Multi-scene self-adaptive man-machine interaction system and method based on emotion recognition

Abstract

The invention discloses a multi-scene self-adaptive man-machine interaction system and method based on emotion recognition, relates to the technical field of man-machine interaction, and solves the technical problems of realizing fusion perception of multi-modal emotion features and improving the accuracy of emotion judgment in complex scenes. According to the method, facial, voice and text emotion features are extracted by adopting a multi-modal fusion technology, the limitation of single-modal recognition is solved, an emotion-scene association rule base and a user portrait are constructed, real-time scene classification is combined, accurate mapping of emotions, scenes and demands is realized, one-step interaction of strategies is avoided, and the user experience is improved. Language interaction adaptation is designed from the form, content and style three-dimensional degree, it is ensured that languages are natural and fit scenes, functional response adaptation improves efficiency through priority ranking and execution mode optimization, environment linkage adaptation is combined with user emotion dynamic adjustment directions, collaborative linkage of languages, functions and environments is achieved, and strategy splitting is avoided.

Description

Technical Field

[0001] This invention relates to the field of human-computer interaction technology, specifically to a multi-scenario adaptive human-computer interaction system and method based on emotion recognition. Background Technology

[0002] With the development of artificial intelligence technology, human-computer interaction has evolved from the traditional command-response model to a scenario-based and personalized approach.

[0003] According to patent application CN110570847A, a human-computer interaction system and method for multi-user scenarios are disclosed. The multi-user human-computer interaction system includes: a voice input module for inputting user speech content; an identity recognition module for recognizing user identity and obtaining identity recognition features; a cloud server for storing user identity recognition features and recording user dialogue content in chronological order; a semantic annotation and analysis module for annotating the stored identity recognition features and performing semantic analysis on the recorded dialogue content; and a dialogue feedback module for recognizing speaker identities when multiple speakers interact with the human-computer interaction system, analyzing historical voice data based on the recorded dialogue content, and quickly retrieving matching answers in the context.

[0004] However, while existing human-computer interaction systems possess certain scene recognition and basic emotion recognition capabilities, they still have the following limitations:

[0005] It relies heavily on single-modal emotion recognition, resulting in insufficient accuracy in judging complex emotions; interaction strategies are mostly preset fixed patterns, unable to be dynamically adjusted according to real-time scene characteristics; the adaptation logic of language interaction, functional response, and environmental linkage are independent of each other and lack coordination; the handling of user feedback is mostly a one-time adjustment, without forming a long-term optimization mechanism of "feedback-analysis-iteration", making it difficult to continuously adapt to changes in user habits. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a multi-scenario adaptive human-computer interaction system and method based on emotion recognition, which solves the problem of achieving fusion perception of multimodal emotion features and improving the accuracy of emotion judgment in complex scenarios.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a multi-scenario adaptive human-computer interaction system based on emotion recognition, comprising:

[0008] The interactive information analysis module is used to extract user emotional features and interactive scene features corresponding to the interactive voice and interactive video transmitted by the interactive information acquisition module, and to establish user profiles and emotion-scene association rule base. At the same time, it obtains association results and interaction results, and then transmits them to the adaptive interactive analysis module.

[0009] The adaptive interaction analysis module is used to analyze interaction strategies based on association results and interaction results. It filters interaction strategies based on language interaction adaptation, specifically by comprehensively analyzing language form, language content, and language style to obtain preliminary strategies. It analyzes the preliminary strategies based on function response adaptation, specifically by comprehensively analyzing function priority and execution method to obtain pre-selected strategies. It performs environmental linkage adaptation analysis on the obtained pre-selected strategies, combines real-time environmental data to generate standard interaction strategies, and transmits them to the comprehensive feedback processing module.

[0010] The comprehensive feedback processing module is used to analyze the standard interaction strategy based on user feedback, calculate the negative proportion of negative feedback and compare it with the preset proportion, generate an interaction adjustment signal, analyze the interaction adjustment signal, determine the feedback direction based on the content of negative feedback, generate adjustment information, and transmit it to the interaction result output module.

[0011] As a further aspect of the present invention, it also includes an interactive information analysis module and an interactive result output module;

[0012] The interactive information analysis module is used to collect user interaction language through the microphone and interactive video through the camera, and then transmit them to the interactive information analysis module.

[0013] The interactive result output module is used to display the generated adjustment information to the corresponding management personnel.

[0014] As a further aspect of the present invention, the user emotional features include facial expressions, voice emotions, and text emotions. Facial expressions are analyzed by computer vision technology to analyze facial muscle movements. Voice emotions are extracted by extracting acoustic features from the voice signal, including tone, speech rate, volume, and spectral changes. Text emotions are identified by recognizing interactive voice and analyzing the emotional tendencies of words and the semantic context using NLP technology.

[0015] Interactive scenarios include home, office, medical, and educational scenarios.

[0016] As a further aspect of the present invention, the specific method by which the interactive information analysis module obtains the association results and interaction results is as follows:

[0017] The system acquires user emotional characteristics and interaction scenario characteristics, mines high-frequency co-occurrence patterns from historical data through association rule algorithms, supplements the data with domain experts, and trains the system using a regression model. Taking scenarios and emotions as input, the system predicts user needs, transforms the extracted association rules into a structured format, and displays them in a condition-conclusion logical form.

[0018] Next, the association results are determined based on the emotion-scene association rule library, and specific interactive behaviors are generated. The interactive behaviors include language, function and environmental linkage. At the same time, real-time interactive information is obtained and its corresponding interactive results are obtained.

[0019] As a further aspect of the present invention, the adaptive interaction analysis module obtains the initial selection strategy in the following specific manner:

[0020] Based on the analysis of language form, the interaction modality, sentence structure, and rhythm of speech rate and pauses are determined according to the interaction results, scene characteristics, and related results.

[0021] For language content analysis, combined with real-time user interaction information, necessary content directly related to real-time needs is selected from the interaction results, and the content is determined by combining the correlation results;

[0022] For language style analysis, it is determined comprehensively based on interaction results, association results, and scene characteristics;

[0023] Combining the above three factors, a preliminary selection strategy that aligns with language interaction adaptation is formed.

[0024] As a further aspect of the present invention, the adaptive interactive analysis module obtains the pre-selected strategy by filtering the initial selection strategy in the following specific manner:

[0025] Based on functional priority, obtain interaction results and classify the interaction results into core needs and peripheral needs. The classification of core needs and peripheral needs is determined according to the characteristics of different interaction scenarios. At the same time, respond to the core needs as the first priority task.

[0026] For the execution method, the execution steps of the interaction result are obtained. At the same time, the execution steps are simplified and analyzed in combination with the characteristics of the interaction scenario. Unnecessary steps in the execution steps are omitted, and the corresponding key steps are retained. Then, the functional response adaptation analysis of the initial strategy is performed to filter and obtain the pre-selected strategy.

[0027] As a further aspect of the present invention, the adaptive interaction analysis module generates a standard interaction strategy by combining real-time environmental data in the following specific manner:

[0028] The system acquires interaction results and user emotional characteristics, adjusts the direction of environmental linkage based on user emotions, collects real-time environmental data and performs environmental linkage processing based on the characteristics of the interaction scenario, filters pre-selected strategies, generates standard interaction strategies, and transmits them to the interaction result output module.

[0029] As a further aspect of the present invention, the specific method by which the integrated feedback processing module generates the interactive adjustment signal is as follows:

[0030] Simultaneously, user feedback is acquired, including both explicit and implicit feedback. Historical user data is also obtained, including the number of historical interactions and the number of negative explicit and implicit feedback responses. The corresponding negative percentage is calculated and compared with a preset percentage, the specific value of which is set by the operator. If the negative percentage is greater than the preset percentage, it indicates that interaction adjustment is needed, and an interaction adjustment signal is generated. Conversely, if the negative percentage is less than the preset percentage, it indicates that no interaction adjustment is needed.

[0031] As a further aspect of the present invention, the specific method by which the integrated feedback processing module generates adjustment information is as follows:

[0032] The system acquires the number of explicit and implicit feedback responses, obtains the content of explicit feedback, categorizes the feedback content to obtain explicit category content, acquires the content of implicit feedback, matches the explicit category content with the implicit feedback content to determine the feedback direction, and generates adjustment information based on the feedback direction.

[0033] A multi-scenario adaptive human-computer interaction method based on emotion recognition, which specifically includes the following steps:

[0034] Step 1: Collect real-time user interaction information and extract user emotional characteristics and interaction scenario characteristics. At the same time, establish user profiles and an emotion-scenario association rule base, and determine the association results and interaction results.

[0035] Step 2: Analyze the interaction strategies based on the association results and interaction results, and filter the interaction strategies based on language interaction adaptation. Specifically, the initial strategy is obtained by comprehensively analyzing language form, language content, and language style.

[0036] Step 3: Analyze the initial selection strategy based on the functional response adaptation, specifically by comprehensively analyzing the functional priority and execution method, and then filter the initial selection strategy to obtain the pre-selected strategy;

[0037] Step 4: Perform environmental linkage adaptation analysis on the obtained pre-selected strategies, and generate standard interaction strategies by combining real-time environmental data.

[0038] Step 5: Analyze the standard interaction strategy based on user feedback, calculate the negative percentage of negative feedback and compare it with the preset percentage, and generate interaction adjustment signals.

[0039] Step 6: Analyze the interactive adjustment signals, determine the feedback direction based on the negative feedback content, and generate adjustment information.

[0040] This invention provides a multi-scenario adaptive human-computer interaction system and method based on emotion recognition. Compared with existing technologies, it has the following advantages:

[0041] This invention employs multimodal fusion technology to extract facial, voice, and text emotional features, overcoming the limitations of single-modal recognition. It constructs an emotion-scene association rule base and user profile, and combines real-time scene classification to achieve accurate mapping of emotions, scenes, and needs, avoiding a one-size-fits-all approach to interaction strategies. Language interaction adaptation is designed from three dimensions: form, content, and style, ensuring that the language is natural and context-appropriate. Functional response adaptation improves efficiency through priority sorting and execution method optimization. Environmental linkage adaptation dynamically adjusts direction based on user emotions, achieving coordinated linkage between language, functions, and environment, and avoiding fragmented strategies. Attached Figure Description

[0042] Figure 1 This is a system block diagram of the present invention;

[0043] Figure 2 This is a diagram illustrating the steps and methods of the present invention. Detailed Implementation

[0044] 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.

[0045] Example 1

[0046] Please see Figure 1 This application provides a multi-scenario adaptive human-computer interaction system based on emotion recognition, including: an interaction information acquisition module, an interaction information analysis module, an adaptive interaction analysis module, a comprehensive feedback processing module, and an interaction result output module, and in conjunction with the appendix... Figure 1 It can be seen that the information between the above functional modules is transmitted in one direction.

[0047] The interactive information acquisition module is used to collect real-time interactive information from users and transmit it to the interactive information analysis module. The real-time interactive information includes interactive voice and interactive video. The interactive voice is acquired through a microphone, while the interactive video is acquired through a camera.

[0048] The interactive information analysis module is used to identify and process the acquired real-time interactive information, extracting user emotional features from the real-time interactive information. These user emotional features include facial expressions, voice emotion, and text emotion. Specifically, for facial expressions, computer vision technology is used to analyze facial muscle movements to determine emotion, such as deep learning models like CNN and Transformer. Voice emotion is determined by extracting acoustic features from the speech signal, including tone, speech rate, volume, and spectral changes, combined with natural language processing (NLP) to analyze semantics and determine emotion. Text emotion is determined by recognizing interactive speech and using NLP technology to analyze word sentiment and contextual semantics. Simultaneously, the module acquires features of the current interactive scenario and classifies the interactive scenario based on these features. Specific interactive scenario classifications include home scenarios, office scenarios, medical scenarios, and educational scenarios, etc. Next, the module acquires the user's historical data and constructs corresponding user profiles and an emotion-scenario association rule base based on this historical data. The construction method of the emotion-scenario association rule base is as follows:

[0049] The algorithm acquires user emotional characteristics and interaction scenario characteristics, mines high-frequency co-occurrence patterns from historical data through association rule algorithms, supplements the data with domain experts, and trains the model using regression models. It uses scenarios and emotions as inputs to predict user needs, transforms the extracted association rules into a structured format, and adopts a condition-conclusion logical form. The construction of user profiles is an existing technology and will not be elaborated on here.

[0050] Next, the association results are determined based on the emotion-scene association rule library, and specific interactive behaviors are generated. The interactive behaviors include language, function and environmental linkage. At the same time, real-time interactive information and its corresponding interactive results are obtained. Then, the association results and interactive results are transmitted to the adaptive interaction analysis module.

[0051] The adaptive interaction analysis module is used to analyze interaction strategies based on the obtained association results and interaction results, specifically from three aspects: language interaction adaptation, function response adaptation, and environment linkage adaptation.

[0052] The language interaction adaptation is analyzed, specifically from the perspectives of language form, language content, and language style. For language form, the presentation mode of the corresponding language carrier is obtained, including interaction modality, sentence structure, and rhythm. The interaction modality specifically includes voice and text, the sentence structure specifically includes long sentences and short sentences, and the structure specifically includes speech rate and pauses. The interaction results and corresponding scene features are obtained, and the language form is determined by combining the corresponding association results. For example, for driving scenarios, a voice + short sentence presentation mode should be used to reduce visual dependence, while a text + long sentence presentation mode can be used for office scenarios to facilitate accurate recall.

[0053] Specifically, the interaction modality is determined by the visual dependence of the scene (e.g., high visual dependence in driving scenes → disable text and prioritize voice) and the user's historical interaction data (e.g., 80% of office scenes use text → default to text modality determination). Modality switching is supported, such as when the speech recognition accuracy is <60% in noisy environments → automatically switch to text.

[0054] Sentence structure: Based on task complexity, for example, simple tasks → short sentences, average length ≤ 5 words; complex tasks → long sentences + paragraphs, each paragraph ≤ 15 words, and determined by user comprehension ability, such as children / elderly → short sentences account for ≥ 80%, for example: navigation scenario use "turn right in 300 meters", office scenario use "We have prepared 3 types of solutions for you, namely A (efficiency first) and B (cost first), do you need more details?".

[0055] Pace control: Adjust speech rate, such as 150 words / minute in normal scenarios and 100 words / minute in elderly scenarios, and pause duration, such as 0.5 seconds between short sentences and 1 second between logical breaks in long sentences. Combine this with emotional intensity, such as speaking slightly faster but with clear pauses when expressing anger, to prevent users from missing key information.

[0056] For language content, the interaction results are obtained and their information content and necessity are analyzed. The necessary content in the interaction results is determined by combining the user's real-time interaction information. Necessary content is content that is directly related to the real-time interaction information. At the same time, the language content of the interaction results is determined by combining the correlation results. For example, when a user is in urgent need of help, the content should focus on the solution, such as "I have contacted the repairman for you, and he will arrive within 10 minutes." When chatting casually, extended topics can be added, such as "Do you want to remind your favorite band that they have a performance next week?"

[0057] Based on the language style, obtain the interaction results and corresponding related results, and comprehensively determine the language style in combination with the characteristics of the interaction scenario. For example, the medical scenario requires professionalism + gentleness, such as postoperative recovery requires patience, and I will remind you to take your medicine every day. The family scenario is more colloquial and intimate, such as "The food is ready, come and eat, or it will get cold."

[0058] The interaction strategy that meets the language interaction adaptation criteria by comprehensively considering language form, language content, and language style is called the initial selection strategy;

[0059] Based on the initial selection strategy, the functional response adaptation is analyzed, specifically from two aspects: functional priority and execution method. For functional priority, the interaction results are obtained and classified into core needs and peripheral needs. The classification of core needs and peripheral needs is determined according to the characteristics of different interaction scenarios. At the same time, the core needs are responded to as the first priority task. For example, in the driving scenario, navigation instructions have higher priority than music recommendations; in the medical scenario, emergency calls have higher priority than information queries.

[0060] For the execution method, the execution steps of the interaction result are obtained. At the same time, the execution steps are simplified and analyzed in combination with the characteristics of the interaction scenario. Unnecessary steps are omitted and the corresponding key steps are retained.

[0061] Next, a functional response adaptation analysis was performed on the initial selection strategy to obtain the pre-selected strategy;

[0062] Next, based on the pre-selected strategy, environmental linkage adaptation analysis is performed to obtain interaction results and user emotional characteristics. The direction of environmental linkage is adjusted in combination with user emotions. At the same time, real-time environmental data is collected and environmental linkage is processed according to its interaction scenario characteristics. The pre-selected strategy is filtered to generate a standard interaction strategy, which is then transmitted to the interaction result output module.

[0063] For example, negative emotions (anger / anxiety) can be addressed by reducing distractions, such as reducing noise or minimizing changes in lighting.

[0064] Positive emotions (pleasure / excitement), linked to a strong atmosphere (such as synchronized light colors and music rhythm);

[0065] Neutral emotion (focused / calm), with a stable orientation, such as maintaining the current temperature and lighting.

[0066] The interaction result output module is used to output the interaction results according to the acquired standard interaction strategy.

[0067] Example 2

[0068] As a second embodiment of the present invention, it is implemented based on the first embodiment, and the difference from the first embodiment is as follows:

[0069] The adaptive interaction analysis module transmits the generated standard interaction strategy to the comprehensive feedback processing module, while simultaneously acquiring user feedback, including explicit and implicit feedback. Explicit feedback specifically refers to users clearly marking negative evaluations such as "dissatisfied" or "vague," or text feedback containing negative words. Implicit negative feedback refers to behaviors judged as negative by the behavior model. The module also acquires the user's historical data, including the corresponding number of historical interactions, and the number of explicit and implicit negative feedbacks. Here, negative feedback indicates dissatisfaction with the current interaction result. The module calculates the corresponding negative percentage and compares it with a preset percentage, the specific value of which is set by the operator. If the negative percentage is greater than the preset percentage, it indicates that interaction adjustment is needed, and an interaction adjustment signal is generated. Conversely, if the negative percentage is less than the preset percentage, it indicates that no interaction adjustment is needed.

[0070] Next, the generated interactive adjustment signals are analyzed to obtain the number of explicit and implicit feedbacks. At the same time, the feedback content corresponding to the explicit feedback is obtained and classified to obtain explicit category content. Then, the feedback content corresponding to the implicit feedback is obtained, and the explicit category content is matched with the implicit feedback content to determine the feedback direction. Based on the feedback direction, adjustment information is generated and transmitted to the interactive result output module.

[0071] For example, language interaction adjustments: such as language content → reduce extended information, core content accounts for ≥80%; language form → sentence length ≤10 characters;

[0072] Function response adjustments: such as function priority → in office scenarios, document saving takes precedence over message notifications; execution method → ​​merging confirmation and submission steps into a one-click operation;

[0073] Environmental linkage adjustment: such as linkage intensity → in the home leisure scene, environmental linkage only triggers the core device (lighting) and turns off the auxiliary device (fragrance machine); physical parameters → the light brightness is reduced from 500 lux to 300 lux.

[0074] The interactive result output module is used to display the generated adjustment information to the corresponding management personnel.

[0075] Example 3

[0076] As a third embodiment of the present invention, the focus is on combining the implementation processes of the first and second embodiments.

[0077] Example 4

[0078] Please see Figure 2 This application provides a multi-scenario adaptive human-computer interaction method based on emotion recognition, which specifically includes the following steps:

[0079] Step 1: Collect real-time user interaction information and extract user emotional characteristics and interaction scenario characteristics. At the same time, establish user profiles and emotion-scenario association rule base, and determine the association results and interaction results. The specific processing method is the same as the processing process of the interaction information analysis module.

[0080] Step 2: Analyze the interaction strategies based on the association results and interaction results, and filter the interaction strategies based on language interaction adaptation. Specifically, the initial strategy is selected by comprehensively analyzing language form, language content, and language style. The specific processing method is the same as that of the adaptive interaction analysis module.

[0081] Step 3: Analyze the initial selection strategy based on the functional response adaptation. Specifically, conduct a comprehensive analysis from the perspectives of functional priority and execution method, and filter the initial selection strategy to obtain the pre-selected strategy. The specific processing method is the same as that of the adaptive interaction analysis module.

[0082] Step 4: Perform environmental linkage adaptation analysis on the obtained pre-selected strategies, and generate standard interaction strategies by combining real-time environmental data. The specific processing method is the same as that of the adaptive interaction analysis module.

[0083] Step 5: Analyze the standard interaction strategy based on user feedback, calculate the negative percentage of negative feedback and compare it with the preset percentage, generate an interaction adjustment signal, and the specific processing method is the same as that of the comprehensive feedback processing module.

[0084] Step 6: Analyze the interactive adjustment signals, determine the feedback direction based on the negative feedback content, and generate adjustment information. The specific processing method is the same as that of the comprehensive feedback processing module.

[0085] The data in the above formulas are all calculated using numerical values, without substituting the units of the parameters. In addition, the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0086] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A multi-scenario adaptive human-computer interaction system based on emotion recognition, characterized in that, include: The interactive information analysis module is used to extract user emotional features and interactive scene features corresponding to the interactive voice and interactive video transmitted by the interactive information acquisition module, and to establish user profiles and emotion-scene association rule base. At the same time, it obtains association results and interaction results, and then transmits them to the adaptive interactive analysis module. The adaptive interaction analysis module is used to analyze interaction strategies based on association results and interaction results. It filters interaction strategies based on language interaction adaptation, specifically by comprehensively analyzing language form, language content, and language style to obtain preliminary strategies. It analyzes the preliminary strategies based on function response adaptation, specifically by comprehensively analyzing function priority and execution method to obtain pre-selected strategies. It performs environmental linkage adaptation analysis on the obtained pre-selected strategies, combines real-time environmental data to generate standard interaction strategies, and transmits them to the comprehensive feedback processing module. The comprehensive feedback processing module is used to analyze the standard interaction strategy based on user feedback, calculate the negative proportion of negative feedback and compare it with the preset proportion, generate an interaction adjustment signal, analyze the interaction adjustment signal, determine the feedback direction based on the content of negative feedback, generate adjustment information, and transmit it to the interaction result output module.

2. The multi-scenario adaptive human-computer interaction system based on emotion recognition according to claim 1, characterized in that, It also includes an interactive information analysis module and an interactive result output module; The interactive information analysis module is used to collect user interaction language through the microphone and interactive video through the camera, and then transmit them to the interactive information analysis module. The interactive result output module is used to display the generated adjustment information to the corresponding management personnel.

3. The multi-scenario adaptive human-computer interaction system based on emotion recognition according to claim 1, characterized in that, The user emotional features include facial expressions, voice emotions, and text emotions. Facial expressions are analyzed by computer vision technology to analyze facial muscle movements. Voice emotions are extracted by extracting acoustic features from the voice signal, including tone, speech rate, volume, and spectral changes. Text emotions are identified by recognizing interactive voice and analyzing the emotional tendencies of words and the semantic context using NLP technology. Interactive scenarios include home, office, medical, and educational scenarios.

4. The multi-scenario adaptive human-computer interaction system based on emotion recognition according to claim 1, characterized in that, The specific method by which the interactive information analysis module obtains the association results and interaction results is as follows: The system acquires user emotional characteristics and interaction scenario characteristics, mines high-frequency co-occurrence patterns from historical data through association rule algorithms, supplements the data with domain experts, and trains the system using a regression model. Taking scenarios and emotions as input, the system predicts user needs, transforms the extracted association rules into a structured format, and displays them in a condition-conclusion logical form. Next, the association results are determined based on the emotion-scene association rule library, and specific interactive behaviors are generated. The interactive behaviors include language, function and environmental linkage. At the same time, real-time interactive information is obtained and its corresponding interactive results are obtained.

5. The multi-scenario adaptive human-computer interaction system based on emotion recognition according to claim 1, characterized in that, The specific method by which the adaptive interactive analysis module obtains the initial selection strategy is as follows: Based on the analysis of language form, the interaction modality, sentence structure, and rhythm of speech rate and pauses are determined according to the interaction results, scene characteristics, and related results. For language content analysis, combined with real-time user interaction information, necessary content directly related to real-time needs is selected from the interaction results, and the content is determined by combining the correlation results; For language style analysis, it is determined comprehensively based on interaction results, association results, and scene characteristics; Combining the above three factors, a preliminary selection strategy that aligns with language interaction adaptation is formed.

6. The multi-scenario adaptive human-computer interaction system based on emotion recognition according to claim 1, characterized in that, The adaptive interactive analysis module obtains the pre-selected strategy by filtering the initial selection strategy in the following way: Based on functional priority, obtain interaction results and classify the interaction results into core needs and peripheral needs. The classification of core needs and peripheral needs is determined according to the characteristics of different interaction scenarios. At the same time, respond to the core needs as the first priority task. For the execution method, the execution steps of the interaction result are obtained. At the same time, the execution steps are simplified and analyzed in combination with the characteristics of the interaction scenario. Unnecessary steps in the execution steps are omitted, and the corresponding key steps are retained. Then, the functional response adaptation analysis of the initial strategy is performed to filter and obtain the pre-selected strategy.

7. The multi-scenario adaptive human-computer interaction system based on emotion recognition according to claim 1, characterized in that, The adaptive interaction analysis module generates standard interaction strategies by combining real-time environmental data in the following specific way: The system acquires interaction results and user emotional characteristics, adjusts the direction of environmental linkage based on user emotions, collects real-time environmental data and performs environmental linkage processing based on the characteristics of the interaction scenario, filters pre-selected strategies, generates standard interaction strategies, and transmits them to the interaction result output module.

8. The multi-scenario adaptive human-computer interaction system based on emotion recognition according to claim 1, characterized in that, The specific method by which the integrated feedback processing module generates the interactive adjustment signal is as follows: Simultaneously, user feedback is acquired, including both explicit and implicit feedback. Historical user data is also obtained, including the number of historical interactions and the number of negative explicit and implicit feedback responses. The corresponding negative percentage is calculated and compared with a preset percentage, the specific value of which is set by the operator. If the negative percentage is greater than the preset percentage, it indicates that interaction adjustment is needed, and an interaction adjustment signal is generated. Conversely, if the negative percentage is less than the preset percentage, it indicates that no interaction adjustment is needed.

9. The multi-scenario adaptive human-computer interaction system based on emotion recognition according to claim 1, characterized in that, The specific method by which the integrated feedback processing module generates adjustment information is as follows: The system acquires the number of explicit and implicit feedback responses, obtains the content of explicit feedback, categorizes the feedback content to obtain explicit category content, acquires the content of implicit feedback, matches the explicit category content with the implicit feedback content to determine the feedback direction, and generates adjustment information based on the feedback direction.

10. A multi-scene adaptive human-computer interaction method based on emotion recognition, used to execute the multi-scene adaptive human-computer interaction system according to any one of claims 1-9, characterized in that, The method specifically includes the following steps: Step 1: Collect real-time user interaction information and extract user emotional characteristics and interaction scenario characteristics. At the same time, establish user profiles and an emotion-scenario association rule base, and determine the association results and interaction results. Step 2: Analyze the interaction strategies based on the association results and interaction results, and filter the interaction strategies based on language interaction adaptation. Specifically, the initial strategy is obtained by comprehensively analyzing language form, language content, and language style. Step 3: Analyze the initial selection strategy based on the functional response adaptation, specifically by comprehensively analyzing the functional priority and execution method, and then filter the initial selection strategy to obtain the pre-selected strategy; Step 4: Perform environmental linkage adaptation analysis on the obtained pre-selected strategies, and generate standard interaction strategies by combining real-time environmental data. Step 5: Analyze the standard interaction strategy based on user feedback, calculate the negative percentage of negative feedback and compare it with the preset percentage, and generate interaction adjustment signals. Step 6: Analyze the interactive adjustment signals, determine the feedback direction based on the negative feedback content, and generate adjustment information.

Images