system

Abstract

The system according to this embodiment aims to monitor and detect abnormal behavior of an AI agent in real time and automatically prevent it. [Solution] The system according to the embodiment comprises a collection unit, a learning unit, a monitoring unit, a detection unit, a blocking unit, and a notification unit. The collection unit collects behavioral data. The learning unit analyzes the data collected by the collection unit and learns normal behavioral patterns. The monitoring unit monitors in real time based on the normal behavioral patterns learned by the learning unit. The detection unit detects the behavior monitored by the monitoring unit. The blocking unit blocks the abnormal behavior detected by the detection unit. The notification unit notifies the relevant parties of the abnormal behavior blocked by the blocking unit.

Description

Technical Field

[0001] The technology of the present disclosure relates to a system.

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor, and includes steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a character of the chatbot, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the prior art, it is difficult to monitor and detect abnormal behaviors of an AI agent in real time and automatically prevent them, and efficient countermeasures are required.

[0005] The system according to the embodiment aims to monitor and detect abnormal behaviors of an AI agent in real time and automatically prevent them.

Means for Solving the Problems

[0006] The system according to the embodiment comprises a collection unit, a learning unit, a monitoring unit, a detection unit, a blocking unit, and a notification unit. The collection unit collects behavioral data. The learning unit analyzes the data collected by the collection unit and learns normal behavioral patterns. The monitoring unit monitors in real time based on the normal behavioral patterns learned by the learning unit. The detection unit detects behavior monitored by the monitoring unit. The blocking unit blocks abnormal behavior detected by the detection unit. The notification unit notifies relevant parties of abnormal behavior blocked by the blocking unit. [Effects of the Invention]

[0007] The system according to this embodiment can monitor and detect abnormal behavior of an AI agent in real time and automatically prevent it. [Brief explanation of the drawing]

[0008] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Modes for carrying out the invention]

[0009] Hereinafter, an example of an embodiment of the system relating to the technology of this disclosure will be described with reference to the attached drawings.

[0010] First, let's explain the terminology used in the following explanation.

[0011] In the following embodiments, the signed processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Furthermore, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), APU (Accelerated Processing Unit), or TPU (Tensor Processing Unit).

[0012] In the following embodiments, signed RAM (Random Access Memory) is a memory that temporarily stores information and is used as work memory by the processor.

[0013] In the following embodiments, the signed storage is one or more non-volatile storage devices that store various programs and various parameters. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes.

[0014] In the following embodiments, the labeled communication I / F (Interface) is an interface including a communication processor, an antenna, and the like. The communication I / F manages communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0015] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B". That is, "A and / or B" means that it may be only A, only B, or a combination of A and B. Also, in this specification, when expressing three or more matters connected by "and / or", the same concept as "A and / or B" is applied.

[0016] [First Embodiment] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0017] As shown in FIG. 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0018] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. Also, the database 24 and the communication I / F 26 are connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0019] The smart device 14 comprises a computer 36, a receiving device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The receiving device 38, output device 40, and camera 42 are also connected to the bus 52.

[0020] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, and accepts user input. The touch panel 38A accepts user input via touch by detecting contact with an object (e.g., a pen or finger). The microphone 38B accepts user input via voice by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 (see Figure 2) acquires the data indicating the user input.

[0021] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user by outputting the data in a form perceptible to the user (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0022] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0023] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0024] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0025] Storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the user's emotions using the emotion identification model 59 and perform identification processing using the user's emotions. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotions, including but not limited to these examples. Furthermore, emotion estimation and prediction also include, for example, emotion analysis.

[0026] In the smart device 14, specific processing is performed by the processor 46. The storage 50 stores a specific processing program 60. The specific processing program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the specific processing program 60 from the storage 50 and executes the read specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as a control unit 46A according to the specific processing program 60 executed on the RAM 48. The smart device 14 also has a data generation model 58 and an emotion identification model 59, similar to the data generation model and emotion identification model 59, and can perform processing similar to that of the specific processing unit 290 using these models.

[0027] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device (e.g., a generation server) may have the data generation model 58. In this case, the data processing device 12 obtains processing results (such as prediction results) using the data generation model 58 by communicating with the server device having the data generation model 58. The data processing device 12 may also be a server device or a terminal device owned by a user (e.g., a mobile phone, robot, home appliance, etc.). Next, an example of processing by the data processing system 10 according to the first embodiment will be described.

[0028] (Example of form 1) The AI ​​agent monitoring system according to an embodiment of the present invention is a system that monitors and detects abnormal behavior of AI agents in real time and automatically blocks it. This AI agent monitoring system uses generative AI to learn the normal behavior patterns of agents and provides a mechanism for early detection of abnormalities. Furthermore, it analyzes communication between agents using natural language processing technology to detect malicious instructions and communications. When an abnormality is detected, it immediately blocks the action and notifies the relevant parties to minimize damage. For example, the AI ​​agent monitoring system collects agent behavior data, and the generative AI analyzes this data to learn normal behavior patterns. For example, it analyzes the tasks and communication content that agents normally perform and sets criteria for identifying abnormal behavior. Next, the AI ​​agent monitoring system monitors the agent's behavior in real time and immediately blocks the action when an abnormality is detected. For example, if an agent receives a malicious instruction, it invalidates the instruction and notifies the relevant parties. Furthermore, the AI ​​agent monitoring system analyzes communication between agents using natural language processing technology to detect malicious instructions and communications. For example, if an agent sends a malicious instruction to another agent, it detects the instruction and blocks the action. Furthermore, the AI ​​agent monitoring system provides tools to visualize and analyze agent activity logs, making it easy to understand the agent's behavior history. This significantly reduces the risk of leakage of personal and confidential information. By preventing fraudulent agent behavior, it improves user and corporate trust and reduces legal risks. Additionally, the automation of agent monitoring and management reduces effort and costs, enabling efficient operation. Ultimately, it promotes a safe coexistence between humans and AI, contributing to the creation of a sustainable future. In short, the AI ​​agent monitoring system significantly reduces the risk of leakage of personal and confidential information.

[0029] The AI ​​agent monitoring system according to this embodiment comprises a collection unit, a learning unit, a monitoring unit, a detection unit, a blocking unit, and a notification unit. The collection unit collects agent behavior data. The collection unit can collect, for example, user operation logs, sensor data, activity logs, etc. The collection unit can, for example, collect agent behavior data in real time and store it in a database. The collection unit can also periodically collect agent behavior data and provide it for analysis. Furthermore, the collection unit can filter the agent behavior data and collect only the necessary data. For example, the collection unit collects agent behavior data in real time and provides basic data for early detection of abnormal behavior. The learning unit uses generative AI to analyze the data collected by the collection unit and learns normal behavior patterns. The learning unit can, for example, learn normal behavior patterns using statistical criteria based on past data or rule-based criteria. The learning unit uses generative AI to analyze agent behavior data and learn normal behavior patterns. Furthermore, the learning unit can use generative AI to set criteria for identifying abnormal behavior based on the agent's behavioral data. In addition, the learning unit can use generative AI to analyze the agent's behavioral data and build a model for early detection of abnormal behavior. For example, the learning unit uses generative AI to analyze the agent's behavioral data and set criteria for identifying abnormal behavior. The monitoring unit monitors in real time based on the normal behavioral patterns learned by the learning unit. The monitoring unit can, for example, monitor the agent's behavioral data in real time and detect abnormal behavior early. The monitoring unit can, for example, monitor the agent's behavioral data in real time and set criteria for detecting abnormal behavior. Furthermore, the monitoring unit can monitor the agent's behavioral data in real time and build a model for detecting abnormal behavior. In addition, the monitoring unit can monitor the agent's behavioral data in real time and build a system for early detection of abnormal behavior. For example, the monitoring unit monitors the agent's behavioral data in real time and sets criteria for early detection of abnormal behavior.The detection unit detects actions monitored by the monitoring unit. For example, the detection unit can detect agent behavior data in real time and detect abnormal behavior early. The detection unit can, for example, detect agent behavior data in real time and set criteria for detecting abnormal behavior. The detection unit can also detect agent behavior data in real time and build models for detecting abnormal behavior. Furthermore, the detection unit can detect agent behavior data in real time and build systems for detecting abnormal behavior early. For example, the detection unit can detect agent behavior data in real time and set criteria for detecting abnormal behavior early. The blocking unit blocks abnormal behavior detected by the detection unit. For example, the blocking unit can block agent behavior data in real time and block abnormal behavior early. The blocking unit can, for example, block agent behavior data in real time and set criteria for blocking abnormal behavior. Furthermore, the blocking unit can block agent behavior data in real time and build models for blocking abnormal behavior. Furthermore, the blocking unit can block agent behavior data in real time and build systems for blocking abnormal behavior early. For example, the blocking unit blocks agent behavior data in real time and sets criteria for early detection of abnormal behavior. The notification unit notifies relevant parties of abnormal behavior blocked by the blocking unit. The notification unit can, for example, notify agent behavior data in real time and provide early notification of abnormal behavior. The notification unit can, for example, notify agent behavior data in real time and set criteria for notifying abnormal behavior. The notification unit can also build a model for notifying agent behavior data in real time and providing early notification of abnormal behavior. Furthermore, the notification unit can build a system for notifying agent behavior data in real time and providing early notification of abnormal behavior. For example, the notification unit can notify agent behavior data in real time and set criteria for early notification of abnormal behavior.As a result, the AI ​​agent monitoring system according to the embodiment can monitor and detect abnormal behavior of AI agents in real time and automatically prevent it.

[0030] The data collection unit collects agent behavior data. For example, the data collection unit can collect user operation logs, sensor data, and activity logs. Specifically, user operation logs include a history of instructions and operations performed by the user on the agent, enabling tracing of the agent's responses and actions. Sensor data includes environmental information such as temperature, humidity, and light intensity in the environment where the agent operates, allowing for understanding changes in the agent's operating environment. Activity logs record details of tasks and processes performed by the agent, enabling detailed tracking of the agent's behavior history. The data collection unit collects this data in real time and stores it in a database. Real-time data collection allows for immediate understanding of agent behavior and helps in the early detection of abnormal behavior. The data collection unit can also periodically collect agent behavior data and provide it for analysis. Periodic data collection allows for understanding long-term changes and trends in behavioral patterns. Furthermore, the data collection unit can filter agent behavior data, collecting only the necessary data. For example, extracting only behavioral data under specific conditions or data showing signs of abnormal behavior enables efficient data management. This allows the data collection unit to collect agent behavioral data from multiple perspectives, providing the fundamental data necessary for early detection and analysis of abnormal behavior.

[0031] The learning unit uses generative AI to analyze data collected by the collection unit and learns normal behavior patterns. The learning unit can learn normal behavior patterns using, for example, statistical criteria based on past data or rule-based criteria. Specifically, the generative AI analyzes a large amount of collected behavioral data and extracts the agent's normal behavior patterns. This involves a process of finding common patterns and features from the data using machine learning algorithms. For example, a series of operations and response times when the agent performs a specific task are learned as normal behavior patterns. The learning unit can also use generative AI to set criteria for identifying abnormal behavior based on the agent's behavioral data. Criteria for abnormal behavior are set based on behaviors that deviate significantly from normal behavior patterns or characteristics of behaviors previously judged as abnormal. Furthermore, the learning unit can use generative AI to analyze the agent's behavioral data and build models for early detection of abnormal behavior. This model is used to analyze behavioral data collected in real time and detect signs of abnormal behavior. For example, the learning unit uses generative AI to analyze the agent's behavioral data and set criteria for identifying abnormal behavior. This allows the learning unit to learn the normal behavior patterns of the agent and build criteria and models for early detection of abnormal behavior.

[0032] The monitoring unit monitors in real time based on normal behavior patterns learned by the learning unit. For example, the monitoring unit can monitor agent behavior data in real time and detect abnormal behavior early. Specifically, the monitoring unit uses an anomaly detection model built by the learning unit to analyze agent behavior data in real time and detect signs of abnormal behavior. For example, if an agent performs an unusual operation or shows an unexpected response, it is detected as abnormal behavior. The monitoring unit also monitors agent behavior data in real time and sets criteria for detecting abnormal behavior. This allows the monitoring unit to continuously monitor agent behavior and achieve early detection of abnormal behavior. Furthermore, the monitoring unit can monitor agent behavior data in real time and build a model for detecting abnormal behavior. This model is used to analyze agent behavior data and identify patterns of abnormal behavior. For example, the monitoring unit monitors agent behavior data in real time and sets criteria for early detection of abnormal behavior. This allows the monitoring unit to build a system for real-time monitoring of agent behavior and early detection of abnormal behavior.

[0033] The detection unit detects actions monitored by the monitoring unit. For example, the detection unit can detect agent behavior data in real time and detect abnormal behavior early. Specifically, the detection unit analyzes agent behavior data using criteria and models for abnormal behavior set by the monitoring unit and detects abnormal behavior. For example, if an agent deviates from normal operating procedures or performs an unexpected action, it is detected as abnormal behavior. Furthermore, the detection unit detects agent behavior data in real time and sets criteria for detecting abnormal behavior. This allows the detection unit to continuously monitor agent behavior and achieve early detection of abnormal behavior. In addition, the detection unit can detect agent behavior data in real time and build models for detecting abnormal behavior. This model is used to analyze agent behavior data and identify patterns of abnormal behavior. For example, the detection unit detects agent behavior data in real time and sets criteria for early detection of abnormal behavior. This allows the detection unit to build a system for real-time detection of agent behavior and early detection of abnormal behavior.

[0034] The blocking unit blocks abnormal behavior detected by the detection unit. For example, the blocking unit can block agent behavior data in real time, enabling early detection of abnormal behavior. Specifically, the blocking unit responds immediately to abnormal behavior detected by the detection unit, either stopping the agent's operation or issuing instructions to correct the abnormal behavior. For example, if an agent attempts an unauthorized operation, the blocking unit disables that operation. Furthermore, the blocking unit blocks agent behavior data in real time and sets criteria for blocking abnormal behavior. This allows the blocking unit to continuously monitor agent behavior and achieve early detection of abnormal behavior. Additionally, the blocking unit can block agent behavior data in real time and build models for blocking abnormal behavior. These models are used to analyze agent behavior data and identify patterns of abnormal behavior. For example, the blocking unit blocks agent behavior data in real time and sets criteria for early detection of abnormal behavior. This allows the blocking unit to build a system that blocks agent behavior in real time and enables early detection of abnormal behavior.

[0035] The notification unit notifies relevant parties of abnormal behavior blocked by the blocking unit. For example, the notification unit can notify agents of their behavior data in real time, enabling early notification of abnormal behavior. Specifically, the notification unit immediately transmits information about abnormal behavior detected by the blocking unit to relevant parties, prompting appropriate action. For example, if an agent attempts to perform an unauthorized operation, the notification unit notifies the administrator of the details and requests action. The notification unit also notifies agents of their behavior data in real time and sets criteria for notifying abnormal behavior. This allows the notification unit to continuously monitor agent behavior and achieve early notification of abnormal behavior. Furthermore, the notification unit can notify agents of their behavior data in real time and build models for notifying abnormal behavior. These models are used to analyze agent behavior data and identify patterns of abnormal behavior. For example, the notification unit notifies agents of their behavior data in real time and sets criteria for early notification of abnormal behavior. This allows the notification unit to build a system for notifying agents of their behavior in real time and for early notification of abnormal behavior. As a result, the AI ​​agent monitoring system according to the embodiment can monitor and detect abnormal behavior of AI agents in real time and automatically prevent it.

[0036] The analysis unit can analyze communication between agents. For example, the analysis unit can analyze the type of message, communication protocol, and data format between agents. For example, the analysis unit can analyze the content of messages between agents to detect malicious instructions or communications. The analysis unit can also analyze the communication protocol between agents to detect malicious communications. Furthermore, the analysis unit can analyze the data format between agents to detect malicious data. For example, the analysis unit can analyze the type of message between agents to detect malicious instructions. In this way, by analyzing communication between agents, the analysis unit can detect malicious instructions and communications and detect abnormal behavior at an early stage.

[0037] The visualization unit can visualize and analyze agent operation logs. For example, the visualization unit visualizes agent operation history, system events, and error logs. For instance, it can display agent operation history as graphs or charts, making it easy to understand the agent's behavior history. The visualization unit can also display system events in a timeline format, allowing for a visual understanding of the agent's operation flow. Furthermore, the visualization unit can display error logs in a list format, allowing for an understanding of the agent's error occurrences. For example, the visualization unit displays agent operation history as a graph, providing a visual understanding of the agent's behavior history. In this way, the visualization unit can easily understand the agent's behavior history by visualizing and analyzing agent operation logs.

[0038] The data collection unit can analyze the agent's past behavioral data and select the optimal data collection method. For example, the data collection unit optimizes the data collection method based on actions the agent has frequently performed in the past. For example, the data collection unit can customize the data collection method based on the agent's past behavioral data. The data collection unit can also analyze the agent's behavioral patterns and select an efficient data collection method. For example, the data collection unit optimizes the data collection method based on the agent's past behavioral data. This allows the data collection unit to select an efficient data collection method by analyzing the agent's past behavioral data. Some or all of the above processing in the data collection unit may be performed using AI, for example, or without AI. For example, the data collection unit can input the agent's past behavioral data into a generating AI and have the generating AI select the optimal data collection method.

[0039] The data collection unit can filter behavioral data based on the agent's current task and environment. For example, the data collection unit can collect only relevant data based on the task the agent is currently performing. The data collection unit can filter the necessary data by considering the agent's environmental information. The data collection unit can also select the data to collect based on the priority of the agent's tasks. For example, the data collection unit can collect only relevant data based on the task the agent is currently performing. This allows the data collection unit to collect only relevant data by filtering the data based on the agent's current task and environment. Some or all of the above processing in the data collection unit may be performed using AI, for example, or without AI. For example, the data collection unit can input the agent's current task and environmental information into a generating AI and have the generating AI perform the filtering.

[0040] The data collection unit can prioritize the collection of highly relevant data by considering the agent's geographical location information when collecting behavioral data. For example, if the agent is in a specific region, the data collection unit will prioritize the collection of data related to that region. The data collection unit can also select highly relevant data based on the agent's location information. Furthermore, the data collection unit can optimize the collected data by considering the agent's movement patterns. For example, if the agent is in a specific region, the data collection unit will prioritize the collection of data related to that region. In this way, the data collection unit can prioritize the collection of highly relevant data by considering the agent's geographical location information. Some or all of the above processing in the data collection unit may be performed using AI, for example, or without AI. For example, the data collection unit can input the agent's geographical location information into a generating AI and have the generating AI perform the selection of highly relevant data.

[0041] The data collection unit can analyze the agent's social media activity and collect relevant data when collecting behavioral data. For example, the data collection unit can analyze the agent's social media posts and collect relevant data. For example, the data collection unit can extract important data from the agent's social media activity. The data collection unit can also select the data to be collected by considering the agent's level of social media involvement. For example, the data collection unit can analyze the agent's social media posts and collect relevant data. In this way, the data collection unit can collect relevant data by analyzing the agent's social media activity. Some or all of the above processing in the data collection unit may be performed using AI, for example, or without AI. For example, the data collection unit can input the agent's social media activity data into a generating AI and have the generating AI perform the extraction of relevant data.

[0042] The learning unit can optimize the learning algorithm by referring to the agent's past behavior data during learning. For example, the learning unit can adjust the learning algorithm based on the agent's past behavior data. The learning unit can, for example, analyze the agent's behavior patterns and select the optimal learning algorithm. The learning unit can also optimize the learning algorithm by referring to the agent's past data. For example, the learning unit can adjust the learning algorithm based on the agent's past behavior data. In this way, the learning unit can optimize the learning algorithm by referring to the agent's past behavior data and improve the accuracy of learning. Some or all of the above processing in the learning unit may be performed using AI, for example, or without using AI. For example, the learning unit can input the agent's past behavior data into a generating AI and have the generating AI perform the optimization of the learning algorithm.

[0043] The learning unit can apply different learning methods to each agent task during training. For example, the learning unit can select an appropriate learning method according to the agent task. For example, the learning unit can apply different learning algorithms to each agent task. The learning unit can also customize the learning method based on the characteristics of the agent task. For example, the learning unit can select an appropriate learning method according to the agent task. This allows the learning unit to improve the accuracy of learning by applying different learning methods to each agent task. Some or all of the above processing in the learning unit may be performed using AI, for example, or without AI. For example, the learning unit can input agent task information into a generating AI and have the generating AI select an appropriate learning method.

[0044] The learning unit can weight the training data during training based on when the agent's behavior data is submitted. For example, the learning unit can weight important data based on when the agent's behavior data is submitted. The learning unit can optimize the training data by considering when the agent's behavior data is submitted. The learning unit can also adjust the weighting of the training data according to when the agent's behavior data is submitted. For example, the learning unit can weight important data based on when the agent's behavior data is submitted. This allows the learning unit to improve the accuracy of training by weighting the training data based on when the agent's behavior data is submitted. Some or all of the above processing in the learning unit may be performed using AI, for example, or without AI. For example, the learning unit can input information about when the agent's behavior data is submitted into a generating AI and have the generating AI perform the weighting of the training data.

[0045] The learning unit can improve the accuracy of learning by referring to the agent's relevant literature during training. For example, the learning unit can optimize the learning algorithm by referring to the agent's relevant literature. The learning unit can supplement the training data based on the agent's relevant literature. The learning unit can also improve the accuracy of learning by referring to the agent's relevant literature. For example, the learning unit can optimize the learning algorithm by referring to the agent's relevant literature. In this way, the learning unit can improve the accuracy of learning by referring to the agent's relevant literature. Some or all of the above processing in the learning unit may be performed using AI, for example, or without using AI. For example, the learning unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the improvement of learning accuracy.

[0046] The monitoring unit can improve the accuracy of monitoring by considering the interrelationships between agents during monitoring. For example, the monitoring unit can analyze communication between agents to improve monitoring accuracy. For example, the monitoring unit can optimize monitoring data by considering the interrelationships between agents. The monitoring unit can also adjust the accuracy of monitoring based on the interrelationships between agents. For example, the monitoring unit can analyze communication between agents to improve monitoring accuracy. In this way, the monitoring unit can improve the accuracy of monitoring by considering the interrelationships between agents. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without using AI. For example, the monitoring unit can input agent interrelationship data into a generating AI and have the generating AI perform the improvement of monitoring accuracy.

[0047] The monitoring unit can perform monitoring while considering the attribute information of the agent's task. For example, the monitoring unit can select monitoring data based on the attribute information of the agent's task. For example, the monitoring unit can improve the accuracy of monitoring by considering the attribute information of the agent's task. The monitoring unit can also adjust the monitoring criteria based on the attribute information of the agent's task. For example, the monitoring unit can select monitoring data based on the attribute information of the agent's task. In this way, the monitoring unit can improve the accuracy of monitoring by considering the attribute information of the agent's task. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without using AI. For example, the monitoring unit can input the agent's task attribute information into a generating AI and have the generating AI perform the task to improve the accuracy of monitoring.

[0048] The monitoring unit can perform monitoring while considering the geographical distribution of agents. For example, the monitoring unit can select monitoring data based on the geographical distribution of agents. For example, the monitoring unit can improve the accuracy of monitoring by considering the geographical distribution of agents. The monitoring unit can also adjust the monitoring criteria based on the geographical distribution of agents. For example, the monitoring unit can select monitoring data based on the geographical distribution of agents. In this way, the monitoring unit can improve the accuracy of monitoring by considering the geographical distribution of agents. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without using AI. For example, the monitoring unit can input the geographical distribution data of agents into a generating AI and have the generating AI perform the task of improving the accuracy of monitoring.

[0049] The monitoring unit can improve the accuracy of monitoring by referring to the agent's relevant literature during monitoring. For example, the monitoring unit can optimize the monitoring algorithm by referring to the agent's relevant literature. The monitoring unit can supplement monitoring data based on the agent's relevant literature. The monitoring unit can also improve the accuracy of monitoring by referring to the agent's relevant literature. For example, the monitoring unit can optimize the monitoring algorithm by referring to the agent's relevant literature. In this way, the monitoring unit can improve the accuracy of monitoring by referring to the agent's relevant literature. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without using AI. For example, the monitoring unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the improvement of monitoring accuracy.

[0050] The detection unit can optimize its detection algorithm by referring to the agent's past abnormal behavior data when detection occurs. For example, the detection unit adjusts the detection algorithm based on the agent's past abnormal behavior data. For example, the detection unit can analyze the agent's abnormal behavior patterns and select the optimal detection algorithm. The detection unit can also optimize its detection algorithm by referring to the agent's past data. For example, the detection unit adjusts the detection algorithm based on the agent's past abnormal behavior data. In this way, the detection unit can optimize its detection algorithm by referring to the agent's past abnormal behavior data and improve the accuracy of detection. Some or all of the above processing in the detection unit may be performed using AI, for example, or without using AI. For example, the detection unit can input the agent's past abnormal behavior data into a generating AI and have the generating AI perform the optimization of the detection algorithm.

[0051] The detection unit can apply different detection methods to each agent's task when detection occurs. For example, the detection unit can select an appropriate detection method according to the agent's task. For example, the detection unit can apply different detection algorithms to each agent's task. The detection unit can also customize detection methods based on the characteristics of the agent's task. For example, the detection unit can select an appropriate detection method according to the agent's task. This allows the detection unit to improve detection accuracy by applying different detection methods to each agent's task. Some or all of the above processing in the detection unit may be performed using AI, for example, or without AI. For example, the detection unit can input agent task information into a generating AI and have the generating AI select an appropriate detection method.

[0052] The detection unit can weight the detected data based on the timing of the agent's behavior data submission when detection occurs. For example, the detection unit weights important data based on the timing of the agent's behavior data submission. The detection unit can optimize the detected data by considering the timing of the agent's behavior data submission. The detection unit can also adjust the weighting of the detected data according to the timing of the agent's behavior data submission. For example, the detection unit weights important data based on the timing of the agent's behavior data submission. This allows the detection unit to improve detection accuracy by weighting the detected data based on the timing of the agent's behavior data submission. Some or all of the above processing in the detection unit may be performed using AI, for example, or without AI. For example, the detection unit can input information on the timing of the agent's behavior data submission into a generating AI and have the generating AI perform the weighting of the detected data.

[0053] The detection unit can improve detection accuracy by referring to the agent's relevant literature during detection. For example, the detection unit optimizes the detection algorithm by referring to the agent's relevant literature. The detection unit can supplement detection data based on the agent's relevant literature. The detection unit can also improve detection accuracy by referring to the agent's relevant literature. For example, the detection unit optimizes the detection algorithm by referring to the agent's relevant literature. In this way, the detection unit can improve detection accuracy by referring to the agent's relevant literature. Some or all of the above processing in the detection unit may be performed using AI, for example, or without AI. For example, the detection unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the detection accuracy improvement.

[0054] The blocking unit can optimize its blocking algorithm by referring to the agent's past abnormal behavior data when blocking. For example, the blocking unit adjusts the blocking algorithm based on the agent's past abnormal behavior data. For example, the blocking unit can analyze the agent's abnormal behavior patterns and select the optimal blocking algorithm. The blocking unit can also optimize its blocking algorithm by referring to the agent's past data. For example, the blocking unit adjusts the blocking algorithm based on the agent's past abnormal behavior data. In this way, the blocking unit can optimize its blocking algorithm by referring to the agent's past abnormal behavior data and improve the accuracy of blocking. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without using AI. For example, the blocking unit can input the agent's past abnormal behavior data into a generating AI and have the generating AI perform the optimization of the blocking algorithm.

[0055] The blocking unit can apply different blocking methods to each agent task when blocking. For example, the blocking unit can select an appropriate blocking method according to the agent task. For example, the blocking unit can apply different blocking algorithms to each agent task. The blocking unit can also customize blocking methods based on the characteristics of the agent task. For example, the blocking unit can select an appropriate blocking method according to the agent task. This allows the blocking unit to improve the accuracy of blocking by applying different blocking methods to each agent task. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without AI. For example, the blocking unit can input agent task information into a generating AI and have the generating AI select an appropriate blocking method.

[0056] The blocking unit can weight the blocking data based on the timing of the agent's behavior data submission when blocking. For example, the blocking unit can weight important data based on the timing of the agent's behavior data submission. The blocking unit can optimize the blocking data by considering the timing of the agent's behavior data submission. The blocking unit can also adjust the weighting of the blocking data according to the timing of the agent's behavior data submission. For example, the blocking unit can weight important data based on the timing of the agent's behavior data submission. This allows the blocking unit to improve the accuracy of blocking by weighting the blocking data based on the timing of the agent's behavior data submission. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without AI. For example, the blocking unit can input information on the timing of the agent's behavior data submission into a generating AI and have the generating AI perform the weighting of the blocking data.

[0057] The blocking unit can improve the accuracy of blocking by referring to the agent's relevant literature during blocking. For example, the blocking unit can optimize the blocking algorithm by referring to the agent's relevant literature. The blocking unit can supplement the blocking data based on the agent's relevant literature. The blocking unit can also improve the accuracy of blocking by referring to the agent's relevant literature. For example, the blocking unit can optimize the blocking algorithm by referring to the agent's relevant literature. In this way, the blocking unit can improve the accuracy of blocking by referring to the agent's relevant literature. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without AI. For example, the blocking unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the improvement of blocking accuracy.

[0058] The notification unit can optimize the notification algorithm by referring to the agent's past abnormal behavior data when issuing a notification. For example, the notification unit can adjust the notification algorithm based on the agent's past abnormal behavior data. For example, the notification unit can analyze the agent's abnormal behavior patterns and select the optimal notification algorithm. The notification unit can also optimize the notification algorithm by referring to the agent's past data. For example, the notification unit can adjust the notification algorithm based on the agent's past abnormal behavior data. In this way, the notification unit can optimize the notification algorithm by referring to the agent's past abnormal behavior data and improve the accuracy of notifications. Some or all of the above processing in the notification unit may be performed using AI, for example, or without using AI. For example, the notification unit can input the agent's past abnormal behavior data into a generating AI and have the generating AI perform the optimization of the notification algorithm.

[0059] The notification unit can apply different notification methods to each agent's task when issuing notifications. For example, the notification unit can select an appropriate notification method according to the agent's task. For example, the notification unit can apply different notification algorithms to each agent's task. The notification unit can also customize notification methods based on the characteristics of the agent's task. For example, the notification unit can select an appropriate notification method according to the agent's task. This allows the notification unit to improve notification accuracy by applying different notification methods to each agent's task. Some or all of the above processing in the notification unit may be performed using AI, for example, or without AI. For example, the notification unit can input agent task information into a generating AI and have the generating AI select an appropriate notification method.

[0060] The notification unit can weight notification data based on the timing of agent behavior data submission at the time of notification. For example, the notification unit can weight important data based on the timing of agent behavior data submission. The notification unit can optimize notification data by considering the timing of agent behavior data submission. The notification unit can also adjust the weighting of notification data according to the timing of agent behavior data submission. For example, the notification unit can weight important data based on the timing of agent behavior data submission. This allows the notification unit to improve the accuracy of notifications by weighting notification data based on the timing of agent behavior data submission. Some or all of the above processing in the notification unit may be performed using AI, for example, or without AI. For example, the notification unit can input agent behavior data submission timing information into a generating AI and have the generating AI perform the weighting of notification data.

[0061] The notification unit can improve the accuracy of notifications by referring to the agent's relevant literature at the time of notification. For example, the notification unit can optimize the notification algorithm by referring to the agent's relevant literature. The notification unit can supplement notification data based on the agent's relevant literature. The notification unit can also improve the accuracy of notifications by referring to the agent's relevant literature. For example, the notification unit can optimize the notification algorithm by referring to the agent's relevant literature. In this way, the notification unit can improve the accuracy of notifications by referring to the agent's relevant literature. Some or all of the above processing in the notification unit may be performed using AI, for example, or without using AI. For example, the notification unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the improvement of notification accuracy.

[0062] The analysis unit can optimize the analysis algorithm by referring to the agent's past communication data during analysis. For example, the analysis unit adjusts the analysis algorithm based on the agent's past communication data. For example, the analysis unit can analyze the agent's communication patterns and select the optimal analysis algorithm. The analysis unit can also optimize the analysis algorithm by referring to the agent's past data. For example, the analysis unit adjusts the analysis algorithm based on the agent's past communication data. In this way, the analysis unit can optimize the analysis algorithm by referring to the agent's past communication data and improve the accuracy of the analysis. Some or all of the above processing in the analysis unit may be performed using AI, for example, or without using AI. For example, the analysis unit can input the agent's past communication data into a generating AI and have the generating AI perform the optimization of the analysis algorithm.

[0063] The analysis unit can apply different analysis methods to each agent task during analysis. For example, the analysis unit can select an appropriate analysis method according to the agent task. For example, the analysis unit can apply different analysis algorithms to each agent task. Furthermore, the analysis unit can customize the analysis method based on the characteristics of the agent task. For example, the analysis unit can select an appropriate analysis method according to the agent task. This allows the analysis unit to improve the accuracy of the analysis by applying different analysis methods to each agent task. Some or all of the above-described processes in the analysis unit may be performed using AI, for example, or without AI. For example, the analysis unit can input agent task information into a generating AI and have the generating AI select an appropriate analysis method.

[0064] The analysis unit can weight the analysis data based on when the agent's communication data was submitted during the analysis. For example, the analysis unit can weight important data based on when the agent's communication data was submitted. The analysis unit can optimize the analysis data by considering when the agent's communication data was submitted. The analysis unit can also adjust the weighting of the analysis data according to when the agent's communication data was submitted. For example, the analysis unit can weight important data based on when the agent's communication data was submitted. This allows the analysis unit to improve the accuracy of the analysis by weighting the analysis data based on when the agent's communication data was submitted. Some or all of the above processing in the analysis unit may be performed using AI, for example, or without AI. For example, the analysis unit can input information on when the agent's communication data was submitted into a generating AI and have the generating AI perform the weighting of the analysis data.

[0065] The analysis unit can improve the accuracy of its analysis by referring to the agent's relevant literature during the analysis. For example, the analysis unit can optimize its analysis algorithm by referring to the agent's relevant literature. The analysis unit can supplement its analysis data based on the agent's relevant literature. Furthermore, the analysis unit can improve the accuracy of its analysis by referring to the agent's relevant literature. For example, the analysis unit can optimize its analysis algorithm by referring to the agent's relevant literature. In this way, the analysis unit can improve the accuracy of its analysis by referring to the agent's relevant literature. Some or all of the above processes in the analysis unit may be performed using AI, for example, or without AI. For example, the analysis unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the analysis accuracy improvement.

[0066] The visualization unit can optimize the visualization algorithm by referring to the agent's past operation logs during visualization. For example, the visualization unit adjusts the visualization algorithm based on the agent's past operation logs. For example, the visualization unit can analyze the agent's operation patterns and select the optimal visualization algorithm. The visualization unit can also optimize the visualization algorithm by referring to the agent's past data. For example, the visualization unit adjusts the visualization algorithm based on the agent's past operation logs. In this way, the visualization unit can optimize the visualization algorithm by referring to the agent's past operation logs and improve the accuracy of the visualization. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without AI. For example, the visualization unit can input the agent's past operation logs into a generating AI and have the generating AI perform the optimization of the visualization algorithm.

[0067] The visualization unit can apply different visualization methods to each agent task during visualization. For example, the visualization unit can select an appropriate visualization method according to the agent task. For example, the visualization unit can apply different visualization algorithms to each agent task. The visualization unit can also customize the visualization method based on the characteristics of the agent task. For example, the visualization unit can select an appropriate visualization method according to the agent task. This allows the visualization unit to improve the accuracy of visualization by applying different visualization methods to each agent task. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without AI. For example, the visualization unit can input agent task information into a generating AI and have the generating AI select an appropriate visualization method.

[0068] The visualization unit can weight the visualization data based on the timing of agent operation log submissions during visualization. For example, the visualization unit can weight important data based on the timing of agent operation log submissions. The visualization unit can optimize the visualization data by considering the timing of agent operation log submissions. The visualization unit can also adjust the weighting of the visualization data according to the timing of agent operation log submissions. For example, the visualization unit can weight important data based on the timing of agent operation log submissions. This allows the visualization unit to improve the accuracy of visualization by weighting the visualization data based on the timing of agent operation log submissions. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without AI. For example, the visualization unit can input agent operation log submission timing information into a generating AI and have the generating AI perform the weighting of the visualization data.

[0069] The visualization unit can improve the accuracy of visualization by referring to the agent's relevant literature during visualization. For example, the visualization unit can optimize the visualization algorithm by referring to the agent's relevant literature. The visualization unit can supplement the visualization data based on the agent's relevant literature. The visualization unit can also improve the accuracy of visualization by referring to the agent's relevant literature. For example, the visualization unit can optimize the visualization algorithm by referring to the agent's relevant literature. In this way, the visualization unit can improve the accuracy of visualization by referring to the agent's relevant literature. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without using AI. For example, the visualization unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the visualization accuracy improvement.

[0070] The system according to the embodiment is not limited to the example described above, and various modifications are possible, for example, as follows.

[0071] The data collection unit can optimize its collection method by considering the frequency and patterns of agent behavior when collecting agent behavior data. For example, if an agent frequently performs a particular action, it can prioritize the collection of data related to that action. It can also analyze agent behavior patterns and focus on collecting data useful for detecting abnormal behavior. Furthermore, it can collect agent behavior data in real time to provide data useful for the early detection of abnormal behavior. In this way, the data collection unit can efficiently collect agent behavior data and provide data useful for the early detection of abnormal behavior.

[0072] The analysis unit can optimize its analysis method by considering the frequency and patterns of agent communication when analyzing communication between agents. For example, it can analyze the frequency and patterns of messages between agents to detect abnormal communication. It can also analyze the content of communication between agents to detect malicious instructions or communications. Furthermore, it can analyze the communication protocols and data formats between agents to detect malicious communications. As a result, the analysis unit can efficiently analyze communication between agents and detect abnormal communication at an early stage.

[0073] The visualization unit can optimize the visualization method when visualizing agent operation logs, taking into account the frequency and patterns of agent operations. For example, it can display the agent's operation history as graphs or charts to visually understand the agent's behavior patterns. It can also display the flow of agent operations in a timeline format to visually understand the overall picture of agent operations. Furthermore, it can display agent error logs in a list format to understand the agent's error occurrence status. In this way, the visualization unit can efficiently visualize agent operation logs and easily understand agent behavior patterns.

[0074] The data collection unit can analyze the agent's past behavioral data and select the optimal collection method. For example, it can optimize the collection method based on actions the agent has frequently performed in the past. It can also analyze the agent's behavioral patterns and select an efficient collection method. Furthermore, it can customize the collection method based on the agent's behavioral data. As a result, the data collection unit can efficiently collect agent behavioral data by analyzing the agent's past behavioral data and selecting an efficient collection method.

[0075] The data collection unit can filter behavioral data based on the agent's current task and environment. For example, it can collect only relevant data based on the task the agent is currently performing. It can also filter necessary data considering the agent's environmental information. Furthermore, it can select collected data based on the priority of the agent's tasks. In this way, the data collection unit can collect only relevant data by filtering it based on the agent's current task and environment.

[0076] The data collection unit can prioritize the collection of highly relevant data by considering the agent's geographical location when collecting behavioral data. For example, if an agent is in a specific region, it will prioritize the collection of data related to that region. It can also select highly relevant data based on the agent's location information. Furthermore, it can optimize the collected data by considering the agent's movement patterns. As a result, the data collection unit can prioritize the collection of highly relevant data by considering the agent's geographical location information.

[0077] The data collection unit can analyze the agent's social media activity and collect relevant data when collecting behavioral data. For example, it can analyze the agent's social media posts and collect relevant data. It can also extract important data from the agent's social media activity. Furthermore, it can select the data to be collected considering the agent's level of social media involvement. In this way, the data collection unit can collect relevant data by analyzing the agent's social media activity.

[0078] The following briefly describes the processing flow for example form 1.

[0079] Step 1: The collection unit collects agent behavior data. For example, it collects user operation logs, sensor data, activity logs, etc., and stores them in a database in real time. It can also collect data periodically and provide it for analysis. Furthermore, it filters and collects only the necessary data to provide foundational data for early detection of abnormal behavior. Step 2: The learning unit uses generative AI to analyze the data collected by the collection unit and learns normal behavior patterns. For example, it learns normal behavior patterns using statistical criteria based on past data or rule-based criteria, and sets criteria for identifying abnormal behavior. It also builds a model for early detection of abnormal behavior. Step 3: The monitoring unit monitors in real time based on the normal behavior patterns learned by the learning unit. For example, it monitors agent behavior data in real time, sets criteria for early detection of abnormal behavior, and builds a model for detecting abnormal behavior. Step 4: The detection unit detects the actions monitored by the monitoring unit. For example, it detects agent behavior data in real time, sets criteria for early detection of abnormal behavior, and builds a model for detecting abnormal behavior. Step 5: The blocking unit blocks the abnormal behavior detected by the detection unit. For example, it blocks agent behavior data in real time, sets criteria for early blocking of abnormal behavior, and builds a model for blocking abnormal behavior. Step 6: The notification unit notifies relevant parties of abnormal behavior blocked by the blocking unit. For example, it notifies agent behavior data in real time, sets criteria for early notification of abnormal behavior, and builds a model for notifying of abnormal behavior.

[0080] (Example of form 2) The AI ​​agent monitoring system according to an embodiment of the present invention is a system that monitors and detects abnormal behavior of AI agents in real time and automatically blocks it. This AI agent monitoring system uses generative AI to learn the normal behavior patterns of agents and provides a mechanism for early detection of abnormalities. Furthermore, it analyzes communication between agents using natural language processing technology to detect malicious instructions and communications. When an abnormality is detected, it immediately blocks the action and notifies the relevant parties to minimize damage. For example, the AI ​​agent monitoring system collects agent behavior data, and the generative AI analyzes this data to learn normal behavior patterns. For example, it analyzes the tasks and communication content that agents normally perform and sets criteria for identifying abnormal behavior. Next, the AI ​​agent monitoring system monitors the agent's behavior in real time and immediately blocks the action when an abnormality is detected. For example, if an agent receives a malicious instruction, it invalidates the instruction and notifies the relevant parties. Furthermore, the AI ​​agent monitoring system analyzes communication between agents using natural language processing technology to detect malicious instructions and communications. For example, if an agent sends a malicious instruction to another agent, it detects the instruction and blocks the action. Furthermore, the AI ​​agent monitoring system provides tools to visualize and analyze agent activity logs, making it easy to understand the agent's behavior history. This significantly reduces the risk of leakage of personal and confidential information. By preventing fraudulent agent behavior, it improves user and corporate trust and reduces legal risks. Additionally, the automation of agent monitoring and management reduces effort and costs, enabling efficient operation. Ultimately, it promotes a safe coexistence between humans and AI, contributing to the creation of a sustainable future. In short, the AI ​​agent monitoring system significantly reduces the risk of leakage of personal and confidential information.

[0081] The AI ​​agent monitoring system according to this embodiment comprises a collection unit, a learning unit, a monitoring unit, a detection unit, a blocking unit, and a notification unit. The collection unit collects agent behavior data. The collection unit can collect, for example, user operation logs, sensor data, activity logs, etc. The collection unit can, for example, collect agent behavior data in real time and store it in a database. The collection unit can also periodically collect agent behavior data and provide it for analysis. Furthermore, the collection unit can filter the agent behavior data and collect only the necessary data. For example, the collection unit collects agent behavior data in real time and provides basic data for early detection of abnormal behavior. The learning unit uses generative AI to analyze the data collected by the collection unit and learns normal behavior patterns. The learning unit can, for example, learn normal behavior patterns using statistical criteria based on past data or rule-based criteria. The learning unit uses generative AI to analyze agent behavior data and learn normal behavior patterns. Furthermore, the learning unit can use generative AI to set criteria for identifying abnormal behavior based on the agent's behavioral data. In addition, the learning unit can use generative AI to analyze the agent's behavioral data and build a model for early detection of abnormal behavior. For example, the learning unit uses generative AI to analyze the agent's behavioral data and set criteria for identifying abnormal behavior. The monitoring unit monitors in real time based on the normal behavioral patterns learned by the learning unit. The monitoring unit can, for example, monitor the agent's behavioral data in real time and detect abnormal behavior early. The monitoring unit can, for example, monitor the agent's behavioral data in real time and set criteria for detecting abnormal behavior. Furthermore, the monitoring unit can monitor the agent's behavioral data in real time and build a model for detecting abnormal behavior. In addition, the monitoring unit can monitor the agent's behavioral data in real time and build a system for early detection of abnormal behavior. For example, the monitoring unit monitors the agent's behavioral data in real time and sets criteria for early detection of abnormal behavior.The detection unit detects actions monitored by the monitoring unit. For example, the detection unit can detect agent behavior data in real time and detect abnormal behavior early. The detection unit can, for example, detect agent behavior data in real time and set criteria for detecting abnormal behavior. The detection unit can also detect agent behavior data in real time and build models for detecting abnormal behavior. Furthermore, the detection unit can detect agent behavior data in real time and build systems for detecting abnormal behavior early. For example, the detection unit can detect agent behavior data in real time and set criteria for detecting abnormal behavior early. The blocking unit blocks abnormal behavior detected by the detection unit. For example, the blocking unit can block agent behavior data in real time and block abnormal behavior early. The blocking unit can, for example, block agent behavior data in real time and set criteria for blocking abnormal behavior. Furthermore, the blocking unit can block agent behavior data in real time and build models for blocking abnormal behavior. Furthermore, the blocking unit can block agent behavior data in real time and build systems for blocking abnormal behavior early. For example, the blocking unit blocks agent behavior data in real time and sets criteria for early detection of abnormal behavior. The notification unit notifies relevant parties of abnormal behavior blocked by the blocking unit. The notification unit can, for example, notify agent behavior data in real time and provide early notification of abnormal behavior. The notification unit can, for example, notify agent behavior data in real time and set criteria for notifying abnormal behavior. The notification unit can also build a model for notifying agent behavior data in real time and providing early notification of abnormal behavior. Furthermore, the notification unit can build a system for notifying agent behavior data in real time and providing early notification of abnormal behavior. For example, the notification unit can notify agent behavior data in real time and set criteria for early notification of abnormal behavior.As a result, the AI ​​agent monitoring system according to the embodiment can monitor and detect abnormal behavior of AI agents in real time and automatically prevent it.

[0082] The data collection unit collects agent behavior data. For example, the data collection unit can collect user operation logs, sensor data, and activity logs. Specifically, user operation logs include a history of instructions and operations performed by the user on the agent, enabling tracing of the agent's responses and actions. Sensor data includes environmental information such as temperature, humidity, and light intensity in the environment where the agent operates, allowing for understanding changes in the agent's operating environment. Activity logs record details of tasks and processes performed by the agent, enabling detailed tracking of the agent's behavior history. The data collection unit collects this data in real time and stores it in a database. Real-time data collection allows for immediate understanding of agent behavior and helps in the early detection of abnormal behavior. The data collection unit can also periodically collect agent behavior data and provide it for analysis. Periodic data collection allows for understanding long-term changes and trends in behavioral patterns. Furthermore, the data collection unit can filter agent behavior data, collecting only the necessary data. For example, extracting only behavioral data under specific conditions or data showing signs of abnormal behavior enables efficient data management. This allows the data collection unit to collect agent behavioral data from multiple perspectives, providing the fundamental data necessary for early detection and analysis of abnormal behavior.

[0083] The learning unit uses generative AI to analyze data collected by the collection unit and learns normal behavior patterns. The learning unit can learn normal behavior patterns using, for example, statistical criteria based on past data or rule-based criteria. Specifically, the generative AI analyzes a large amount of collected behavioral data and extracts the agent's normal behavior patterns. This involves a process of finding common patterns and features from the data using machine learning algorithms. For example, a series of operations and response times when the agent performs a specific task are learned as normal behavior patterns. The learning unit can also use generative AI to set criteria for identifying abnormal behavior based on the agent's behavioral data. Criteria for abnormal behavior are set based on behaviors that deviate significantly from normal behavior patterns or characteristics of behaviors previously judged as abnormal. Furthermore, the learning unit can use generative AI to analyze the agent's behavioral data and build models for early detection of abnormal behavior. This model is used to analyze behavioral data collected in real time and detect signs of abnormal behavior. For example, the learning unit uses generative AI to analyze the agent's behavioral data and set criteria for identifying abnormal behavior. This allows the learning unit to learn the normal behavior patterns of the agent and build criteria and models for early detection of abnormal behavior.

[0084] The monitoring unit monitors in real time based on normal behavior patterns learned by the learning unit. For example, the monitoring unit can monitor agent behavior data in real time and detect abnormal behavior early. Specifically, the monitoring unit uses an anomaly detection model built by the learning unit to analyze agent behavior data in real time and detect signs of abnormal behavior. For example, if an agent performs an unusual operation or shows an unexpected response, it is detected as abnormal behavior. The monitoring unit also monitors agent behavior data in real time and sets criteria for detecting abnormal behavior. This allows the monitoring unit to continuously monitor agent behavior and achieve early detection of abnormal behavior. Furthermore, the monitoring unit can monitor agent behavior data in real time and build a model for detecting abnormal behavior. This model is used to analyze agent behavior data and identify patterns of abnormal behavior. For example, the monitoring unit monitors agent behavior data in real time and sets criteria for early detection of abnormal behavior. This allows the monitoring unit to build a system for real-time monitoring of agent behavior and early detection of abnormal behavior.

[0085] The detection unit detects actions monitored by the monitoring unit. For example, the detection unit can detect agent behavior data in real time and detect abnormal behavior early. Specifically, the detection unit analyzes agent behavior data using criteria and models for abnormal behavior set by the monitoring unit and detects abnormal behavior. For example, if an agent deviates from normal operating procedures or performs an unexpected action, it is detected as abnormal behavior. Furthermore, the detection unit detects agent behavior data in real time and sets criteria for detecting abnormal behavior. This allows the detection unit to continuously monitor agent behavior and achieve early detection of abnormal behavior. In addition, the detection unit can detect agent behavior data in real time and build models for detecting abnormal behavior. This model is used to analyze agent behavior data and identify patterns of abnormal behavior. For example, the detection unit detects agent behavior data in real time and sets criteria for early detection of abnormal behavior. This allows the detection unit to build a system for real-time detection of agent behavior and early detection of abnormal behavior.

[0086] The blocking unit blocks abnormal behavior detected by the detection unit. For example, the blocking unit can block agent behavior data in real time, enabling early detection of abnormal behavior. Specifically, the blocking unit responds immediately to abnormal behavior detected by the detection unit, either stopping the agent's operation or issuing instructions to correct the abnormal behavior. For example, if an agent attempts an unauthorized operation, the blocking unit disables that operation. Furthermore, the blocking unit blocks agent behavior data in real time and sets criteria for blocking abnormal behavior. This allows the blocking unit to continuously monitor agent behavior and achieve early detection of abnormal behavior. Additionally, the blocking unit can block agent behavior data in real time and build models for blocking abnormal behavior. These models are used to analyze agent behavior data and identify patterns of abnormal behavior. For example, the blocking unit blocks agent behavior data in real time and sets criteria for early detection of abnormal behavior. This allows the blocking unit to build a system that blocks agent behavior in real time and enables early detection of abnormal behavior.

[0087] The notification unit notifies relevant parties of abnormal behavior blocked by the blocking unit. For example, the notification unit can notify agents of their behavior data in real time, enabling early notification of abnormal behavior. Specifically, the notification unit immediately transmits information about abnormal behavior detected by the blocking unit to relevant parties, prompting appropriate action. For example, if an agent attempts to perform an unauthorized operation, the notification unit notifies the administrator of the details and requests action. The notification unit also notifies agents of their behavior data in real time and sets criteria for notifying abnormal behavior. This allows the notification unit to continuously monitor agent behavior and achieve early notification of abnormal behavior. Furthermore, the notification unit can notify agents of their behavior data in real time and build models for notifying abnormal behavior. These models are used to analyze agent behavior data and identify patterns of abnormal behavior. For example, the notification unit notifies agents of their behavior data in real time and sets criteria for early notification of abnormal behavior. This allows the notification unit to build a system for notifying agents of their behavior in real time and for early notification of abnormal behavior. As a result, the AI ​​agent monitoring system according to the embodiment can monitor and detect abnormal behavior of AI agents in real time and automatically prevent it.

[0088] The analysis unit can analyze communication between agents. For example, the analysis unit can analyze the type of message, communication protocol, and data format between agents. For example, the analysis unit can analyze the content of messages between agents to detect malicious instructions or communications. The analysis unit can also analyze the communication protocol between agents to detect malicious communications. Furthermore, the analysis unit can analyze the data format between agents to detect malicious data. For example, the analysis unit can analyze the type of message between agents to detect malicious instructions. In this way, by analyzing communication between agents, the analysis unit can detect malicious instructions and communications and detect abnormal behavior at an early stage.

[0089] The visualization unit can visualize and analyze agent operation logs. For example, the visualization unit visualizes agent operation history, system events, and error logs. For instance, it can display agent operation history as graphs or charts, making it easy to understand the agent's behavior history. The visualization unit can also display system events in a timeline format, allowing for a visual understanding of the agent's operation flow. Furthermore, the visualization unit can display error logs in a list format, allowing for an understanding of the agent's error occurrences. For example, the visualization unit displays agent operation history as a graph, providing a visual understanding of the agent's behavior history. In this way, the visualization unit can easily understand the agent's behavior history by visualizing and analyzing agent operation logs.

[0090] The data collection unit can estimate the user's emotions and adjust the timing of behavioral data collection based on the estimated emotions. For example, if the user is stressed, the data collection unit can delay the collection timing to reduce the user's burden. For example, if the user is relaxed, the data collection unit can advance the collection timing to collect more detailed data. Furthermore, if the user is in a hurry, the data collection unit can optimize the collection timing to quickly collect data. For example, the data collection unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the data collection unit can calculate an emotion score based on changes in facial expressions and adjust the collection timing accordingly. In this way, the data collection unit can reduce the user's burden and collect more detailed data by adjusting the timing of behavioral data collection according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or generative AI. Generative AI includes, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above-described processes in the data collection unit may be performed using AI, for example, or without AI. For example, the data collection unit can input image data of the user captured by the camera into a generating AI, which can then perform the estimation of the user's emotions.

[0091] The data collection unit can analyze the agent's past behavioral data and select the optimal data collection method. For example, the data collection unit optimizes the data collection method based on actions the agent has frequently performed in the past. For example, the data collection unit can customize the data collection method based on the agent's past behavioral data. The data collection unit can also analyze the agent's behavioral patterns and select an efficient data collection method. For example, the data collection unit optimizes the data collection method based on the agent's past behavioral data. This allows the data collection unit to select an efficient data collection method by analyzing the agent's past behavioral data. Some or all of the above processing in the data collection unit may be performed using AI, for example, or without AI. For example, the data collection unit can input the agent's past behavioral data into a generating AI and have the generating AI select the optimal data collection method.

[0092] The data collection unit can filter behavioral data based on the agent's current task and environment. For example, the data collection unit can collect only relevant data based on the task the agent is currently performing. The data collection unit can filter the necessary data by considering the agent's environmental information. The data collection unit can also select the data to collect based on the priority of the agent's tasks. For example, the data collection unit can collect only relevant data based on the task the agent is currently performing. This allows the data collection unit to collect only relevant data by filtering the data based on the agent's current task and environment. Some or all of the above processing in the data collection unit may be performed using AI, for example, or without AI. For example, the data collection unit can input the agent's current task and environmental information into a generating AI and have the generating AI perform the filtering.

[0093] The data collection unit can estimate the user's emotions and determine the priority of behavioral data to collect based on the estimated user emotions. For example, if the user is stressed, the data collection unit will prioritize collecting important data. For example, if the user is relaxed, the data collection unit can prioritize collecting detailed data. Also, if the user is in a hurry, the data collection unit can prioritize data that can be collected quickly. For example, the data collection unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the data collection unit can calculate an emotion score based on changes in facial expressions and determine the priority of behavioral data to collect. In this way, the data collection unit can prioritize the collection of important data by determining the priority of behavioral data to collect according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the data collection unit may be performed using AI, for example, or without using AI. For example, the data collection unit can input image data of the user captured by the camera into a generating AI, which can then perform the estimation of the user's emotions.

[0094] The data collection unit can prioritize the collection of highly relevant data by considering the agent's geographical location information when collecting behavioral data. For example, if the agent is in a specific region, the data collection unit will prioritize the collection of data related to that region. The data collection unit can also select highly relevant data based on the agent's location information. Furthermore, the data collection unit can optimize the collected data by considering the agent's movement patterns. For example, if the agent is in a specific region, the data collection unit will prioritize the collection of data related to that region. In this way, the data collection unit can prioritize the collection of highly relevant data by considering the agent's geographical location information. Some or all of the above processing in the data collection unit may be performed using AI, for example, or without AI. For example, the data collection unit can input the agent's geographical location information into a generating AI and have the generating AI perform the selection of highly relevant data.

[0095] The data collection unit can analyze the agent's social media activity and collect relevant data when collecting behavioral data. For example, the data collection unit can analyze the agent's social media posts and collect relevant data. For example, the data collection unit can extract important data from the agent's social media activity. The data collection unit can also select the data to be collected by considering the agent's level of social media involvement. For example, the data collection unit can analyze the agent's social media posts and collect relevant data. In this way, the data collection unit can collect relevant data by analyzing the agent's social media activity. Some or all of the above processing in the data collection unit may be performed using AI, for example, or without AI. For example, the data collection unit can input the agent's social media activity data into a generating AI and have the generating AI perform the extraction of relevant data.

[0096] The learning unit can estimate the user's emotions and select training data based on the estimated emotions. For example, if the user is stressed, the learning unit will prioritize learning simple data. For example, if the user is relaxed, the learning unit can prioritize learning detailed data. Also, if the user is in a hurry, the learning unit can select data that can be learned quickly. For example, the learning unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the learning unit can calculate an emotion score based on changes in facial expressions and select training data. In this way, the learning unit can improve the efficiency of learning by selecting training data according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the learning unit may be performed using AI, for example, or without AI. For example, the learning unit can input user image data captured by a camera into a generative AI, allowing the generative AI to estimate the user's emotions.

[0097] The learning unit can optimize the learning algorithm by referring to the agent's past behavior data during learning. For example, the learning unit can adjust the learning algorithm based on the agent's past behavior data. The learning unit can, for example, analyze the agent's behavior patterns and select the optimal learning algorithm. The learning unit can also optimize the learning algorithm by referring to the agent's past data. For example, the learning unit can adjust the learning algorithm based on the agent's past behavior data. In this way, the learning unit can optimize the learning algorithm by referring to the agent's past behavior data and improve the accuracy of learning. Some or all of the above processing in the learning unit may be performed using AI, for example, or without using AI. For example, the learning unit can input the agent's past behavior data into a generating AI and have the generating AI perform the optimization of the learning algorithm.

[0098] The learning unit can apply different learning methods to each agent task during training. For example, the learning unit can select an appropriate learning method according to the agent task. For example, the learning unit can apply different learning algorithms to each agent task. The learning unit can also customize the learning method based on the characteristics of the agent task. For example, the learning unit can select an appropriate learning method according to the agent task. This allows the learning unit to improve the accuracy of learning by applying different learning methods to each agent task. Some or all of the above processing in the learning unit may be performed using AI, for example, or without AI. For example, the learning unit can input agent task information into a generating AI and have the generating AI select an appropriate learning method.

[0099] The learning unit can estimate the user's emotions and adjust the learning frequency based on the estimated emotions. For example, if the user is stressed, the learning unit can reduce the learning frequency to alleviate the burden. For example, if the user is relaxed, the learning unit can increase the learning frequency to learn more detailed data. The learning unit can also optimize the learning frequency to learn quickly if the user is in a hurry. For example, the learning unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the learning unit can calculate an emotion score based on changes in facial expressions and adjust the learning frequency. In this way, the learning unit can reduce the burden on the user and improve the efficiency of learning by adjusting the learning frequency according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the learning unit may be performed using AI, for example, or without AI. For example, the learning unit can input user image data captured by a camera into a generative AI, allowing the generative AI to estimate the user's emotions.

[0100] The learning unit can weight the training data during training based on when the agent's behavior data is submitted. For example, the learning unit can weight important data based on when the agent's behavior data is submitted. The learning unit can optimize the training data by considering when the agent's behavior data is submitted. The learning unit can also adjust the weighting of the training data according to when the agent's behavior data is submitted. For example, the learning unit can weight important data based on when the agent's behavior data is submitted. This allows the learning unit to improve the accuracy of training by weighting the training data based on when the agent's behavior data is submitted. Some or all of the above processing in the learning unit may be performed using AI, for example, or without AI. For example, the learning unit can input information about when the agent's behavior data is submitted into a generating AI and have the generating AI perform the weighting of the training data.

[0101] The learning unit can improve the accuracy of learning by referring to the agent's relevant literature during training. For example, the learning unit can optimize the learning algorithm by referring to the agent's relevant literature. The learning unit can supplement the training data based on the agent's relevant literature. The learning unit can also improve the accuracy of learning by referring to the agent's relevant literature. For example, the learning unit can optimize the learning algorithm by referring to the agent's relevant literature. In this way, the learning unit can improve the accuracy of learning by referring to the agent's relevant literature. Some or all of the above processing in the learning unit may be performed using AI, for example, or without using AI. For example, the learning unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the improvement of learning accuracy.

[0102] The monitoring unit can estimate the user's emotions and adjust the monitoring criteria based on the estimated emotions. For example, if the user is stressed, the monitoring unit can relax the monitoring criteria to reduce the burden. For example, if the user is relaxed, the monitoring unit can tighten the monitoring criteria to monitor detailed data. Also, if the user is in a hurry, the monitoring unit can optimize the monitoring criteria to monitor quickly. For example, the monitoring unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the monitoring unit can calculate an emotion score based on changes in facial expressions and adjust the monitoring criteria. In this way, the monitoring unit can reduce the burden on the user and monitor detailed data by adjusting the monitoring criteria according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without AI. For example, the monitoring unit can input image data of the user captured by the camera into a generating AI, and have the generating AI perform the estimation of the user's emotions.

[0103] The monitoring unit can improve the accuracy of monitoring by considering the interrelationships between agents during monitoring. For example, the monitoring unit can analyze communication between agents to improve monitoring accuracy. For example, the monitoring unit can optimize monitoring data by considering the interrelationships between agents. The monitoring unit can also adjust the accuracy of monitoring based on the interrelationships between agents. For example, the monitoring unit can analyze communication between agents to improve monitoring accuracy. In this way, the monitoring unit can improve the accuracy of monitoring by considering the interrelationships between agents. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without using AI. For example, the monitoring unit can input agent interrelationship data into a generating AI and have the generating AI perform the improvement of monitoring accuracy.

[0104] The monitoring unit can perform monitoring while considering the attribute information of the agent's task. For example, the monitoring unit can select monitoring data based on the attribute information of the agent's task. For example, the monitoring unit can improve the accuracy of monitoring by considering the attribute information of the agent's task. The monitoring unit can also adjust the monitoring criteria based on the attribute information of the agent's task. For example, the monitoring unit can select monitoring data based on the attribute information of the agent's task. In this way, the monitoring unit can improve the accuracy of monitoring by considering the attribute information of the agent's task. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without using AI. For example, the monitoring unit can input the agent's task attribute information into a generating AI and have the generating AI perform the task to improve the accuracy of monitoring.

[0105] The monitoring unit can estimate the user's emotions and adjust the order in which monitoring results are displayed based on the estimated user emotions. For example, if the user is stressed, the monitoring unit can prioritize displaying important results. For example, if the user is relaxed, the monitoring unit can prioritize displaying detailed results. The monitoring unit can also prioritize results that can be quickly viewed if the user is in a hurry. For example, the monitoring unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the monitoring unit can calculate an emotion score based on changes in facial expressions and adjust the order in which monitoring results are displayed. In this way, the monitoring unit can prioritize displaying important results by adjusting the order in which monitoring results are displayed according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without AI. For example, the monitoring unit can input image data of the user captured by the camera into a generating AI, and have the generating AI perform the estimation of the user's emotions.

[0106] The monitoring unit can perform monitoring while considering the geographical distribution of agents. For example, the monitoring unit can select monitoring data based on the geographical distribution of agents. For example, the monitoring unit can improve the accuracy of monitoring by considering the geographical distribution of agents. The monitoring unit can also adjust the monitoring criteria based on the geographical distribution of agents. For example, the monitoring unit can select monitoring data based on the geographical distribution of agents. In this way, the monitoring unit can improve the accuracy of monitoring by considering the geographical distribution of agents. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without using AI. For example, the monitoring unit can input the geographical distribution data of agents into a generating AI and have the generating AI perform the task of improving the accuracy of monitoring.

[0107] The monitoring unit can improve the accuracy of monitoring by referring to the agent's relevant literature during monitoring. For example, the monitoring unit can optimize the monitoring algorithm by referring to the agent's relevant literature. The monitoring unit can supplement monitoring data based on the agent's relevant literature. The monitoring unit can also improve the accuracy of monitoring by referring to the agent's relevant literature. For example, the monitoring unit can optimize the monitoring algorithm by referring to the agent's relevant literature. In this way, the monitoring unit can improve the accuracy of monitoring by referring to the agent's relevant literature. Some or all of the above processing in the monitoring unit may be performed using AI, for example, or without using AI. For example, the monitoring unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the improvement of monitoring accuracy.

[0108] The detection unit can estimate the user's emotions and adjust the detection criteria for abnormal behavior based on the estimated user emotions. For example, if the user is stressed, the detection unit can relax the detection criteria to reduce the burden. For example, if the user is relaxed, the detection unit can tighten the detection criteria to detect more detailed data. The detection unit can also optimize the detection criteria to detect quickly if the user is in a hurry. For example, the detection unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the detection unit can calculate an emotion score based on changes in facial expressions and adjust the detection criteria for abnormal behavior. In this way, the detection unit can reduce the burden on the user and detect more detailed data by adjusting the detection criteria for abnormal behavior according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the detection unit may be performed using AI, for example, or without using AI. For example, the detection unit can input image data of the user captured by the camera into a generating AI, allowing the generating AI to perform the estimation of the user's emotions.

[0109] The detection unit can optimize its detection algorithm by referring to the agent's past abnormal behavior data when detection occurs. For example, the detection unit adjusts the detection algorithm based on the agent's past abnormal behavior data. For example, the detection unit can analyze the agent's abnormal behavior patterns and select the optimal detection algorithm. The detection unit can also optimize its detection algorithm by referring to the agent's past data. For example, the detection unit adjusts the detection algorithm based on the agent's past abnormal behavior data. In this way, the detection unit can optimize its detection algorithm by referring to the agent's past abnormal behavior data and improve the accuracy of detection. Some or all of the above processing in the detection unit may be performed using AI, for example, or without using AI. For example, the detection unit can input the agent's past abnormal behavior data into a generating AI and have the generating AI perform the optimization of the detection algorithm.

[0110] The detection unit can apply different detection methods to each agent's task when detection occurs. For example, the detection unit can select an appropriate detection method according to the agent's task. For example, the detection unit can apply different detection algorithms to each agent's task. The detection unit can also customize detection methods based on the characteristics of the agent's task. For example, the detection unit can select an appropriate detection method according to the agent's task. This allows the detection unit to improve detection accuracy by applying different detection methods to each agent's task. Some or all of the above processing in the detection unit may be performed using AI, for example, or without AI. For example, the detection unit can input agent task information into a generating AI and have the generating AI select an appropriate detection method.

[0111] The detection unit can estimate the user's emotions and determine detection priorities based on the estimated emotions. For example, if the user is stressed, the detection unit will prioritize detecting important abnormal behaviors. For example, if the user is relaxed, the detection unit can prioritize detecting detailed abnormal behaviors. Also, if the user is in a hurry, the detection unit can prioritize abnormal behaviors that can be detected quickly. For example, the detection unit can capture the user's facial expression with a camera and estimate their emotions using an emotion estimation algorithm. For example, the detection unit can calculate an emotion score based on changes in facial expression and determine detection priorities. In this way, the detection unit can prioritize the detection of important abnormal behaviors by determining detection priorities according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the detection unit may be performed using AI, for example, or without using AI. For example, the detection unit can input image data of the user captured by the camera into a generating AI, allowing the generating AI to perform the estimation of the user's emotions.

[0112] The detection unit can weight the detected data based on the timing of the agent's behavior data submission when detection occurs. For example, the detection unit weights important data based on the timing of the agent's behavior data submission. The detection unit can optimize the detected data by considering the timing of the agent's behavior data submission. The detection unit can also adjust the weighting of the detected data according to the timing of the agent's behavior data submission. For example, the detection unit weights important data based on the timing of the agent's behavior data submission. This allows the detection unit to improve detection accuracy by weighting the detected data based on the timing of the agent's behavior data submission. Some or all of the above processing in the detection unit may be performed using AI, for example, or without AI. For example, the detection unit can input information on the timing of the agent's behavior data submission into a generating AI and have the generating AI perform the weighting of the detected data.

[0113] The detection unit can improve detection accuracy by referring to the agent's relevant literature during detection. For example, the detection unit optimizes the detection algorithm by referring to the agent's relevant literature. The detection unit can supplement detection data based on the agent's relevant literature. The detection unit can also improve detection accuracy by referring to the agent's relevant literature. For example, the detection unit optimizes the detection algorithm by referring to the agent's relevant literature. In this way, the detection unit can improve detection accuracy by referring to the agent's relevant literature. Some or all of the above processing in the detection unit may be performed using AI, for example, or without AI. For example, the detection unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the detection accuracy improvement.

[0114] The blocking unit can estimate the user's emotions and adjust the method of preventing abnormal behavior based on the estimated user emotions. For example, if the user is stressed, the blocking unit can apply a quick and simple blocking method. For example, if the user is relaxed, the blocking unit can apply a more detailed blocking method. The blocking unit can also select a method that can quickly prevent the behavior if the user is in a hurry. For example, the blocking unit can capture the user's facial expression with a camera and estimate the emotion using an emotion estimation algorithm. For example, the blocking unit can calculate an emotion score based on changes in facial expression and adjust the method of preventing abnormal behavior. In this way, the blocking unit can quickly and appropriately prevent abnormal behavior by adjusting the method of preventing abnormal behavior according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without using AI. For example, the blocking unit can input image data of the user captured by the camera into a generating AI, allowing the generating AI to perform an estimation of the user's emotions.

[0115] The blocking unit can optimize its blocking algorithm by referring to the agent's past abnormal behavior data when blocking. For example, the blocking unit adjusts the blocking algorithm based on the agent's past abnormal behavior data. For example, the blocking unit can analyze the agent's abnormal behavior patterns and select the optimal blocking algorithm. The blocking unit can also optimize its blocking algorithm by referring to the agent's past data. For example, the blocking unit adjusts the blocking algorithm based on the agent's past abnormal behavior data. In this way, the blocking unit can optimize its blocking algorithm by referring to the agent's past abnormal behavior data and improve the accuracy of blocking. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without using AI. For example, the blocking unit can input the agent's past abnormal behavior data into a generating AI and have the generating AI perform the optimization of the blocking algorithm.

[0116] The blocking unit can apply different blocking methods to each agent task when blocking. For example, the blocking unit can select an appropriate blocking method according to the agent task. For example, the blocking unit can apply different blocking algorithms to each agent task. The blocking unit can also customize blocking methods based on the characteristics of the agent task. For example, the blocking unit can select an appropriate blocking method according to the agent task. This allows the blocking unit to improve the accuracy of blocking by applying different blocking methods to each agent task. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without AI. For example, the blocking unit can input agent task information into a generating AI and have the generating AI select an appropriate blocking method.

[0117] The blocking unit can estimate the user's emotions and determine blocking priorities based on the estimated user emotions. For example, if the user is stressed, the blocking unit will prioritize blocking important abnormal behaviors. For example, if the user is relaxed, the blocking unit can prioritize blocking minor abnormal behaviors. Also, if the user is in a hurry, the blocking unit can prioritize abnormal behaviors that can be blocked quickly. For example, the blocking unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the blocking unit can calculate an emotion score based on changes in facial expressions and determine blocking priorities. In this way, the blocking unit can prioritize blocking important abnormal behaviors by determining blocking priorities according to the user's emotions. Emotion estimation is implemented using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without using AI. For example, the blocking unit can input image data of the user captured by the camera into a generating AI, allowing the generating AI to perform an estimation of the user's emotions.

[0118] The blocking unit can weight the blocking data based on the timing of the agent's behavior data submission when blocking. For example, the blocking unit can weight important data based on the timing of the agent's behavior data submission. The blocking unit can optimize the blocking data by considering the timing of the agent's behavior data submission. The blocking unit can also adjust the weighting of the blocking data according to the timing of the agent's behavior data submission. For example, the blocking unit can weight important data based on the timing of the agent's behavior data submission. This allows the blocking unit to improve the accuracy of blocking by weighting the blocking data based on the timing of the agent's behavior data submission. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without AI. For example, the blocking unit can input information on the timing of the agent's behavior data submission into a generating AI and have the generating AI perform the weighting of the blocking data.

[0119] The blocking unit can improve the accuracy of blocking by referring to the agent's relevant literature during blocking. For example, the blocking unit can optimize the blocking algorithm by referring to the agent's relevant literature. The blocking unit can supplement the blocking data based on the agent's relevant literature. The blocking unit can also improve the accuracy of blocking by referring to the agent's relevant literature. For example, the blocking unit can optimize the blocking algorithm by referring to the agent's relevant literature. In this way, the blocking unit can improve the accuracy of blocking by referring to the agent's relevant literature. Some or all of the above processing in the blocking unit may be performed using AI, for example, or without AI. For example, the blocking unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the improvement of blocking accuracy.

[0120] The notification unit can estimate the user's emotions and adjust the notification method based on the estimated emotions. For example, if the user is stressed, the notification unit can apply a concise and easy-to-understand notification method. For example, if the user is relaxed, the notification unit can apply a detailed notification method. The notification unit can also select a method that allows for quick notification if the user is in a hurry. For example, the notification unit can capture the user's facial expression with a camera and estimate their emotions using an emotion estimation algorithm. For example, the notification unit can calculate an emotion score based on changes in facial expression and adjust the notification method. In this way, the notification unit can provide quick and appropriate notifications by adjusting the notification method according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the notification unit may be performed using AI, for example, or without AI. For example, the notification unit can input image data of the user captured by the camera into a generating AI, and have the generating AI perform an estimation of the user's emotions.

[0121] The notification unit can optimize the notification algorithm by referring to the agent's past abnormal behavior data when issuing a notification. For example, the notification unit can adjust the notification algorithm based on the agent's past abnormal behavior data. For example, the notification unit can analyze the agent's abnormal behavior patterns and select the optimal notification algorithm. The notification unit can also optimize the notification algorithm by referring to the agent's past data. For example, the notification unit can adjust the notification algorithm based on the agent's past abnormal behavior data. In this way, the notification unit can optimize the notification algorithm by referring to the agent's past abnormal behavior data and improve the accuracy of notifications. Some or all of the above processing in the notification unit may be performed using AI, for example, or without using AI. For example, the notification unit can input the agent's past abnormal behavior data into a generating AI and have the generating AI perform the optimization of the notification algorithm.

[0122] The notification unit can apply different notification methods to each agent's task when issuing notifications. For example, the notification unit can select an appropriate notification method according to the agent's task. For example, the notification unit can apply different notification algorithms to each agent's task. The notification unit can also customize notification methods based on the characteristics of the agent's task. For example, the notification unit can select an appropriate notification method according to the agent's task. This allows the notification unit to improve notification accuracy by applying different notification methods to each agent's task. Some or all of the above processing in the notification unit may be performed using AI, for example, or without AI. For example, the notification unit can input agent task information into a generating AI and have the generating AI select an appropriate notification method.

[0123] The notification unit can estimate the user's emotions and determine notification priorities based on the estimated emotions. For example, if the user is stressed, the notification unit will prioritize important notifications. For example, if the user is relaxed, the notification unit will prioritize detailed notifications. The notification unit can also prioritize content that can be delivered quickly if the user is in a hurry. For example, the notification unit can capture the user's facial expression with a camera and estimate their emotions using an emotion estimation algorithm. For example, the notification unit can calculate an emotion score based on changes in facial expression and determine notification priorities. In this way, the notification unit can prioritize important notifications by determining notification priorities according to the user's emotions. Emotion estimation is implemented using an emotion estimation function, for example, an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the notification unit may be performed using AI, for example, or without AI. For example, the notification unit can input image data of the user captured by the camera into a generating AI, and have the generating AI perform an estimation of the user's emotions.

[0124] The notification unit can weight notification data based on the timing of agent behavior data submission at the time of notification. For example, the notification unit can weight important data based on the timing of agent behavior data submission. The notification unit can optimize notification data by considering the timing of agent behavior data submission. The notification unit can also adjust the weighting of notification data according to the timing of agent behavior data submission. For example, the notification unit can weight important data based on the timing of agent behavior data submission. This allows the notification unit to improve the accuracy of notifications by weighting notification data based on the timing of agent behavior data submission. Some or all of the above processing in the notification unit may be performed using AI, for example, or without AI. For example, the notification unit can input agent behavior data submission timing information into a generating AI and have the generating AI perform the weighting of notification data.

[0125] The notification unit can improve the accuracy of notifications by referring to the agent's relevant literature at the time of notification. For example, the notification unit can optimize the notification algorithm by referring to the agent's relevant literature. The notification unit can supplement notification data based on the agent's relevant literature. The notification unit can also improve the accuracy of notifications by referring to the agent's relevant literature. For example, the notification unit can optimize the notification algorithm by referring to the agent's relevant literature. In this way, the notification unit can improve the accuracy of notifications by referring to the agent's relevant literature. Some or all of the above processing in the notification unit may be performed using AI, for example, or without using AI. For example, the notification unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the improvement of notification accuracy.

[0126] The analysis unit can estimate the user's emotions and adjust the analysis method based on the estimated emotions. For example, if the user is stressed, the analysis unit can apply a simple and easy-to-understand analysis method. For example, if the user is relaxed, the analysis unit can apply a detailed analysis method. The analysis unit can also select a method that allows for rapid analysis if the user is in a hurry. For example, the analysis unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the analysis unit can calculate an emotion score based on changes in facial expressions and adjust the analysis method. In this way, the analysis unit can perform analysis quickly and appropriately by adjusting the analysis method according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above-described processes in the analysis unit may be performed using AI, for example, or without AI. For example, the analysis unit can input user image data captured by a camera into a generating AI, allowing the AI ​​to estimate the user's emotions.

[0127] The analysis unit can optimize the analysis algorithm by referring to the agent's past communication data during analysis. For example, the analysis unit adjusts the analysis algorithm based on the agent's past communication data. For example, the analysis unit can analyze the agent's communication patterns and select the optimal analysis algorithm. The analysis unit can also optimize the analysis algorithm by referring to the agent's past data. For example, the analysis unit adjusts the analysis algorithm based on the agent's past communication data. In this way, the analysis unit can optimize the analysis algorithm by referring to the agent's past communication data and improve the accuracy of the analysis. Some or all of the above processing in the analysis unit may be performed using AI, for example, or without using AI. For example, the analysis unit can input the agent's past communication data into a generating AI and have the generating AI perform the optimization of the analysis algorithm.

[0128] The analysis unit can apply different analysis methods to each agent task during analysis. For example, the analysis unit can select an appropriate analysis method according to the agent task. For example, the analysis unit can apply different analysis algorithms to each agent task. Furthermore, the analysis unit can customize the analysis method based on the characteristics of the agent task. For example, the analysis unit can select an appropriate analysis method according to the agent task. This allows the analysis unit to improve the accuracy of the analysis by applying different analysis methods to each agent task. Some or all of the above-described processes in the analysis unit may be performed using AI, for example, or without AI. For example, the analysis unit can input agent task information into a generating AI and have the generating AI select an appropriate analysis method.

[0129] The analysis unit can estimate the user's emotions and determine the priority of analysis based on the estimated emotions. For example, if the user is stressed, the analysis unit will prioritize important analyses. For example, if the user is relaxed, the analysis unit will prioritize detailed analyses. Also, if the user is in a hurry, the analysis unit can prioritize content that can be analyzed quickly. For example, the analysis unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the analysis unit can calculate an emotion score based on changes in facial expressions and determine the priority of analysis. In this way, the analysis unit can prioritize important analyses by determining the priority of analysis according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the analysis unit may be performed using AI, for example, or without using AI. For example, the analysis unit can input user image data captured by a camera into a generating AI, allowing the AI ​​to estimate the user's emotions.

[0130] The analysis unit can weight the analysis data based on when the agent's communication data was submitted during the analysis. For example, the analysis unit can weight important data based on when the agent's communication data was submitted. The analysis unit can optimize the analysis data by considering when the agent's communication data was submitted. The analysis unit can also adjust the weighting of the analysis data according to when the agent's communication data was submitted. For example, the analysis unit can weight important data based on when the agent's communication data was submitted. This allows the analysis unit to improve the accuracy of the analysis by weighting the analysis data based on when the agent's communication data was submitted. Some or all of the above processing in the analysis unit may be performed using AI, for example, or without AI. For example, the analysis unit can input information on when the agent's communication data was submitted into a generating AI and have the generating AI perform the weighting of the analysis data.

[0131] The analysis unit can improve the accuracy of its analysis by referring to the agent's relevant literature during the analysis. For example, the analysis unit can optimize its analysis algorithm by referring to the agent's relevant literature. The analysis unit can supplement its analysis data based on the agent's relevant literature. Furthermore, the analysis unit can improve the accuracy of its analysis by referring to the agent's relevant literature. For example, the analysis unit can optimize its analysis algorithm by referring to the agent's relevant literature. In this way, the analysis unit can improve the accuracy of its analysis by referring to the agent's relevant literature. Some or all of the above processes in the analysis unit may be performed using AI, for example, or without AI. For example, the analysis unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the analysis accuracy improvement.

[0132] The visualization unit can estimate the user's emotions and adjust the visualization method based on the estimated emotions. For example, if the user is stressed, the visualization unit can apply a simple and easy-to-understand visualization method. For example, if the user is relaxed, the visualization unit can apply a detailed visualization method. The visualization unit can also select a method that allows for quick visualization if the user is in a hurry. For example, the visualization unit can capture the user's facial expression with a camera and estimate the emotion using an emotion estimation algorithm. For example, the visualization unit can calculate an emotion score based on changes in facial expression and adjust the visualization method. In this way, the visualization unit can perform visualization quickly and appropriately by adjusting the visualization method according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, using an emotion engine or generative AI. Generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without AI. For example, the visualization unit can input image data of the user captured by the camera into a generating AI, which can then perform the estimation of the user's emotions.

[0133] The visualization unit can optimize the visualization algorithm by referring to the agent's past operation logs during visualization. For example, the visualization unit adjusts the visualization algorithm based on the agent's past operation logs. For example, the visualization unit can analyze the agent's operation patterns and select the optimal visualization algorithm. The visualization unit can also optimize the visualization algorithm by referring to the agent's past data. For example, the visualization unit adjusts the visualization algorithm based on the agent's past operation logs. In this way, the visualization unit can optimize the visualization algorithm by referring to the agent's past operation logs and improve the accuracy of the visualization. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without AI. For example, the visualization unit can input the agent's past operation logs into a generating AI and have the generating AI perform the optimization of the visualization algorithm.

[0134] The visualization unit can apply different visualization methods to each agent task during visualization. For example, the visualization unit can select an appropriate visualization method according to the agent task. For example, the visualization unit can apply different visualization algorithms to each agent task. The visualization unit can also customize the visualization method based on the characteristics of the agent task. For example, the visualization unit can select an appropriate visualization method according to the agent task. This allows the visualization unit to improve the accuracy of visualization by applying different visualization methods to each agent task. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without AI. For example, the visualization unit can input agent task information into a generating AI and have the generating AI select an appropriate visualization method.

[0135] The visualization unit can estimate the user's emotions and determine the priority of visualizations based on the estimated emotions. For example, if the user is stressed, the visualization unit will prioritize important visualizations. For example, if the user is relaxed, the visualization unit will prioritize detailed visualizations. Also, if the user is in a hurry, the visualization unit can prioritize content that can be visualized quickly. For example, the visualization unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. For example, the visualization unit can calculate an emotion score based on changes in facial expressions and determine the priority of visualizations. In this way, the visualization unit can prioritize important visualizations by determining the priority of visualizations according to the user's emotions. Emotion estimation is achieved using an emotion estimation function, for example, an emotion engine or a generative AI. The generative AI is, but is not limited to, text generation AI (e.g., LLM) or multimodal generation AI. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without AI. For example, the visualization unit can input image data of the user captured by the camera into a generating AI, which can then perform the estimation of the user's emotions.

[0136] The visualization unit can weight the visualization data based on the timing of agent operation log submissions during visualization. For example, the visualization unit can weight important data based on the timing of agent operation log submissions. The visualization unit can optimize the visualization data by considering the timing of agent operation log submissions. The visualization unit can also adjust the weighting of the visualization data according to the timing of agent operation log submissions. For example, the visualization unit can weight important data based on the timing of agent operation log submissions. This allows the visualization unit to improve the accuracy of visualization by weighting the visualization data based on the timing of agent operation log submissions. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without AI. For example, the visualization unit can input agent operation log submission timing information into a generating AI and have the generating AI perform the weighting of the visualization data.

[0137] The visualization unit can improve the accuracy of visualization by referring to the agent's relevant literature during visualization. For example, the visualization unit can optimize the visualization algorithm by referring to the agent's relevant literature. The visualization unit can supplement the visualization data based on the agent's relevant literature. The visualization unit can also improve the accuracy of visualization by referring to the agent's relevant literature. For example, the visualization unit can optimize the visualization algorithm by referring to the agent's relevant literature. In this way, the visualization unit can improve the accuracy of visualization by referring to the agent's relevant literature. Some or all of the above processing in the visualization unit may be performed using AI, for example, or without using AI. For example, the visualization unit can input the agent's relevant literature information into a generating AI and have the generating AI perform the visualization accuracy improvement.

[0138] The system according to the embodiment is not limited to the example described above, and various modifications are possible, for example, as follows.

[0139] The data collection unit can optimize its collection method by considering the frequency and patterns of agent behavior when collecting agent behavior data. For example, if an agent frequently performs a particular action, it can prioritize the collection of data related to that action. It can also analyze agent behavior patterns and focus on collecting data useful for detecting abnormal behavior. Furthermore, it can collect agent behavior data in real time to provide data useful for the early detection of abnormal behavior. In this way, the data collection unit can efficiently collect agent behavior data and provide data useful for the early detection of abnormal behavior.

[0140] The analysis unit can optimize its analysis method by considering the frequency and patterns of agent communication when analyzing communication between agents. For example, it can analyze the frequency and patterns of messages between agents to detect abnormal communication. It can also analyze the content of communication between agents to detect malicious instructions or communications. Furthermore, it can analyze the communication protocols and data formats between agents to detect malicious communications. As a result, the analysis unit can efficiently analyze communication between agents and detect abnormal communication at an early stage.

[0141] The visualization unit can optimize the visualization method when visualizing agent operation logs, taking into account the frequency and patterns of agent operations. For example, it can display the agent's operation history as graphs or charts to visually understand the agent's behavior patterns. It can also display the flow of agent operations in a timeline format to visually understand the overall picture of agent operations. Furthermore, it can display agent error logs in a list format to understand the agent's error occurrence status. In this way, the visualization unit can efficiently visualize agent operation logs and easily understand agent behavior patterns.

[0142] The data collection unit can estimate the user's emotions and adjust the method of collecting behavioral data based on the estimated emotions. For example, if the user is stressed, the collection method can be simplified to reduce the user's burden. Conversely, if the user is relaxed, detailed data can be collected. Furthermore, if the user is in a hurry, a method that allows for rapid collection can be selected. For example, the data collection unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. This allows the data collection unit to adjust the method of collecting behavioral data according to the user's emotions, thereby reducing the user's burden and collecting detailed data.

[0143] The data collection unit can analyze the agent's past behavioral data and select the optimal collection method. For example, it can optimize the collection method based on actions the agent has frequently performed in the past. It can also analyze the agent's behavioral patterns and select an efficient collection method. Furthermore, it can customize the collection method based on the agent's behavioral data. As a result, the data collection unit can efficiently collect agent behavioral data by analyzing the agent's past behavioral data and selecting an efficient collection method.

[0144] The data collection unit can filter behavioral data based on the agent's current task and environment. For example, it can collect only relevant data based on the task the agent is currently performing. It can also filter necessary data considering the agent's environmental information. Furthermore, it can select collected data based on the priority of the agent's tasks. In this way, the data collection unit can collect only relevant data by filtering it based on the agent's current task and environment.

[0145] The data collection unit can estimate the user's emotions and prioritize the behavioral data to collect based on those emotions. For example, if the user is stressed, it can prioritize collecting important data. If the user is relaxed, it can prioritize collecting detailed data. Furthermore, if the user is in a hurry, it can prioritize data that can be collected quickly. For instance, the data collection unit captures the user's facial expressions with a camera and estimates their emotions using an emotion estimation algorithm. This allows the data collection unit to prioritize the collection of important data by determining the priority of behavioral data according to the user's emotions.

[0146] The data collection unit can prioritize the collection of highly relevant data by considering the agent's geographical location when collecting behavioral data. For example, if an agent is in a specific region, it will prioritize the collection of data related to that region. It can also select highly relevant data based on the agent's location information. Furthermore, it can optimize the collected data by considering the agent's movement patterns. As a result, the data collection unit can prioritize the collection of highly relevant data by considering the agent's geographical location information.

[0147] The data collection unit can analyze the agent's social media activity and collect relevant data when collecting behavioral data. For example, it can analyze the agent's social media posts and collect relevant data. It can also extract important data from the agent's social media activity. Furthermore, it can select the data to be collected considering the agent's level of social media involvement. In this way, the data collection unit can collect relevant data by analyzing the agent's social media activity.

[0148] The learning unit can estimate the user's emotions and select training data based on those estimated emotions. For example, if the user is stressed, it will prioritize learning simple data. If the user is relaxed, it will prioritize learning detailed data. Furthermore, if the user is in a hurry, it can select data that allows for rapid learning. For instance, the learning unit can capture the user's facial expressions with a camera and estimate their emotions using an emotion estimation algorithm. This allows the learning unit to improve learning efficiency by selecting training data according to the user's emotions.

[0149] The following briefly describes the processing flow for example form 2.

[0150] Step 1: The collection unit collects agent behavior data. For example, it collects user operation logs, sensor data, activity logs, etc., and stores them in a database in real time. It can also collect data periodically and provide it for analysis. Furthermore, it filters and collects only the necessary data to provide foundational data for early detection of abnormal behavior. Step 2: The learning unit uses generative AI to analyze the data collected by the collection unit and learns normal behavior patterns. For example, it learns normal behavior patterns using statistical criteria based on past data or rule-based criteria, and sets criteria for identifying abnormal behavior. It also builds a model for early detection of abnormal behavior. Step 3: The monitoring unit monitors in real time based on the normal behavior patterns learned by the learning unit. For example, it monitors agent behavior data in real time, sets criteria for early detection of abnormal behavior, and builds a model for detecting abnormal behavior. Step 4: The detection unit detects the actions monitored by the monitoring unit. For example, it detects agent behavior data in real time, sets criteria for early detection of abnormal behavior, and builds a model for detecting abnormal behavior. Step 5: The blocking unit blocks the abnormal behavior detected by the detection unit. For example, it blocks agent behavior data in real time, sets criteria for early blocking of abnormal behavior, and builds a model for blocking abnormal behavior. Step 6: The notification unit notifies relevant parties of abnormal behavior blocked by the blocking unit. For example, it notifies agent behavior data in real time, sets criteria for early notification of abnormal behavior, and builds a model for notifying of abnormal behavior.

[0151] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0152] Data generation model 58 is a form of so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> Examples of generative AI include text generation AI, image generation AI, and multimodal generation AI. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images (e.g., still image data or video data). The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference result in one or more data formats from audio data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the specific processing described above using the data generation model 58. The data generation model 58 may be a fine-tuned model that outputs inference results from prompts that do not contain instructions, in which case the data generation model 58 can output inference results from prompts that do not contain instructions. In the data processing device 12, etc., there are multiple types of data generation models 58, and the data generation model 58 includes AI other than generative AI. AI other than generative AI includes, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVMs), k-means clustering, convolutional neural networks (CNNs), recurrent neural networks (RNNs), generative adversarial networks (GANs), or naive Bayes, and can perform various processes, but is not limited to these examples. Also, the AI ​​may be an AI agent. Furthermore, when the processing of each of the above parts is performed by the AI, the processing may be performed by the AI ​​in part or in whole, but is not limited to this example.Furthermore, processing performed by AI, including generative AI, may be replaced with rule-based processing, and rule-based processing may be replaced with processing performed by AI, including generative AI.

[0153] Furthermore, the processing performed by the data processing system 10 described above is carried out by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart device 14, but it may also be carried out by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart device 14. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information necessary for processing from the smart device 14 or an external device, and the smart device 14 acquires or collects information necessary for processing from the data processing device 12 or an external device.

[0154] For example, each of the multiple elements, including the collection unit, learning unit, monitoring unit, detection unit, blocking unit, notification unit, analysis unit, and visualization unit, is implemented in at least one of the smart device 14 and the data processing unit 12. For example, the collection unit collects agent behavior data using the camera 42 and sensor data of the smart device 14 and analyzes it using the specific processing unit 290 of the data processing unit 12. The learning unit learns normal behavior patterns using generated AI by the specific processing unit 290 of the data processing unit 12. The monitoring unit monitors the agent's behavior in real time using the control unit 46A of the smart device 14. The detection unit detects abnormal behavior using the specific processing unit 290 of the data processing unit 12. The blocking unit blocks abnormal behavior using the control unit 46A of the smart device 14. The notification unit notifies relevant parties using the specific processing unit 290 of the data processing unit 12. The analysis unit analyzes communication between agents using the specific processing unit 290 of the data processing unit 12. The visualization unit visualizes the agent's operation log, for example, using the display 40A of the smart device 14. The correspondence between each unit and the device or control unit is not limited to the example described above and can be modified in various ways.

[0155] [Second Embodiment] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0156] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0157] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN and / or LAN.

[0158] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0159] The microphone 238 receives voice signals from the user and accepts instructions from the user. The microphone 238 captures the voice signals from the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0160] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, which captures images of the area around the user (for example, an imaging range defined by a field of view equivalent to the field of vision of a typical healthy person).

[0161] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0162] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing by the processor 28. The storage 32 stores the specific processing program 56.

[0163] The processor 28 reads a specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 acting as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0164] Storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the user's emotions using the emotion identification model 59 and perform identification processing using the user's emotions. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotions, including but not limited to these examples. Furthermore, emotion estimation and prediction also include, for example, emotion analysis.

[0165] In the smart glasses 214, specific processing is performed by the processor 46. The storage 50 stores a specific processing program 60. The processor 46 reads the specific processing program 60 from the storage 50 and executes the read specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 acting as a control unit 46A according to the specific processing program 60 executed on the RAM 48. The smart glasses 214 also have a data generation model 58 and an emotion identification model 59, similar to the data generation model and emotion identification model 59, and can perform processing similar to that of the specific processing unit 290 using these models.

[0166] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains processing results (such as prediction results) using the data generation model 58 by communicating with the server device that has the data generation model 58. Also, the data processing device 12 may be a server device or a terminal device owned by the user (for example, a mobile phone, robot, home appliance, etc.).

[0167] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0168] The data generation model 58 is a so-called generative AI. An example of a data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and inference data such as audio data representing speech, text data representing text, and image data representing images (e.g., still image data or video data). The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference result in one or more data formats such as audio data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the specific processing described above using the data generation model 58. The data generation model 58 may be a fine-tuned model that outputs inference results from prompts that do not contain instructions, in which case the data generation model 58 can output inference results from prompts that do not contain instructions. In the data processing device 12, etc., there are multiple types of data generation models 58, and the data generation model 58 includes AI other than generative AI. AI other than generative AI includes, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. Also, the AI ​​may be an AI agent. Furthermore, when the processing of each part described above is performed by the AI, the processing may be performed by the AI ​​in part or in whole, but is not limited to this example. Also, processing performed by an AI including a generative AI may be replaced by rule-based processing, and rule-based processing may be replaced by processing performed by an AI including a generative AI.

[0169] The data processing system 210 according to the second embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 210 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart glasses 214, but it may also be performed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart glasses 214. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information necessary for processing from the smart glasses 214 or an external device, and the smart glasses 214 acquires or collects information necessary for processing from the data processing device 12 or an external device.

[0170] For example, each of the multiple elements, including the collection unit, learning unit, monitoring unit, detection unit, blocking unit, notification unit, analysis unit, and visualization unit, is implemented by, for example, at least one of the smart glasses 214 and the data processing unit 12. For example, the collection unit collects agent behavior data using the camera 42 and sensor data of the smart glasses 214 and analyzes it by the specific processing unit 290 of the data processing unit 12. The learning unit learns normal behavior patterns using generated AI by, for example, the specific processing unit 290 of the data processing unit 12. The monitoring unit monitors the agent's behavior in real time by, for example, the control unit 46A of the smart glasses 214. The detection unit detects abnormal behavior by, for example, the specific processing unit 290 of the data processing unit 12. The blocking unit blocks abnormal behavior by, for example, the control unit 46A of the smart glasses 214. The notification unit notifies relevant parties by, for example, the specific processing unit 290 of the data processing unit 12. The analysis unit analyzes communication between agents by, for example, the specific processing unit 290 of the data processing unit 12. The visualization unit visualizes the agent's operation log, for example, using the display 40A of the smart glasses 214. The correspondence between each unit and the device or control unit is not limited to the example described above and can be modified in various ways.

[0171] [Third Embodiment] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0172] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0173] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN and / or LAN.

[0174] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0175] The microphone 238 receives voice signals from the user and accepts instructions from the user. The microphone 238 captures the voice signals from the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0176] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, which captures images of the area around the user (for example, an imaging range defined by a field of view equivalent to the field of vision of a typical healthy person).

[0177] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0178] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0179] The processor 28 reads a specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 acting as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0180] Storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the user's emotions using the emotion identification model 59 and perform identification processing using the user's emotions. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotions, including but not limited to these examples. Furthermore, emotion estimation and prediction also include, for example, emotion analysis.

[0181] In the headset terminal 314, specific processing is performed by the processor 46. The storage 50 stores a specific program 60. The processor 46 reads the specific program 60 from the storage 50 and executes the read specific program 60 on the RAM 48. The specific processing is realized by the processor 46 acting as a control unit 46A according to the specific program 60 executed on the RAM 48. The headset terminal 314 also has a data generation model 58 and an emotion identification model 59, similar to the data generation model and emotion identification model 59, and can perform processing similar to that of the specific processing unit 290 using these models.

[0182] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains processing results (such as prediction results) using the data generation model 58 by communicating with the server device that has the data generation model 58. Also, the data processing device 12 may be a server device or a terminal device owned by the user (for example, a mobile phone, robot, home appliance, etc.).

[0183] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0184] The data generation model 58 is a so-called generative AI. An example of a data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and inference data such as audio data representing speech, text data representing text, and image data representing images (e.g., still image data or video data). The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference result in one or more data formats such as audio data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the specific processing described above using the data generation model 58. The data generation model 58 may be a fine-tuned model that outputs inference results from prompts that do not contain instructions, in which case the data generation model 58 can output inference results from prompts that do not contain instructions. In the data processing device 12, etc., there are multiple types of data generation models 58, and the data generation model 58 includes AI other than generative AI. AI other than generative AI includes, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. Also, the AI ​​may be an AI agent. Furthermore, when the processing of each part described above is performed by the AI, the processing may be performed by the AI ​​in part or in whole, but is not limited to this example. Also, processing performed by an AI including a generative AI may be replaced by rule-based processing, and rule-based processing may be replaced by processing performed by an AI including a generative AI.

[0185] The data processing system 310 according to the third embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 310 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the headset terminal 314, but may also be performed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the headset terminal 314. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information necessary for processing from the headset terminal 314 or an external device, and the headset terminal 314 acquires or collects information necessary for processing from the data processing device 12 or an external device.

[0186] For example, each of the multiple elements, including the collection unit, learning unit, monitoring unit, detection unit, blocking unit, notification unit, analysis unit, and visualization unit, is implemented by, for example, at least one of the headset terminal 314 and the data processing unit 12. For example, the collection unit collects agent behavior data using the camera 42 and sensor data of the headset terminal 314 and analyzes it by the specific processing unit 290 of the data processing unit 12. The learning unit learns normal behavior patterns using generated AI by, for example, the specific processing unit 290 of the data processing unit 12. The monitoring unit monitors the agent's behavior in real time by, for example, the control unit 46A of the headset terminal 314. The detection unit detects abnormal behavior by, for example, the specific processing unit 290 of the data processing unit 12. The blocking unit blocks abnormal behavior by, for example, the control unit 46A of the headset terminal 314. The notification unit notifies relevant parties by, for example, the specific processing unit 290 of the data processing unit 12. The analysis unit analyzes communication between agents by, for example, the specific processing unit 290 of the data processing unit 12. The visualization unit visualizes the agent's operation log, for example, using the display 40A of the headset terminal 314. The correspondence between each unit and the device or control unit is not limited to the example described above and can be modified in various ways.

[0187] [Fourth Embodiment] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0188] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0189] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN and / or LAN.

[0190] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[0191] The microphone 238 receives voice signals from the user and accepts instructions from the user. The microphone 238 captures the voice signals from the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0192] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS image sensor or CCD image sensor, which captures images of the area around the user (for example, an imaging range defined by a field of view equivalent to the field of vision of a typical healthy person).

[0193] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0194] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. The robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[0195] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0196] The processor 28 reads a specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 acting as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0197] Storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate the user's emotions using the emotion identification model 59 and perform identification processing using the user's emotions. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotions, including but not limited to these examples. Furthermore, emotion estimation and prediction also include, for example, emotion analysis.

[0198] In robot 414, specific processing is performed by processor 46. A specific program 60 is stored in storage 50. Processor 46 reads the specific program 60 from storage 50 and executes it on RAM 48. The specific processing is achieved by processor 46 acting as a control unit 46A according to the specific program 60 executed on RAM 48. Robot 414 also has data generation model 58 and emotion identification model 59, similar to those of the robot, and can perform processing similar to that of the specific processing unit 290 using these models.

[0199] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains processing results (such as prediction results) using the data generation model 58 by communicating with the server device that has the data generation model 58. Also, the data processing device 12 may be a server device or a terminal device owned by the user (for example, a mobile phone, robot, home appliance, etc.).

[0200] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0201] The data generation model 58 is a so-called generative AI. An example of a data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and inference data such as audio data representing speech, text data representing text, and image data representing images (e.g., still image data or video data). The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference result in one or more data formats such as audio data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the specific processing described above using the data generation model 58. The data generation model 58 may be a fine-tuned model that outputs inference results from prompts that do not contain instructions, in which case the data generation model 58 can output inference results from prompts that do not contain instructions. In the data processing device 12, etc., there are multiple types of data generation models 58, and the data generation model 58 includes AI other than generative AI. AI other than generative AI includes, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. Also, the AI ​​may be an AI agent. Furthermore, when the processing of each part described above is performed by the AI, the processing may be performed by the AI ​​in part or in whole, but is not limited to this example. Also, processing performed by an AI including a generative AI may be replaced by rule-based processing, and rule-based processing may be replaced by processing performed by an AI including a generative AI.

[0202] The data processing system 410 according to the fourth embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 410 is performed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the robot 414, but it may also be performed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the robot 414. In addition, the specific processing unit 290 of the data processing device 12 acquires or collects information necessary for processing from the robot 414 or an external device, and the robot 414 acquires or collects information necessary for processing from the data processing device 12 or an external device.

[0203] For example, each of the multiple elements, including the collection unit, learning unit, monitoring unit, detection unit, blocking unit, notification unit, analysis unit, and visualization unit, is implemented by, for example, at least one of the robot 414 and the data processing unit 12. For example, the collection unit collects agent behavior data using the camera 42 and sensor data of the robot 414 and analyzes it by the specific processing unit 290 of the data processing unit 12. The learning unit learns normal behavior patterns using generated AI by, for example, the specific processing unit 290 of the data processing unit 12. The monitoring unit monitors the agent's behavior in real time by, for example, the control unit 46A of the robot 414. The detection unit detects abnormal behavior by, for example, the specific processing unit 290 of the data processing unit 12. The blocking unit blocks abnormal behavior by, for example, the control unit 46A of the robot 414. The notification unit notifies relevant parties by, for example, the specific processing unit 290 of the data processing unit 12. The analysis unit analyzes communication between agents by, for example, the specific processing unit 290 of the data processing unit 12. The visualization unit visualizes the agent's operation log, for example, using the display 40A of the robot 414. The correspondence between each unit and the device or control unit is not limited to the example described above and can be modified in various ways.

[0204] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[0205] Figure 9 shows the emotion map 400, in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[0206] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[0207] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[0208] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, and motorcycles, emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated based, for example, on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[0209] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[0210] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[0211] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing method for the specific process may be used, which includes computer 22 and multiple other computers.

[0212] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[0213] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[0214] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[0215] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[0216] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[0217] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[0218] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[0219] Furthermore, although the above-described examples were divided into four embodiments, some or all of these embodiments may be combined. Also, the smart device 14, smart glasses 214, headset terminal 314, and robot 414 are just examples, and they may be combined, or other devices may be used. Also, although the above-described examples were divided into two embodiments, Embodiment 1 and Embodiment 2, these may be combined.

[0220] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and other things that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[0221] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[0222] (Note 1) A data collection unit that collects behavioral data, A learning unit analyzes the data collected by the aforementioned collection unit and learns normal behavioral patterns, A monitoring unit that monitors in real time based on the normal behavior patterns learned by the learning unit, A detection unit that detects the actions monitored by the aforementioned monitoring unit, A blocking unit that prevents abnormal behavior detected by the detection unit, The system includes a notification unit that notifies the relevant parties of the abnormal behavior blocked by the blocking unit. A system characterized by the following features. (Note 2) It includes an analysis unit that analyzes communication between agents. The system described in Appendix 1, characterized by the features described herein. (Note 3) It includes a visualization unit that visualizes and analyzes the agent's operation logs. The system described in Appendix 1, characterized by the features described herein. (Note 4) The aforementioned collection unit is We estimate the user's emotions and adjust the timing of behavioral data collection based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Note 5) The aforementioned collection unit is Analyze the agent's past behavioral data to select the optimal data collection method. The system described in Appendix 1, characterized by the features described herein. (Note 6) The aforementioned collection unit is When collecting behavioral data, filtering is performed based on the agent's current task and environment. The system described in Appendix 1, characterized by the features described herein. (Note 7) The aforementioned collection unit is It estimates the user's emotions and determines the priority of behavioral data to collect based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Note 8) The aforementioned collection unit is When collecting behavioral data, the system prioritizes collecting highly relevant data by considering the agent's geographical location. The system described in Appendix 1, characterized by the features described herein. (Note 9) The aforementioned collection unit is When collecting behavioral data, analyze the agent's social media activity and collect relevant data. The system described in Appendix 1, characterized by the features described herein. (Note 10) The aforementioned learning unit, The system estimates the user's emotions and selects training data based on those estimated emotions. The system described in Appendix 1, characterized by the features described herein. (Note 11) The aforementioned learning unit, During training, the learning algorithm is optimized by referring to the agent's past behavior data. The system described in Appendix 1, characterized by the features described herein. (Note 12) The aforementioned learning unit, During training, different learning methods are applied to each task of the agent. The system described in Appendix 1, characterized by the features described herein. (Note 13) The aforementioned learning unit, It estimates the user's emotions and adjusts the learning frequency based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Note 14) The aforementioned learning unit, During training, the training data is weighted based on when the agent's behavioral data was submitted. The system described in Appendix 1, characterized by the features described herein. (Note 15) The aforementioned learning unit, During training, the agent's relevant literature is referenced to improve the accuracy of the learning process. The system described in Appendix 1, characterized by the features described herein. (Note 16) The aforementioned monitoring unit, We estimate user sentiment and adjust monitoring criteria based on the estimated user sentiment. The system described in Appendix 1, characterized by the features described herein. (Note 17) The aforementioned monitoring unit, During monitoring, consider the interrelationships between agents to improve monitoring accuracy. The system described in Appendix 1, characterized by the features described herein. (Note 18) The aforementioned monitoring unit, During monitoring, the agent's task attribute information is taken into consideration. The system described in Appendix 1, characterized by the features described herein. (Note 19) The aforementioned monitoring unit, It estimates the user's sentiment and adjusts the order in which monitoring results are displayed based on the estimated user sentiment. The system described in Appendix 1, characterized by the features described herein. (Note 20) The aforementioned monitoring unit, During monitoring, the geographical distribution of agents is taken into consideration. The system described in Appendix 1, characterized by the features described herein. (Note 21) The aforementioned monitoring unit, During monitoring, refer to relevant literature related to the agent to improve monitoring accuracy. The system described in Appendix 1, characterized by the features described herein. (Note 22) The detection unit is The system estimates the user's emotions and adjusts the detection criteria for abnormal behavior based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Note 23) The detection unit is When detection occurs, the detection algorithm is optimized by referring to the agent's past abnormal behavior data. The system described in Appendix 1, characterized by the features described herein. (Note 24) The detection unit is When detection occurs, different detection methods are applied to each agent task. The system described in Appendix 1, characterized by the features described herein. (Note 25) The detection unit is It estimates the user's emotions and determines the detection priority based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Note 26) The detection unit is Upon detection, the detection data is weighted based on when the agent's behavioral data was submitted. The system described in Appendix 1, characterized by the features described herein. (Note 27) The detection unit is During detection, the agent references relevant literature to improve detection accuracy. The system described in Appendix 1, characterized by the features described herein. (Note 28) The aforementioned blocking portion is The system estimates the user's emotions and adjusts methods for preventing abnormal behavior based on those estimated emotions. The system described in Appendix 1, characterized by the features described herein. (Note 29) The aforementioned blocking portion is When blocking an attack, the blocking algorithm is optimized by referencing the agent's past abnormal behavior data. The system described in Appendix 1, characterized by the features described herein. (Note 30) The aforementioned blocking portion is When blocking, different blocking methods are applied for each agent task. The system described in Appendix 1, characterized by the features described herein. (Note 31) The aforementioned blocking portion is The system estimates the user's emotions and determines blocking priorities based on those estimated emotions. The system described in Appendix 1, characterized by the features described herein. (Note 32) The aforementioned blocking portion is When blocking an attack, the blocking data is weighted based on when the agent's behavioral data was submitted. The system described in Appendix 1, characterized by the features described herein. (Note 33) The aforementioned blocking portion is During blocking, the accuracy of blocking is improved by referring to the agent's relevant literature. The system described in Appendix 1, characterized by the features described herein. (Note 34) The aforementioned notification unit, It estimates the user's emotions and adjusts the notification method based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Note 35) The aforementioned notification unit, When sending a notification, the notification algorithm is optimized by referring to the agent's past abnormal behavior data. The system described in Appendix 1, characterized by the features described herein. (Note 36) The aforementioned notification unit, When sending notifications, apply different notification methods for each agent task. The system described in Appendix 1, characterized by the features described herein. (Note 37) The aforementioned notification unit, It estimates the user's emotions and prioritizes notifications based on those emotions. The system described in Appendix 1, characterized by the features described herein. (Note 38) The aforementioned notification unit, When sending notifications, the notification data is weighted based on when the agent's behavioral data was submitted. The system described in Appendix 1, characterized by the features described herein. (Note 39) The aforementioned notification unit, When sending notifications, the system references relevant literature from the agent to improve the accuracy of the notifications. The system described in Appendix 1, characterized by the features described herein. (Note 40) The aforementioned analysis unit, It estimates the user's emotions and adjusts the analysis method based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Supplementary Note 41) The analysis unit optimizes the analysis algorithm by referring to the past communication data of the agent during analysis The system according to Supplementary Note 1, characterized in that (Supplementary Note 42) The analysis unit applies different analysis methods for each task of the agent during analysis The system according to Supplementary Note 1, characterized in that (Supplementary Note 43) The analysis unit estimates the user's emotion and determines the priority of analysis based on the estimated user's emotion The system according to Supplementary Note 1, characterized in that (Supplementary Note 44) The analysis unit performs weighting of analysis data based on the submission time of the agent's communication data during analysis The system according to Supplementary Note 1, characterized in that (Supplementary Note 45) The analysis unit improves the accuracy of analysis by referring to the relevant literature of the agent during analysis The system according to Supplementary Note 1, characterized in that (Supplementary Note 46) The visualization unit estimates the user's emotion and adjusts the visualization method based on the estimated user's emotion The system according to Supplementary Note 1, characterized in that (Supplementary Note 47) The visualization unit optimizes the visualization algorithm by referring to the past operation log of the agent during visualization The system according to Supplementary Note 1, characterized in that (Supplementary Note 48) The visualization unit applies different visualization methods for each task of the agent during visualization The system according to Supplementary Note 1, characterized in that (Supplementary Note 49) The aforementioned visualization unit, It estimates the user's emotions and determines the visualization priority based on the estimated user emotions. The system described in Appendix 1, characterized by the features described herein. (Note 50) The aforementioned visualization unit, During visualization, the visualization data is weighted based on when the agent's operation logs were submitted. The system described in Appendix 1, characterized by the features described herein. (Note 51) The aforementioned visualization unit, During visualization, the accuracy of the visualization is improved by referring to the agent's relevant literature. The system described in Appendix 1, characterized by the features described herein. [Explanation of Symbols]

[0223] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots

Claims

1. A data collection unit that collects behavioral data, A learning unit analyzes the data collected by the aforementioned collection unit and learns normal behavioral patterns, A monitoring unit that monitors in real time based on the normal behavior patterns learned by the learning unit, A detection unit that detects the actions monitored by the aforementioned monitoring unit, A blocking unit that prevents abnormal behavior detected by the detection unit, The system includes a notification unit that notifies the relevant parties of the abnormal behavior blocked by the blocking unit. A system characterized by the following features.

2. It includes an analysis unit that analyzes communication between agents. The system according to feature 1.

3. It includes a visualization unit that visualizes and analyzes the agent's operation logs. The system according to feature 1.

4. The aforementioned collection unit is We estimate the user's emotions and adjust the timing of behavioral data collection based on the estimated user emotions. The system according to feature 1.

5. The aforementioned collection unit is Analyze the agent's past behavioral data to select the optimal data collection method. The system according to feature 1.

6. The aforementioned collection unit is When collecting behavioral data, filtering is performed based on the agent's current task and environment. The system according to feature 1.

7. The aforementioned collection unit is It estimates the user's emotions and determines the priority of behavioral data to collect based on the estimated user emotions. The system according to feature 1.

8. The aforementioned collection unit is When collecting behavioral data, the system prioritizes collecting highly relevant data by considering the agent's geographical location. The system according to feature 1.

9. The aforementioned collection unit is When collecting behavioral data, analyze the agent's social media activity and collect relevant data. The system according to feature 1.

Images