A system for modeling human attention mechanisms in social scenarios

Abstract

The application discloses a human attention mechanism modeling system in a social scene, comprising: a wearable device based on an interaction design established by an edge interaction theory, a user internal information processing module and a user external environment element module connected with the wearable device, an information processing module connected with the user internal information processing module and the user external environment element module, a global attention module connected with the information processing module, and a social scene / wearing scene module connected with the information processing module, wherein the social scene / wearing scene module is connected with the wearable device; the information processing module is provided with an intention layer, a perception layer, a behavior layer and an environment layer; the user internal information processing module is used for processing information of a user behavior target, user cognition and use habits; and the user external environment element module is used for providing real scene information. According to the application, information acquisition efficiency is improved, and the man-machine relationship between the device and the user is improved.

Description

Technical Field

[0001] This invention relates to the technical field of service robots, and in particular to a modeling system for human attention mechanisms in social scenarios. Background Technology

[0002] In recent years, with the rapid development of artificial intelligence technology, the application value of robots has become increasingly significant. Using service robots in social scenarios offers advantages such as low labor costs, safety, and flexibility, mitigating the drawbacks of traditional social scenarios, which require substantial human resources, involve high labor intensity, and suffer from unpredictability. When service robots work in social settings, they need to acquire environmental information through a visual system, process it using relevant algorithms in a computing system, and then output corresponding action strategies. Therefore, an effective deep reinforcement learning network is crucial.

[0003] The human attention mechanism refers to the human visual system rapidly scanning the entire image to identify the target area requiring focused attention—the so-called focus of attention. More attentional resources are then allocated to this area to acquire more detailed information about the target, while suppressing other irrelevant information. Edge interaction theory was proposed in the context of the Internet of Things (IoT) era, characterized by multiple devices serving the same user. The competition among these devices for the user's scarce attention resources causes interference, further impacting the user's information acquisition efficiency. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a modeling system for human attention mechanisms in social scenarios, thereby improving information acquisition efficiency and enhancing the human-computer interaction between devices and users. To achieve the above-mentioned objectives and other advantages of the present invention, a modeling system for human attention mechanisms in social scenarios is provided, comprising:

[0005] The wearable device with an interactive design based on edge interaction theory, the user's internal information processing module and the user's external environment element module connected to the wearable device, the information processing module connected to both the user's internal information processing module and the user's external environment element module, and the association perception global attention module connected to the information processing module, the information processing module is connected to a social scene / wearable scene module, and the social scene / wearable scene module is connected to the wearable device.

[0006] The information processing module is configured with an intention layer, a perception layer, a behavior layer, and an environment layer.

[0007] The user internal information processing module is used to process information about user behavior goals, user cognition, and usage habits; the user external environment element module is used to provide information about real-world scenarios.

[0008] Preferably, the intent layer is used to satisfy user behavior goals, focusing on user intent and exploring user goals across devices.

[0009] Preferably, the perception layer is used to match the clarity and importance of information to the user's cognition and to establish differentiated user cognition. In the perception layer, the user receives multi-channel information from the interactive device through their own sensory system.

[0010] Preferably, the behavior layer is used to conform to the user's instinctive behavior and utilize and cultivate the user's experiential behavior. The interaction design of the interactive device is based on the behavior layer. The behavior layer includes user usage habits and behavioral habits. User usage habits are derived from the user's common biological or genetic characteristics, while behavioral habits are based on the user's acquired habits.

[0011] Preferably, the environment layer is used to make real-time judgments on the specific user interaction scenarios, and at the same time determine high-frequency scenarios in combination with user goals. The environment layer is the overall environmental background and space in which the user interacts with the interactive device. Moreover, the environmental background and space in which the interaction occurs are not only related to the user's behavioral goals, but also because wearable devices have complex interactive features due to the scenarios.

[0012] Preferably, the association-aware global attention module includes the following steps:

[0013] S1. The intermediate layer features of the neural network are decomposed into several feature nodes;

[0014] S2. Calculate the correlation between the given feature node and other feature nodes in the global scope.

[0015] S3. Use the correlation calculation results to infer the importance of each feature node;

[0016] S4. Effectively capture the inherent correlation of features within the global scope, thereby effectively utilizing the global structural information of features.

[0017] Compared with the prior art, the beneficial effects of this invention are:

[0018] (1) This invention introduces edge interaction theory into the interaction design of wearable devices and constructs an interaction design model for wearable devices under edge interaction. The introduction of edge interaction theory into the interaction scenario of wearable devices helps users acquire information by utilizing edge attention, improves information acquisition efficiency, and improves the human-computer relationship between devices and users.

[0019] (2) The inventors explored the relationship between elements such as behavior, scene, habit, and cognition during the interaction with wearable devices, and proposed an interaction design strategy for wearable devices from four different levels of edge interaction.

[0020] (3) In this invention, the limitations of two attention mechanisms based on human attention mechanisms, local attention and global attention, are utilized to propose an association perception global attention mechanism. Attached Figure Description

[0021] Figure 1 A flowchart illustrating a system for modeling human attention mechanisms in social scenarios according to the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Reference Figure 1 A modeling system for human attention mechanisms in social scenarios includes: a wearable device with an interaction design based on edge interaction theory; a user internal information processing module and a user external environment element module connected to the wearable device; an information processing module connected to both the user internal information processing module and the user external environment element module; and an association perception global attention module connected to the information processing module. The information processing module is connected to a social scenario / wearable scenario module, and the social scenario / wearable scenario module is connected to the wearable device.

[0024] The analysis of the impact of information reached by wearable devices on user attention specifically includes:

[0025] Human vision, hearing, and touch are the three main sensory channels for perceiving high-definition and low-definition information. High-definition information usually requires the user's central attention to be effectively acquired, while low-definition information can generally be acquired efficiently using the user's peripheral attention. The clarity of information needs to be matched with the importance of the information. The user's familiarity with the information is also important, as users' attention is naturally more sensitive to familiar things. The delivery rate of information is also crucial, as users may encounter situations where they cannot accurately input certain types of information at once.

[0026] The information processing module is configured with an intention layer, a perception layer, a behavior layer, and an environment layer.

[0027] The user internal information processing module is used to process information about user behavior goals, user cognition, and usage habits; the user external environment element module is used to provide information about real-world scenarios. These four key influencing factors—user behavior goals, user cognition, usage habits, and real-world scenarios—combine with the wearable device interaction mode to jointly constitute the attention mechanism design model in the interaction scenario.

[0028] Modeling four key influencing factors—user behavior goals, user cognition, usage habits, and real-world scenarios—involves the following steps:

[0029] S201. The intent determination node is where the machine performs further analysis and processing on the information directly input by the user and actively acquired by the device during the input process.

[0030] S202. Pay attention to the efficiency of users in acquiring information and the corresponding consumption of attention resources. At the same time, when using multi-channel output, it is necessary to avoid interfering with the user's central attention;

[0031] S203. Attention should be paid to the importance of the information and the clarity required for accurate and efficient communication of the information.

[0032] S204. In real-world scenarios, user input is characterized by multiple channels. For different types of information input methods, it is necessary to consider the user's real-world scenarios and usage habits.

[0033] A user's behavioral objective is a key factor influencing the use of their attention resources during information processing. This factor affects the user's attention resource usage by matching the accuracy of the device's judgment of the user's intent during the information processing process.

[0034] Furthermore, the intent layer is used to satisfy user behavior goals, focusing on user intent and exploring user goals across devices.

[0035] Furthermore, the perception layer is used to match the clarity and importance of information to the user's cognition and establish differentiated user cognition. In the perception layer, the user receives multi-channel information from the interactive device through their own sensory system. Research on edge interaction theory shows that the clarity of information in each sensory channel will affect the user's attention resources and cognitive resources, both of which are also affected by the user's personalized life experience.

[0036] Furthermore, the behavior layer is used to adapt to users' instinctive behaviors and utilize and cultivate users' experiential behaviors. The interaction design of interactive devices is based on the behavior layer. The behavior layer includes user usage habits and behavioral habits. User usage habits are derived from users' common biological or genetic traits, while behavioral habits are based on users' acquired habits.

[0037] Furthermore, the environment layer is used to make real-time judgments on the specific user interaction scenarios, and at the same time determine high-frequency scenarios in combination with user goals. The environment layer is the overall environmental background and space in which the user interacts with the interactive device. Moreover, the environmental background and space in which the interaction occurs are not only related to the user's behavioral goals, but also, because wearable devices have complex interactive characteristics, they will affect the completion of the user's interaction behavior and the usability and ease of use of the device, and further affect the occupation of the user's attention resources.

[0038] Furthermore, the association-aware global attention module includes the following steps:

[0039] S1. The intermediate layer features of the neural network are decomposed into several feature nodes;

[0040] S2. Calculate the correlation between the given feature node and other feature nodes in the global scope.

[0041] S3. Use the correlation calculation results to infer the importance of each feature node;

[0042] S4. Effectively capture the inherent correlation of features within the global scope, thereby effectively utilizing the global structural information of features.

[0043] To address the limitations of both local and global attention mechanisms, an association-aware global attention mechanism is proposed. For a given feature vector X... i Calculate X i The correlation between the vectors X1…X5 and the correlation results are combined into a correlation feature vector r. i In the global attention mechanism of association perception, the importance of each feature is jointly inferred as its attention weight by combining the feature's own information and its corresponding related features. The association vector modeled in this way contains global information and explicitly captures the internal associations of features, integrating the advantages of the self-attention mechanism into the selective attention mechanism, and using a lightweight network to infer the attention weights.

[0044] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention, and applications, modifications and variations thereof will be obvious to those skilled in the art.

[0045] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A modeling system for human attention mechanisms in social scenarios, characterized in that, Includes the following steps: The wearable device with an interactive design based on the edge interaction theory, the user's internal information processing module and the user's external environment element module connected to the wearable device, the information processing module connected to both the user's internal information processing module and the user's external environment element module, and the association perception global attention module connected to the information processing module are all connected to the information processing module. The information processing module is connected to a social scene / wearable scene module, and the social scene / wearable scene module is connected to the wearable device. The information processing module is configured with an intention layer, a perception layer, a behavior layer, and an environment layer. The intent layer is used to satisfy user behavior goals, focusing on user intent and exploring user goals across devices; The perception layer is used to match the clarity and importance of information with the user's cognition and to establish differentiated user cognition. In the perception layer, the user receives multi-channel information from the interactive device through their own sensory system. The behavior layer is used to adapt to users' instinctive behaviors and utilize and cultivate users' experiential behaviors. The interaction design of interactive devices is based on the behavior layer. The behavior layer includes user habits and behavioral habits. User habits are derived from the common biological or genetic characteristics of users, while behavioral habits are based on the user's acquired habits. The environment layer is used to make real-time judgments on the specific user interaction scenarios, and at the same time determine high-frequency scenarios in combination with user goals. The environment layer is the overall environmental background and space in which the user interacts with the interactive device. Moreover, the environmental background and space in which the interaction behavior occurs are not only related to the user's behavioral goals, but also because wearable devices have complex interaction characteristics. The user internal information processing module is used to process information about user behavior goals, user cognition, and usage habits; the user external environment element module is used to provide information about real-world scenarios.

2. The human attention mechanism modeling system in a social scenario as described in claim 1, characterized in that, The associative global attention module includes the following steps: S1. The intermediate layer features of the neural network are decomposed into several feature nodes; S2. Calculate the correlation between the given feature node and other feature nodes in the global scope. S3. Use the correlation calculation results to infer the importance of each feature node; S4. Effectively capture the inherent correlation of features within the global scope, thereby effectively utilizing the global structural information of features.

Images