A scene-based voice intelligent recommendation method, system and terminal

Abstract

The application relates to the technical field of intelligent recommendation, and discloses a scene-based voice intelligent recommendation method, system and terminal, wherein corresponding comprehensive data in a scene is acquired according to the scene classification of a specified user use scene; the comprehensive data in the scene is input into an intelligent interaction module to acquire corresponding output results; an action strategy for the specified user is selected according to the output results; the scene classification comprises an active question and answer communication scene and a passive voice recommendation scene; the active question and answer communication scene: an interactive dialogue with the specified user is generated, and the consultation question needs to be replied and guided; the passive voice recommendation scene: corresponding interactive actions are automatically initiated according to the specified demand of the specified user, which not only can improve user satisfaction, but also improves service efficiency through optimization of an interactive process, and is particularly suitable for application in fields such as intelligent customer service, voice assistants and the like which need to process complex interactive demands, and provides a better solution for constructing a more intelligent and more humanized voice interaction.

Description

Technical Field

[0001] This application relates to the technical field of intelligent recommendation, and in particular to a scenario-based voice intelligent recommendation method, system, and terminal. Background Technology

[0002] With the rapid development of artificial intelligence technology, intelligent voice recommendation systems have been widely applied in e-commerce, healthcare, education, finance, and other fields, becoming a key technology for improving user experience. The construction and use of intelligent voice recommendation systems require prior knowledge reserves in the corresponding domain's knowledge base and its continuous updating. However, traditional knowledge base systems still have significant shortcomings when facing complex and ever-changing user needs. In the initial construction phase, relying on manually compiled knowledge bases is not only costly but also difficult to comprehensively cover all possible variations of problems in various fields. More importantly, when user needs or business scenarios change, static knowledge bases are difficult to update in a timely manner, leading to a rapid decline in system responsiveness and severely impacting service quality and user experience.

[0003] This problem is particularly pronounced in voice interaction scenarios. Different application scenarios present differentiated requirements for intelligent voice recommendations: in marketing, it's necessary to accurately understand users' colloquial descriptions and link them to historical purchase or consultation data for personalized recommendations or answers; in medical scenarios, it requires accurate recognition of professional terminology and reliable treatment suggestions based on patient symptoms; and in educational applications, the system needs to adapt to the expression styles of learners of different ages and provide targeted knowledge explanations. Furthermore, currently, intelligent voice recommendations cannot proactively remind users when they haven't initiated communication, such as in scenarios requiring timely reminders. At present, intelligent voice recommendation systems often use general models to handle various scenarios, lacking deep adaptation to domain characteristics, resulting in inconsistent recommendation performance. In voice interaction scenarios, the system also needs to address additional complex factors such as speech recognition errors, dialect differences, and colloquial expressions. Summary of the Invention

[0004] In order to provide a method, system and terminal for intelligent voice recommendation that can adapt to the needs of different scenarios, this application provides a scenario-based intelligent voice recommendation method, system and terminal.

[0005] Firstly, this application provides a scenario-based voice intelligent recommendation method, system, and terminal, employing the following technical solution:

[0006] A scenario-based intelligent voice recommendation method includes:

[0007] Step 1: Obtain comprehensive data within the corresponding scenario based on the scenario classification of the specified user usage scenario;

[0008] Step 2: Input the comprehensive data within the scenario into the intelligent interaction module and obtain the corresponding output results;

[0009] Step 3: Select the action strategy for the specified user based on the output results;

[0010] The scenario classification includes proactive question-and-answer communication scenarios and passive voice recommendation scenarios;

[0011] The proactive question-and-answer communication scenario refers to: engaging in interactive dialogue with the designated user, requiring responses and guidance regarding inquiries;

[0012] The passive voice recommendation scenario means that, based on the specified needs of the specified user, a corresponding interactive action is automatically initiated.

[0013] By adopting the above technical solutions, and through scenario-based data processing and intelligent analysis, it is possible to adapt to both proactive question-and-answer communication scenarios and passive voice recommendation scenarios. This enables intelligent recommendations for different interaction modes, effectively handling users' immediate inquiry needs while proactively providing valuable recommendation information, forming a complete service loop. This scenario-aware intelligent recommendation mechanism not only significantly improves user satisfaction but also enhances service efficiency by optimizing the interaction process. It provides reliable technical support for the in-depth application of intelligent voice services in various industries, and is particularly suitable for fields such as intelligent customer service and voice assistants that require handling complex interaction needs. It provides an innovative solution for building a more intelligent and human-centered voice interaction system.

[0014] Optionally, for the proactive question-and-answer communication scenario, the comprehensive data within the scenario includes the consultation question data of the specified user, and the intelligent interaction module includes a clustering mining module, a knowledge base, and a response module;

[0015] The clustering mining module is used to automatically mine knowledge points that the specified user is concerned about but has not responded to, and recommend valuable questions to human experts in the form of knowledge points. If the human experts judge that they are valuable, they can be directly added to the knowledge base.

[0016] The reply module is used to reply to the knowledge points that the specified user is concerned about but has a response to, and the reply output content is the output result; the action strategy corresponding to the output result is the associated guided teaching.

[0017] Optionally, the process of automatically mining knowledge points that the specified user is interested in but has not yet addressed includes:

[0018] Collect the original questions from users and obtain a corpus of questions that are not covered;

[0019] NLP techniques are used to semantically cluster questions, identify high-frequency uncovered questions, and discover knowledge blind spots.

[0020] Analyze the clustering results, extract common knowledge points, and form expandable knowledge points;

[0021] High-value questions are submitted for manual review to ensure accuracy and prevent incorrect knowledge from being entered into the database.

[0022] Optionally, the method for collecting the original questions from user inquiries and obtaining a corpus of uncovered questions includes:

[0023] Continuously collect original consultation questions input by users via voice or text;

[0024] The original consultation question is preprocessed; the preprocessing includes removing irrelevant characters, filtering stop words, and restoring word forms.

[0025] The original consultation question is encoded into a high-dimensional semantic vector using a pre-trained language model;

[0026] Calculate the maximum cosine similarity between the high-dimensional semantic vector and all existing knowledge points in the knowledge base to evaluate the coverage of the original consultation question;

[0027] When the maximum cosine similarity is lower than a preset similarity threshold, the original consultation question is automatically collected into a corpus of uncovered questions and labeled with tags to be analyzed.

[0028] Optionally, the method of using NLP technology to perform semantic clustering of questions, identify high-frequency uncovered question points, and discover knowledge blind spots includes:

[0029] The high-dimensional semantic vector integrates contextual features based on the Transformer architecture and optimized keyword features to form a hybrid semantic space with strong representational capabilities.

[0030] During the clustering process, question clusters with high semantic relevance are identified, and core question points are determined through a density peak detection algorithm.

[0031] Calculate multi-dimensional metrics for each generated semantic cluster. When the frequency of a specific question cluster exceeds a preset threshold within a unit of time, it is marked as a high-frequency uncovered question.

[0032] Optionally, the method for analyzing the clustering results, extracting common knowledge points, and forming scalable knowledge points includes:

[0033] Dependency parsing and semantic role labeling are performed on the questions in each semantic cluster to identify the core entities, actions and their relationships in the questions;

[0034] A neural network model based on an attention mechanism is used to extract frequently occurring common semantic patterns and transform these common semantic patterns into structured knowledge triples.

[0035] Optionally, for the passive question-and-answer communication scenario, the comprehensive data within the scenario includes the real-time monitoring data of the specified user, scenario status information, and scenario voice information;

[0036] The working process of the intelligent interaction module includes:

[0037] Based on the comprehensive data within the scenario, a scenario is constructed, and the scenario construction result is output.

[0038] Based on the scenario setup results, a corresponding recommendation strategy is obtained and voice recommendations are performed.

[0039] Optionally, the process of setting up the scene includes:

[0040] Based on real-time monitoring data, the user status of the specified user is analyzed, and a user status analysis index is output.

[0041] Analyze the application status of the scenario based on the scenario status information, and output the scenario device status coefficient;

[0042] Analyze the scene state based on the voice information within the scene and output the scene state coefficient.

[0043] Secondly, this application provides a scenario-based voice intelligent recommendation system, which adopts the following technical solution:

[0044] A scenario-based voice-based intelligent recommendation system includes:

[0045] The data acquisition module is used to obtain comprehensive data within a specified scenario based on the scenario classification of the user's usage scenario.

[0046] The data transmission module is used to input the comprehensive data within the scene into the intelligent interaction module and obtain the corresponding output results;

[0047] The strategy selection module is used to select an action strategy for the specified user based on the output results.

[0048] The scenario classification includes proactive question-and-answer communication scenarios and passive voice recommendation scenarios;

[0049] The proactive question-and-answer communication scenario refers to: engaging in interactive dialogue with the designated user, requiring responses and guidance regarding inquiries;

[0050] The passive voice recommendation scenario means that, based on the specified needs of the specified user, a corresponding interactive action is automatically initiated.

[0051] Thirdly, this application provides a scenario-based voice intelligent recommendation terminal, which adopts the following technical solution:

[0052] A scenario-based voice intelligent recommendation terminal is provided, which is used to execute a scenario-based voice intelligent recommendation method.

[0053] In summary, this application includes at least one of the following beneficial technical effects:

[0054] This application, through scenario-based data processing and intelligent analysis, can adapt to both proactive question-and-answer communication scenarios and passive voice recommendation scenarios, achieving intelligent recommendations for different interaction modes. It can not only properly handle users' immediate consultation needs, but also proactively provide valuable recommendation information, forming a complete service loop. This scenario-aware intelligent recommendation mechanism not only significantly improves user satisfaction, but also improves service efficiency by optimizing the interaction process. It provides reliable technical support for the in-depth application of intelligent voice services in various industries, and is particularly suitable for fields such as intelligent customer service and voice assistants that require handling complex interaction needs. It provides an innovative solution for building a more intelligent and humanized voice interaction system. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating the scene-based intelligent voice recommendation method of the present invention;

[0056] Figure 2 This is a schematic diagram of the modules of the scene-based voice intelligent recommendation system of the present invention. Detailed Implementation

[0057] The embodiments of this application are described in detail below, and examples of the embodiments are shown in the accompanying drawings.

[0058] In the description of this specification, the references to "certain embodiments," "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples" refer to specific features, structures, materials, or characteristics described in connection with the described embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0059] This application discloses a scenario-based intelligent voice recommendation method, referring to... Figure 1 ,include:

[0060] Step 1: Obtain comprehensive data within the corresponding scenario based on the scenario classification of the specified user usage scenario;

[0061] Step 2: Input the comprehensive data within the scenario into the intelligent interaction module and obtain the corresponding output results;

[0062] Step 3: Select the action strategy for the specified user based on the output results;

[0063] The scenario classification includes proactive question-and-answer communication scenarios and passive voice recommendation scenarios;

[0064] The proactive question-and-answer communication scenario refers to: engaging in interactive dialogue with the designated user, requiring responses and guidance regarding inquiries;

[0065] The passive voice recommendation scenario means that, based on the specified needs of the specified user, a corresponding interactive action is automatically initiated.

[0066] This application, through scenario-based data processing and intelligent analysis, can adapt to both proactive question-and-answer communication scenarios and passive voice recommendation scenarios, achieving intelligent recommendations for different interaction modes. It can properly handle users' immediate consultation needs (such as customer inquiries like "Can I get a loan for my insurance policy?") and proactively provide valuable recommendation information (such as automatically sending voice prompts based on customer status), forming a complete service loop. This scenario-aware intelligent recommendation mechanism not only significantly improves user satisfaction but also enhances service efficiency by optimizing the interaction process. It provides reliable technical support for the in-depth application of intelligent voice services in various industries and is particularly suitable for fields such as intelligent customer service and voice assistants that require handling complex interaction needs. It provides an innovative solution for building a more intelligent and human-centered voice interaction system.

[0067] Optionally, for the proactive question-and-answer communication scenario, the comprehensive data within the scenario includes the consultation question data of the specified user, and the intelligent interaction module includes a clustering mining module, a knowledge base, and a response module;

[0068] The clustering mining module is used to automatically mine knowledge points that the specified user is concerned about but has not responded to, and recommend valuable questions to human experts in the form of knowledge points. If the human experts judge that they are valuable, they can be directly added to the knowledge base.

[0069] The reply module is used to reply to the knowledge points that the specified user is concerned about but has a response to, and the reply output content is the output result; the action strategy corresponding to the output result is the associated guided teaching.

[0070] In this embodiment of the application, the response module can quickly provide accurate answers to problems with existing solutions; while for newly emerging typical problems, the clustering and mining module can form knowledge accumulation and updates. This knowledge discovery mechanism not only greatly reduces the cost of manual maintenance, but also ensures that the knowledge base can continuously absorb new questions that users are truly concerned about, and always maintain the ability to update and iterate its knowledge reserves.

[0071] Optionally, the process of automatically mining knowledge points that the specified user is interested in but has not yet addressed includes:

[0072] Collect the original questions from users and obtain a corpus of questions that are not covered;

[0073] NLP techniques are used to semantically cluster questions, identify high-frequency uncovered questions, and discover knowledge blind spots.

[0074] Analyze the clustering results, extract common knowledge points, and form expandable knowledge points;

[0075] High-value questions are submitted for manual review to ensure accuracy and prevent incorrect knowledge from being entered into the database.

[0076] Optionally, the method for collecting the original questions from user inquiries and obtaining a corpus of uncovered questions includes:

[0077] Continuously collect original consultation questions input by users via voice or text;

[0078] The original consultation question is preprocessed; the preprocessing includes removing irrelevant characters, filtering stop words, and restoring word forms.

[0079] Encoding the original consultation question into a high-dimensional semantic vector using a pre-trained language model (such as BERT or RoBERTa) can break through the limitations of surface vocabulary and grasp the essential meaning of the question from a deeper level.

[0080] Calculate the maximum cosine similarity between the high-dimensional semantic vector and all existing knowledge points in the knowledge base to evaluate the coverage of the original consultation question;

[0081] When the maximum cosine similarity is lower than a preset similarity threshold, the original consultation question is automatically collected into a corpus of uncovered questions and labeled with tags to be analyzed.

[0082] A multi-level verification mechanism can be employed during the process: first, a rapid coarse matching is performed to filter out obviously uncovered issues, and then fine-grained semantic analysis is conducted on boundary cases. A pre-defined dynamic similarity threshold adjustment module can automatically optimize the judgment criteria based on the knowledge density of different domains. For example, the threshold can be appropriately increased to 0.3 in domains with dense technical terminology, while decreasing to 0.2 in domains of everyday language, thereby ensuring the accuracy of the screening results. Identified uncovered issues are structured and stored according to timestamps, issue types, and domain labels, providing a high-quality data foundation for subsequent semantic analysis and knowledge mining.

[0083] Optionally, the method of using NLP technology to perform semantic clustering of questions, identify high-frequency uncovered question points, and discover knowledge blind spots includes:

[0084] The high-dimensional semantic vector integrates contextual features based on the Transformer architecture and optimized keyword features to form a hybrid semantic space with strong representational capabilities.

[0085] During the clustering process, question clusters with high semantic relevance are identified, and core question points are determined through a density peak detection algorithm.

[0086] Calculate multi-dimensional metrics (occurrence frequency, time distribution, and user group characteristics) for each generated semantic cluster. When the occurrence frequency of a specific question cluster exceeds a preset threshold within a unit of time, it is marked as a high-frequency uncovered question point.

[0087] In addition, a configurable dynamic weight adjustment mechanism automatically optimizes the judgment criteria based on the importance differences of business scenarios, ensuring that knowledge blind spots in key areas are identified first. Furthermore, by analyzing the semantic relationship network between question clusters, the system can identify potential complex knowledge blind spots—knowledge gaps jointly pointed to by multiple related but differently worded questions.

[0088] In this embodiment of the application, the hybrid semantic space can simultaneously capture the contextual information and key elements of the problem, overcoming the limitations of single feature representation. While maintaining semantic depth, it enhances the interpretability of features, providing a more comprehensive and reliable data foundation for subsequent clustering analysis. The density peak detection technology is used to identify core questions, effectively solving the noise interference and boundary ambiguity problems faced by traditional clustering algorithms in problem discovery scenarios. By analyzing the density distribution characteristics of semantic clusters, the system can accurately identify the core questions that truly represent the user's focus, rather than superficially similar peripheral questions. This density peak-based analysis method is particularly suitable for handling common expression differences and semantic overlap phenomena in natural language, significantly improving the quality and usability of clustering results.

[0089] Optionally, the method for analyzing the clustering results, extracting common knowledge points, and forming scalable knowledge points includes:

[0090] Dependency parsing and semantic role labeling are performed on the questions in each semantic cluster to identify the core entities, actions and their relationships in the questions;

[0091] This system employs an attention-based neural network model to extract frequently occurring common semantic patterns and transforms these patterns into structured knowledge triples. For example, in the field of new energy vehicles, the system can automatically extract standardized knowledge units such as <new energy vehicles, winter range, degradation rate> from multiple similar questions like "how much does the driving range decrease in winter?" and "battery degradation rate in low-temperature environments". Through the organic combination of deep semantic analysis and pattern recognition technology, the system achieves intelligent transformation from original questions to scalable knowledge points. In the analysis phase, the system uses dependency parsing and semantic role labeling techniques to deeply analyze the problem structure, accurately identifying core entities, actions, and their interrelationships. The linguistic rule-based parsing method ensures the accuracy and completeness of knowledge element extraction. Compared to traditional simple keyword extraction, it can accurately grasp the essential semantic relationships of the problem, laying a solid foundation for subsequent knowledge modeling.

[0092] Optionally, for the passive question-and-answer communication scenario, the comprehensive data within the scenario includes the real-time monitoring data of the specified user, scenario status information, and scenario voice information;

[0093] The working process of the intelligent interaction module includes:

[0094] Based on the comprehensive data within the scenario, a scenario is constructed, and the scenario construction result is output.

[0095] Based on the scenario setup results, a corresponding recommendation strategy is obtained and voice recommendations are performed.

[0096] For example, in the field of intelligent voice reminders for autonomous driving, based on the acquisition of scene status information, by acquiring real-time monitoring data and voice information within the scene, it is possible to more comprehensively assess the risk of user fatigue driving, thereby improving the accuracy and timeliness of fatigue monitoring. At the same time, this embodiment presets multiple scenarios and sets corresponding recommendation strategies according to the type of scenario. Therefore, scenarios are built based on real-time monitoring data, scene status information, and voice information within the scene. Recommendation strategies are obtained based on the built scenarios and voice recommendations are made. Through this process, corresponding strategies can be provided in advance according to different user states, thereby adapting to the needs of different scenarios.

[0097] In one embodiment, taking intelligent voice reminders for autonomous driving as an example, a scenario building process is provided, including: Specifically, the scenario building process includes:

[0098] Based on real-time monitoring data, the user status of the specified user is analyzed, and a user status analysis index is output.

[0099] Analyze the application status of the scenario based on the scenario status information, and output the scenario device status coefficient;

[0100] Analyze the scene state based on the voice information in the scene and output the scene state coefficient;

[0101] The interval is determined based on the user status analysis index, the scene device status coefficient, and the scene status coefficient value. The combined interval is compared with the corresponding interval of the preset scene to determine the corresponding preset scene and use it as the scene to build.

[0102] The above technical solution quantifies real-time monitoring data, scene status information, and voice information within the scene to obtain user status analysis index, scene device status coefficient, and scene status coefficient. By comparing the intervals of the user status analysis index, scene device status coefficient, and scene status coefficient, the corresponding preset scene is obtained as the scene to be built.

[0103] In this embodiment, real-time monitoring data, scene status information, and scene voice information are divided into two intervals, resulting in a total of 8 combination methods. Each of the 8 combination methods corresponds to a preset scene, and each preset scene has a recommendation strategy suitable for that scene. Therefore, it can adapt to the needs of the scene and improve user satisfaction.

[0104] In one embodiment, the real-time monitoring data includes user heart rate variability, user blinking frequency, duration of a single eye closure, head nodding frequency, duration of head deviation beyond a preset angle, duration of skin conductance below a preset value, and periocular temperature exceeding the average facial temperature. These parameters can all be achieved through common wearable devices and smart sensing modules in vehicles. Specifically, user heart rate variability and duration of skin conductance below the preset value are obtained through a smart bracelet; user blinking frequency, duration of a single eye closure, head nodding frequency, and duration of head deviation beyond a preset angle can be achieved through a camera and recognition algorithm. The recognition algorithm utilizes existing computer vision technology, which will not be detailed here; and the periocular temperature exceeding the average facial temperature can be achieved through an infrared thermal imager and existing computer vision technology. Therefore, the aforementioned real-time monitoring data not only accurately reflects the user's fatigue state but also has superior feasibility. Furthermore, the process of obtaining the user state analysis index includes comparing the real-time monitoring data with corresponding standards, which are set based on empirical data.

[0105] For example, regarding heart rate variability, if its value SDNN < 50ms, then fatigue is judged, so 50ms is its corresponding standard. Similarly, if the blink rate is less than 8 times / minute, then fatigue is judged, so 8 times / minute is its corresponding standard. Therefore, when a certain real-time monitoring data exceeds the corresponding standard, the user state analysis index U = A is set, and U is judged to belong to the first physiological interval U1, indicating that the user's physiological state is fatigued.

[0106] Otherwise, through the formula The user status analysis index U is calculated. This process allows for a comprehensive assessment of user fatigue status by calculating the degree of closeness between the values ​​of various real-time monitoring data and the corresponding standards, even when none exceed the standards. This enables earlier assessment of user fatigue status and improves the timeliness of the monitoring process. Here, i is a positive integer satisfying i∈[1,7], p i Let pt represent the value of the i-th real-time monitoring data item. i The ideal value for the i-th real-time monitoring data is pa, which is set based on empirical data. i This is the quantization unit value for the i-th real-time monitoring data item, which is determined based on the numerical type of each real-time monitoring data item. For example, the quantization unit value for heart rate variability is 1ms. i Let α be the influence coefficient of the i-th real-time monitoring data. When the value of the i-th real-time monitoring data is positively correlated with the user's fatigue state, α i >0, when the value of the i-th real-time monitoring data is negatively correlated with the user's fatigue state, α i <0; This parameter is fitted and set according to the degree of influence of different real-time monitoring data in the test data. Therefore, by measuring the user state analysis index U, the fatigue risk of the user can be judged. The user state analysis index U is compared with the preset physiological threshold At, which is set according to empirical data. If U≥At, the user state analysis index U is judged to belong to the first physiological interval U1; if U<At, the user state analysis index U is judged to belong to the second physiological interval U2. Through the above calculation and comparison process of the user state analysis index, the user's state can be accurately and timely judged. Then, the user's needs can be adapted through subsequent scenario construction, thereby improving user satisfaction while ensuring driving safety.

[0107] In one embodiment, the scene status information includes the number of oncoming vehicles per unit time, cumulative driving time and the proportion of nighttime driving, the curve of total number of turns as a function of driving time, the curve of total number of braking actions as a function of driving time, and the curve of vehicle speed as a function of driving time. All of the above parameters are obtained through vehicle driving data. The process of obtaining the scene equipment status coefficient includes:

[0108] The scene device state coefficient F is obtained by calculating using formulas (1)-(5);

[0109]

[0110]

[0111] Since user fatigue during vehicle operation is caused by factors including time, operation, environment, and vehicle dynamics, the scene equipment state coefficient F is calculated by comprehensively considering the influence of these four factors and then using formula (1); where j is a positive integer and j∈[1,4], λ j For S jThe corresponding influence coefficients are obtained by fitting data based on the degree of influence of different factors. S1 is the driving coefficient, S2 is the environmental coefficient, S3 is the operation coefficient, S4 is the power coefficient, T is the cumulative driving time, η is the proportion of night driving, and x is the adjustment coefficient, which is obtained by fitting test data (x > 1). T0 is the time threshold, which is set based on empirical data; in this embodiment, 2 hours is selected. N is the number of oncoming vehicles per unit time, N0 is the threshold for the number of oncoming vehicles per unit time, which is set based on empirical data, and r(t) is the total number of turns as driving time increases. The time-varying curves are shown, where z(t) represents the total number of braking actions as a function of driving time, t1 is the current time point, t0 is the starting time point, rt is the steering concentration coefficient, zt is the braking concentration coefficient, γ is the steering adjustment coefficient, and τ is the braking adjustment coefficient. Both the steering and braking adjustment coefficients are set based on test data fitting. r0 is the threshold value for steering actions per unit time, z0 is the threshold value for braking actions per unit time, and both the steering and braking action threshold values ​​are set based on empirical data fitting. y1 and y2 are preset proportional coefficients, whose root... According to the difference in the degree of influence, the test data is fitted and set; v(t) is the curve of vehicle speed change with driving time, v0 is the reference value of vehicle speed, which is set according to empirical data, κ is the speed adjustment coefficient, which is set according to the test data fitting, vt is the speed concentration coefficient. Through the calculation process of formula (2), it is possible to judge the degree to which the user's driving time exceeds the standard. Through the calculation process of formula (3), it is possible to judge the traffic density of the vehicle in the past period. Through the calculation process of formula (4), it is possible to judge the user's state by comprehensively considering the steering and braking states during vehicle operation. Through the calculation process of formula (5), it is possible to judge the user's state based on the speed change factors during the user's driving process. Finally, through formula (1), a multi-factor comprehensive judgment process for the user's fatigue state is realized, which improves the accuracy and timeliness of the judgment. In addition, in the calculation process of operation coefficient and power coefficient, it is not only judged by the size of the number of steering and braking and the size of speed, but also by obtaining the corresponding concentration coefficient based on the overall degree of short-term changes in the data to adjust the results. Therefore, the data obtained is more in line with the actual state.

[0112] In one embodiment, the process of obtaining the steering concentration coefficient, braking concentration coefficient and speed concentration coefficient includes: calculating the steering concentration coefficient rt, braking concentration coefficient zt and speed concentration coefficient vt using formulas (7)-(9);

[0113]

[0114] in, The average value of v(t) during the time interval t0 to t1 is given. Through the calculation process of formulas (7) and (8), the concentration state during steering and braking can be determined. When the concentration is greater, the impact on user fatigue is greater, and the calculated values ​​of rt and zt are larger. Then, the size of the operation coefficient can be adjusted through the obtained rt and zt. Similarly, through the calculation process of formula (9), the uniformity of speed change can be determined. When the speed change is less uniform, the impact on user fatigue is greater, and the calculated value of vt is larger, thus realizing the adjustment process of the power coefficient.

[0115] In one embodiment, the process of obtaining scene state coefficients includes: analyzing the acquired voice information in the scene based on AI to determine the state of conversation in the vehicle: if the conversation is of high density, the corresponding interval of the scene state coefficient is set to E1; if the conversation is of low density, the corresponding interval of the scene state coefficient is set to E2; collecting the dialogue content and dialogue density through the AI ​​model to determine the scene state during the user's driving process, and thus obtaining the corresponding interval of the scene state coefficient.

[0116] In one embodiment, the recommended strategy includes, but is not limited to, adjusting voice type, voice fatigue prompts, adjusting air conditioning temperature, adjusting music playback rhythm, and recommending nearby service areas. For different scenarios, this embodiment combines the above-mentioned recommended strategy content to determine the recommended strategy based on the different levels of user fatigue, thereby improving the user experience while ensuring user driving safety.

[0117] In one embodiment, a scenario-based voice intelligent recommendation system is provided, referring to... Figure 2 ,include:

[0118] The data acquisition module is used to obtain comprehensive data within a specified scenario based on the scenario classification of the user's usage scenario.

[0119] The data transmission module is used to input the comprehensive data within the scene into the intelligent interaction module and obtain the corresponding output results;

[0120] The strategy selection module is used to select an action strategy for the specified user based on the output results.

[0121] The scenario classification includes proactive question-and-answer communication scenarios and passive voice recommendation scenarios;

[0122] The proactive question-and-answer communication scenario refers to: engaging in interactive dialogue with the designated user, requiring responses and guidance regarding inquiries;

[0123] The passive voice recommendation scenario means that, based on the specified needs of the specified user, a corresponding interactive action is automatically initiated.

[0124] In one embodiment, a scenario-based voice intelligent recommendation terminal is provided, which is used to execute a scenario-based voice intelligent recommendation method.

[0125] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A scenario-based intelligent voice recommendation method, characterized in that, include: Step 1: Obtain comprehensive data within the specified user scenarios based on the scenario classification. Step 2: Input the comprehensive data within the scenario into the intelligent interaction module and obtain the corresponding output results; Step 3: Select the action strategy for the specified user based on the output results; The scenario classification includes proactive question-and-answer communication scenarios and passive voice recommendation scenarios; In the aforementioned proactive question-and-answer communication scenario, when engaging in interactive dialogue with the designated user, it is necessary to respond to and guide the user regarding their inquiries. For the passive voice recommendation scenario, based on the specified needs of the specified user, the corresponding interactive action is automatically initiated. The working process of the intelligent interaction module includes: Based on the comprehensive data within the scenario, a scenario is constructed, and the scenario construction result is output. Based on the scenario setup results, a corresponding recommendation strategy is obtained and voice recommendations are performed. For the aforementioned proactive question-and-answer communication scenario, the comprehensive data within the scenario includes the consultation question data of the specified user, and the intelligent interaction module includes a clustering mining module, a knowledge base, and a response module; The clustering mining module is used to automatically mine knowledge points that the specified user is concerned about but has not responded to, and recommend valuable questions to human experts in the form of knowledge points. If the human experts judge that they are valuable, they can be directly added to the knowledge base. The reply module is used to reply to the knowledge points that the specified user is concerned about but has a response to, and the reply output content is the output result; the action strategy corresponding to the output result is the associated guided teaching. For the passive voice recommendation scenario, the comprehensive data within the scenario includes real-time monitoring data of the specified user, scenario status information, and scenario-based voice information; The process of setting up the scene includes: Based on real-time monitoring data, the user status of the specified user is analyzed, and a user status analysis index is output. Analyze the application status of the scenario based on the scenario status information, and output the scenario device status coefficient; Analyze the scene state based on the voice information within the scene and output the scene state coefficient.

2. The scene-based intelligent voice recommendation method according to claim 1, characterized in that, The process of automatically identifying knowledge points that the specified user is concerned about but has not yet been addressed includes: Collect the original questions from users and obtain a corpus of questions that are not covered; NLP techniques are used to semantically cluster questions, identify high-frequency uncovered questions, and discover knowledge blind spots. Analyze the clustering results, extract common knowledge points, and form expandable knowledge points; High-value questions are submitted for manual review to ensure accuracy and prevent incorrect knowledge from being entered into the database.

3. The scene-based intelligent voice recommendation method according to claim 2, characterized in that, The method for collecting the original questions from user inquiries and obtaining a corpus of uncovered questions includes: Continuously collect original consultation questions input by users via voice or text; The original consultation question is preprocessed; the preprocessing includes removing irrelevant characters, filtering stop words, and restoring word forms. The original consultation question is encoded into a high-dimensional semantic vector using a pre-trained language model; Calculate the maximum cosine similarity between the high-dimensional semantic vector and all existing knowledge points in the knowledge base to evaluate the coverage of the original consultation question; When the maximum cosine similarity is lower than a preset similarity threshold, the original consultation question is automatically collected into a corpus of uncovered questions and labeled with tags to be analyzed.

4. The scene-based intelligent voice recommendation method according to claim 3, characterized in that, The method of using NLP technology to perform semantic clustering of questions, identify high-frequency uncovered question points, and discover knowledge blind spots includes: The high-dimensional semantic vector integrates contextual features based on the Transformer architecture and optimized keyword features to form a hybrid semantic space with strong representational capabilities. During the clustering process, question clusters with high semantic relevance are identified, and core question points are determined through a density peak detection algorithm. Calculate multi-dimensional metrics for each generated semantic cluster. When the frequency of a specific question cluster exceeds a preset threshold within a unit of time, it is marked as a high-frequency uncovered question.

5. The scene-based intelligent voice recommendation method according to claim 3, characterized in that, The method for analyzing clustering results, extracting common knowledge points, and forming expandable knowledge points includes: Dependency parsing and semantic role labeling are performed on the questions in each semantic cluster to identify the core entities, actions and their relationships in the questions; A neural network model based on an attention mechanism is used to extract frequently occurring common semantic patterns and transform these common semantic patterns into structured knowledge triples.

6. A scenario-based voice intelligent recommendation system, characterized in that, The system employs a scenario-based intelligent voice recommendation method as described in any one of claims 1-5, comprising: The data acquisition module is used to obtain comprehensive data within a specified scenario based on the scenario classification of the user's usage scenario. The data transmission module is used to input the comprehensive data within the scene into the intelligent interaction module and obtain the corresponding output results; The strategy selection module is used to select an action strategy for the specified user based on the output results. The scenario classification includes proactive question-and-answer communication scenarios and passive voice recommendation scenarios; The proactive question-and-answer communication scenario refers to: engaging in interactive dialogue with the designated user, requiring responses and guidance regarding inquiries; The passive voice recommendation scenario means that, based on the specified needs of the specified user, a corresponding interactive action is automatically initiated.

7. A scenario-based voice intelligent recommendation terminal, characterized in that, The terminal is used to execute a scenario-based voice intelligent recommendation method as described in any one of claims 1-5.

Images