated by the user and determine whether the user specifies the hosting AI; The first response processing module is used to use the large language model to identify the message when the user specifies the hosting AI, and call the corresponding large model tool to return a corresponding response to the user; A monitoring AI binding status judgment module, used to judge whether the user is bound to the monitoring AI when the user has not specified the hosting AI; An agent association status judgment module is used to use the monitoring AI to perform text recognition on the message when the user is bound to the monitoring AI to determine whether the message is associated with the agent; The second response processing module is used for, when the message is associated with an agent, using the agent to call the corresponding large model tool to identify new messages subsequently initiated by the user, so as to return a corresponding response to the user; A subject information identification module, used to identify the message subsequently initiated by the user and obtain the corresponding subject information when the message is not associated with an agent; The active questioning and data push execution module is used to generate a corresponding questioning strategy based on the topic information through the large language model, and actively ask questions to the user based on the questioning strategy; based on the user's reply message to the questioning operation, interact with the user and push matching information data to the user during the interactive communication process.

7. The active AI agent configuration system based on function call according to claim 6, characterized in that: The AI ​​hosting status judgment module is used to analyze the message initiated by the user to determine whether the user specifies the hosting AI, including: Performing text analysis on the message initiated by the user to obtain the vocabulary and symbol content information contained in the message; performing semantic fuzzy comparison on the vocabulary and symbol content information and the preset trusteeship AI specified instruction to determine whether the vocabulary and symbol content information contains the preset trusteeship AI specified instruction; if so, determining that the user specifies the trusteeship AI; if not, determining that the user does not specify the trusteeship AI; The first response processing module is used to use the large language model to identify the message when the user specifies the hosting AI, and call the corresponding large model tool to return a corresponding response to the user, including: When the user specifies a hosted AI, the large language model is used to perform semantic recognition on the message, and based on the result of the semantic recognition, a function call operation is performed to call the corresponding large model tool, and then the large model tool is used to return a corresponding response to the user based on the historical interaction data of the SQL database.

8. The active AI agent configuration system based on function call according to claim 6, characterized in that: The monitoring AI binding status judgment module is used to judge whether the user is bound to the monitoring AI when the user does not specify the hosting AI, including: When the user does not specify a hosting AI, the user's identity information is compared with the monitoring AI binding directory. If the identity information exists in the monitoring AI binding directory, it is determined that the user has been bound to the monitoring AI; otherwise, it is determined that the user has not been bound to the monitoring AI; The agent association status judgment module is used to use the monitoring AI to perform text recognition on the message when the user is bound to the monitoring AI, and to judge whether the message is associated with the agent, including: When the user has bound the monitoring AI, the monitoring AI is used to perform text recognition on the message to obtain the text content of the message; based on the text keywords contained in the text content, it is determined whether the message is associated with the corresponding type of intelligent agent.

9. The active AI agent configuration system based on function call according to claim 6, characterized in that: The second response processing module is used to use the agent to call the corresponding large model tool to identify the new message subsequently initiated by the user when the message is associated with the agent, so as to return a corresponding response to the user, including: When the message is associated with the agent, all new messages initiated by the user in the subsequent preset time period are obtained, and the agent is used to perform a function call operation to call the corresponding large model tool to identify all new messages, so as to return a corresponding response to the user based on the historical interaction data of the SQL database; The subject information identification module is used to identify the message subsequently initiated by the user when the message is not associated with the agent, and obtain the corresponding subject information, including: When the message is not associated with an intelligent agent, all new messages initiated by the user within a subsequent preset time period are obtained, and all new messages are formatted and saved in a SQL database as historical interaction data; and language logic recognition is performed on all new messages to obtain corresponding topic information.

10. The active AI agent configuration system based on function call according to claim 6, characterized in that: The active questioning and data push execution module is used to generate a corresponding questioning strategy based on the topic information through the large language model, and actively ask questions to the user based on the questioning strategy; interact with the user based on the user's reply message to the questioning operation, and push matching information data to the user during the interactive communication process, including: Based on the topic information, a corresponding questioning strategy is generated through the large language model; wherein the questioning strategy includes a plurality of questions; the questioning strategy is compared with the SQL database, and the questions that have been asked in the questioning strategy are eliminated; and based on the logical relationship between all the questions included in the questioning strategy, a questioning operation is actively performed to the user; Semantic recognition is performed on the reply message of the user to the question operation, and the question association expansion is performed based on the result of the semantic recognition, so as to interact with the user and push matching information data to the user during the idle time of the interactive communication process.

Images