A television application recommendation method and system based on multi-physiological signal fusion, a television, and a storage medium

Abstract

The application discloses a television application recommendation method and system based on multi-physiological signal fusion, a television and a storage medium, and relates to the technical field of television application recommendation.The method comprises the following steps: acquiring physiological signals of a user collected by a remote controller, and generating a standardized physiological feature vector according to the physiological signals; acquiring behavior features, time features and scene features; inputting the standardized physiological feature vector, the behavior features, the time features and the scene features into a lightweight end-side calculation model for fusion calculation to obtain user identity data; acquiring multi-modal features and user historical use records; constructing a user portrait according to the user identity data, the multi-modal features and the user historical use records, and obtaining user portrait data; and performing application prediction and recommendation according to the user portrait data, and displaying recommended applications when a proactive recommendation trigger condition is met.The application provides personalized services according to the characteristics of different users, improves the prediction accuracy of recommended content, and thus realizes true non-sensing recognition and proactive prediction.

Description

Technical Field

[0001] This invention relates to the field of smart TV and human-computer interaction technology, and in particular to a TV application recommendation method, system, TV set, and computer-readable storage medium based on multi-physiological signal fusion. Background Technology

[0002] Currently, home televisions are typically shared devices, with significant differences in age, interests, usage time, and viewing preferences among family members. Existing television recommendation systems often rely on single accounts or device usage records, failing to identify "who the current user is," resulting in generalized and untargeted recommendations and a poor user experience. For example, parents primarily watch news or dramas in the evenings, while children use educational or entertainment apps during weekend days.

[0003] Traditional systems cannot automatically switch recommended users, and the frequent manual login and account switching operations are cumbersome, which does not conform to the natural interaction habits of home TV sharing and cannot achieve true "seamless recognition and proactive prediction".

[0004] In other words, existing identity verification methods in home television systems are cumbersome and inefficient, unable to quickly and automatically identify the operator's identity, thus failing to provide personalized services based on the characteristics of different users. Current technologies for predicting user behavior largely rely on simple historical data, resulting in low accuracy in content recommendations and a lack of personalization and contextual adaptability. Most current television system recommendation services are static, unable to proactively recommend content based on dynamic changes in user behavior and different contextual states, leading to shortcomings in user experience.

[0005] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0006] The main objective of this invention is to provide a television application recommendation method, system, television set, and computer-readable storage medium based on multi-physiological signal fusion, aiming to solve the problems of low prediction accuracy of recommended content in the prior art, which cannot provide personalized services according to the characteristics of different users.

[0007] To achieve the above objectives, the present invention provides a television application recommendation method based on multi-physiological signal fusion, the method comprising the following steps:

[0008] The remote control acquires the user's physiological signals through multiple sensor units and generates a standardized physiological feature vector based on the physiological signals.

[0009] Behavioral features, temporal features, and scene features are acquired. The standardized physiological feature vector, behavioral features, temporal features, and scene features are then input into a lightweight edge computing model for fusion calculation to obtain user identity data.

[0010] Obtain multimodal features and user history records, construct a user profile based on the user identity data, the multimodal features, and the user history records, and obtain user profile data;

[0011] Application prediction and recommendation are performed based on the user profile data, and recommended applications are displayed when the active recommendation trigger conditions are met.

[0012] Optionally, in the TV application recommendation method based on the fusion of multiple physiological signals, the physiological signals include skin conductance data, skin temperature data, grip pressure data, and bioimpedance data.

[0013] The process of acquiring the remote control involves collecting the user's physiological signals through multiple sensor units and generating a standardized physiological feature vector based on the physiological signals. Specifically, this includes:

[0014] When a user holds the remote control, the skin point sensing unit collects skin conductance data reflecting the differences in individual skin conductivity characteristics, the temperature sensor collects skin temperature data reflecting individual temperature, the pressure array unit collects grip pressure data regarding differences in hand shape, grip strength, and grip posture, and the bioimpedance detection unit collects bioimpedance data regarding the measurement of skin resistance and phase difference.

[0015] After the data collection time has been preset, the average value, variance, rate of change, and spectral features are extracted from the skin conductance data, skin temperature data, grip pressure data, and bioimpedance data to generate a standardized physiological feature vector.

[0016] Optionally, the TV application recommendation method based on multi-physiological signal fusion, wherein the acquisition of behavioral features, temporal features, and scene features, and the input of the standardized physiological feature vector, the behavioral features, the temporal features, and the scene features into a lightweight edge computing model for fusion calculation to obtain user identity data, specifically includes:

[0017] Behavioral characteristics are collected based on key press rhythm, focus movement speed, and distribution of frequently used applications; time characteristics are collected based on usage period; and scene characteristics are collected based on TV status.

[0018] The standardized physiological feature vector, the behavioral feature, the time feature, and the scene feature are input into a lightweight edge computing model. The lightweight edge computing model performs fusion computing to identify the current operator's identity and obtain user identity data.

[0019] Optionally, the TV application recommendation method based on multi-physiological signal fusion, wherein acquiring multimodal features and user historical usage records, constructing a user profile based on the user identity data, the multimodal features, and the user historical usage records, and obtaining user profile data specifically includes:

[0020] Multimodal features are obtained by combining features from various types of data, and user history records are obtained based on the user's operation and application history data on the TV system.

[0021] A user profile is constructed based on the user identity data, the multimodal features, and the user's historical usage records, and user profile data is generated. The user profile includes behavioral patterns, application usage preferences, time cycle patterns, and scene adaptation characteristics.

[0022] Optionally, the TV application recommendation method based on multi-physiological signal fusion, wherein the step of predicting and recommending applications based on the user profile data, and displaying recommended applications when the active recommendation triggering condition is met, specifically includes:

[0023] Based on the user profile data, time period features, user behavior features, scene state features, physiological signal features, and application usage features are obtained and input into the hierarchical fusion model, which includes a period layer and a context layer.

[0024] The periodic layer calculates long-term weights based on time patterns, and the context layer calculates short-term weights based on the current operating state. The application's overall score (Score) is then calculated based on the long-term and short-term weights.

[0025] Score(App)= α×S_cycle(App)+ β×S_context(App);

[0026] Where S_cycle(App) represents long-term weight, S_context(App) represents short-term weight, and α and β represent two dynamically changing weight coefficients. α is used to control the influence of long-term habits, and β is used to control the influence of current behavior and current TV state.

[0027] The top three apps are ranked based on their overall scores. If the top three apps meet the conditions for proactive recommendation, the recommended apps will be displayed.

[0028] Optionally, in the aforementioned television application recommendation method based on multi-physiological signal fusion, the long-term weight is calculated as follows:

[0029] S_cycle(App)=λ×S_time +γ×S_similar+δ×S_holiday;

[0030] Where S_time represents the weight of the time dimension, S_similar represents the compensation weight for similar time periods, S_holiday represents the weight of holidays, λ represents the exponential decay factor used to adjust the weight of long-term data, γ represents the similarity compensation factor used to compensate for the similarity between different time periods, and δ represents the holiday correction factor used to adjust the weight of special dates.

[0031] Optionally, in the TV application recommendation method based on multi-physiological signal fusion, the active recommendation triggering condition is: the cumulative usage time of the application in the current time period is greater than the first time, the usage time span is greater than the second time, and the current time is located between the third time before the start of the historical interval and the fourth time before the end.

[0032] Furthermore, to achieve the above objectives, the present invention also provides a television application recommendation system based on multi-physiological signal fusion, wherein the television application recommendation system based on multi-physiological signal fusion includes:

[0033] The physiological signal acquisition module is used to acquire the user's physiological signals collected by the remote control through multiple sensor units, and generate a standardized physiological feature vector based on the physiological signals.

[0034] The multimodal fusion recognition module is used to acquire behavioral features, time features, and scene features. The standardized physiological feature vector, the behavioral features, the time features, and the scene features are input into a lightweight edge computing model for fusion calculation to obtain user identity data.

[0035] The user profile modeling module is used to acquire multimodal features and user history usage records, construct user profiles based on the user identity data, the multimodal features, and the user history usage records, and obtain user profile data.

[0036] The application prediction and recommendation module is used to predict and recommend applications based on the user profile data, and to display recommended applications when the active recommendation trigger conditions are met.

[0037] Furthermore, to achieve the above objectives, the present invention also provides a television set, wherein the television set includes: a memory, a processor, and a television application recommendation program based on multi-physiological signal fusion stored in the memory and executable on the processor, wherein when the television application recommendation program based on multi-physiological signal fusion is executed by the processor, it implements the steps of the television application recommendation method based on multi-physiological signal fusion as described above.

[0038] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a television application recommendation program based on multi-physiological signal fusion, and the television application recommendation program based on multi-physiological signal fusion, when executed by a processor, implements the steps of the television application recommendation method based on multi-physiological signal fusion as described above.

[0039] In this invention, a remote control collects the user's physiological signals through multiple sensor units, and generates a standardized physiological feature vector based on these signals. Behavioral features, temporal features, and scene features are acquired, and these are then input into a lightweight edge computing model for fusion calculation to obtain user identity data. Multimodal features and user historical usage records are acquired, and a user profile is constructed based on the user identity data, multimodal features, and user historical usage records, resulting in user profile data. Application prediction and recommendation are performed based on the user profile data, and recommended applications are displayed when proactive recommendation trigger conditions are met. This invention can automatically identify user identity without login and predict the application the user may use next, providing personalized services based on different user characteristics, improving the prediction accuracy of recommended content, and thus achieving true seamless identification and proactive prediction. Attached Figure Description

[0040] Figure 1 This is a flowchart of a preferred embodiment of the television application recommendation method based on multi-physiological signal fusion of the present invention;

[0041] Figure 2 This is a schematic diagram illustrating the application recommendation process completed by various modules in a preferred embodiment of the TV application recommendation method based on multi-physiological signal fusion of the present invention.

[0042] Figure 3 This is a schematic diagram of the physiological signal acquisition module collecting data in a preferred embodiment of the television application recommendation method based on multi-physiological signal fusion of the present invention;

[0043] Figure 4 This is a schematic diagram of the data fusion calculation performed by the multimodal fusion recognition module in a preferred embodiment of the television application recommendation method based on multi-physiological signal fusion of the present invention;

[0044] Figure 5 This is a schematic diagram of the user profile modeling module constructing a user profile in a preferred embodiment of the TV application recommendation method based on multi-physiological signal fusion of the present invention;

[0045] Figure 6 This is a schematic diagram illustrating the application prediction and recommendation module performing application recommendation in a preferred embodiment of the TV application recommendation method based on multi-physiological signal fusion of the present invention;

[0046] Figure 7 This is a structural diagram of a preferred embodiment of the television application recommendation system based on multi-physiological signal fusion of the present invention;

[0047] Figure 8 This is a structural diagram of a preferred embodiment of the television set of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0049] The preferred embodiment of the television application recommendation method based on multi-physiological signal fusion described in this invention, such as... Figure 1 and Figure 2 As shown, the television application recommendation method based on multi-physiological signal fusion includes the following steps:

[0050] Step S10: Obtain the user's physiological signals collected by the remote control through multiple sensor units, and generate a standardized physiological feature vector based on the physiological signals.

[0051] Specifically, such as Figure 2 As shown, step S10 is completed through the physiological signal acquisition module in the remote control. The remote control acquires the user's physiological signals through multiple sensor units. These physiological signals include skin conductance response data, skin temperature data, grip pressure data, and bioimpedance data. Skin conductance response: Detects skin electrical activity through changes in conductivity, reflecting individual differences in skin conductivity characteristics. Skin temperature: Measured by a thermistor or infrared sensor; individual temperature distribution is stable and can be used to distinguish users. Grip pressure distribution: A pressure array sensor acquires differences in hand shape, grip force, and grip posture. Bioimpedance: Microcurrent measures skin resistance and phase difference, used to distinguish characteristics such as body fat percentage and water content.

[0052] like Figure 3 As shown, when a user holds the remote control, the skin point sensing unit collects skin conductance data reflecting individual differences in skin conductivity, the temperature sensor collects individual skin temperature data, the pressure array unit collects grip pressure data reflecting differences in hand shape, grip strength, and grip posture, and the bioimpedance detection unit collects bioimpedance data regarding the measurement of skin resistance and phase difference. After the data collection time has been preset (e.g., 3-5 seconds), the physiological signal acquisition module extracts parameters such as average value, variance, rate of change, and spectral characteristics based on the skin conductance data, skin temperature data, grip pressure data, and bioimpedance data to generate a standardized physiological feature vector, and sends the standardized physiological feature vector to the multimodal fusion recognition module of the television.

[0053] Step S20: Obtain behavioral features, time features, and scene features. Input the standardized physiological feature vector, behavioral features, time features, and scene features into a lightweight edge computing model for fusion calculation to obtain user identity data.

[0054] Specifically, such as Figure 2 As shown, step S20 is completed by the multimodal fusion recognition module in the television. This module acquires behavioral features, temporal features, and scene features. Behavioral features include: button press rhythm, focus movement speed, and distribution of frequently used applications; temporal features include: usage time period (morning / evening / weekend / holiday); and scene features include: television status (power on, playing, paused, voice interaction). Figure 4 As shown, behavioral characteristics are collected based on key rhythm, focus movement speed, and distribution of frequently used applications; time characteristics are collected based on usage period; and scene characteristics are collected based on TV status. The standardized physiological feature vector, behavioral characteristics, time characteristics, and scene characteristics are input into a lightweight end-side computing model. The lightweight end-side computing model performs fusion calculation to identify the current operator's identity, obtain user identity data, and send the user identity data to the user profile modeling module of the TV.

[0055] Step S30: Obtain multimodal features and user history usage records, construct a user profile based on the user identity data, the multimodal features, and the user history usage records, and obtain user profile data.

[0056] Specifically, such as Figure 2 As shown, step S30 is completed through the user profile modeling module in the television set. Multimodal features refer to feature combinations derived from multiple different types of data. These multimodal features are independently collected from different modules, hardware, and data sources. In other words, features from multiple different types of data are combined to obtain multimodal features. User history usage records are obtained based on historical data of user operations and applications on the television system; for example... Figure 5 As shown, a user profile is constructed based on the user identity data, the multimodal features, and the user's historical usage records, generating user profile data. The user profile includes behavioral patterns, application usage preferences, time cycle patterns, and scene adaptation features (referring to the user's historical data on the TV system's operations and application usage). The user profile data is then sent to the TV's application prediction and recommendation module.

[0057] Step S40: Based on the user profile data, perform application prediction and recommendation, and display recommended applications when the active recommendation trigger condition is met.

[0058] Specifically, such as Figure 2As shown, step S40 is completed through the application prediction and recommendation module in the television set, such as... Figure 6 As shown, based on the user profile data, time period features, user behavior features, scene state features, physiological signal features, and application usage features are obtained and input into the hierarchical fusion model. Time period features include: intraday time periods, intraweek distribution, and usage patterns over the past 8 days; user behavior features include: last application use, dwell time, and operation rhythm; scene state features include: TV status, brightness, volume, and external connections; physiological signal features include: grip strength, temperature, conductivity, and rate of change of impedance; and application usage features include: usage records per minute and a weight matrix. The hierarchical fusion model includes a period layer and a context layer. The period layer calculates long-term weights based on time patterns, and the context layer calculates short-term weights based on the current operation state (i.e., short-term corrected scores, which are mainly calculated based on the user's recent operation behavior and usage habits). The comprehensive score (Score) of the application is calculated based on the long-term weights and the short-term weights.

[0059] Score(App)= α×S_cycle(App)+ β×S_context(App);

[0060] Where S_cycle(App) represents long-term weight, S_context(App) represents short-term weight, and α and β represent two dynamically changing weight coefficients that are dynamically adjusted by the scene adaptation mechanism. α is used to control the degree of influence of long-term habits, and β is used to control the degree of influence of current behavior and current TV state.

[0061] The long-term weight is calculated as follows:

[0062] S_cycle(App)=λ×S_time +γ×S_similar+δ×S_holiday;

[0063] Where S_time represents the weight of the time dimension, S_similar represents the compensation weight for similar time periods, S_holiday represents the weight of holidays, λ represents the exponential decay factor used to adjust the weight of long-term data, γ represents the similarity compensation factor used to compensate for the similarity between different time periods, and δ represents the holiday correction factor used to adjust the weight of special dates.

[0064] λ: As time goes on, the importance of historical data gradually decreases. It is used to adjust the weight of long-term data to ensure that the model pays more attention to recent behavior.

[0065] γ: Used to compensate for similarities between different time periods. This factor is weighted according to the patterns observed in certain specific time periods.

[0066] δ: User behavior on holidays may differ significantly from that on weekdays. This is used to adjust the weight of special dates to ensure that holiday data is not overlooked.

[0067] Specifically, similar time period compensation refers to the fact that user behavior patterns in certain time periods may repeat in other similar time periods, rather than being limited to the same specific time point. By linking time periods with similar user habits as supplementary weights, recommendations can be made more accurate. Holiday correction refers to the fact that user habits may differ during holidays, requiring special adjustments to recommendations to better reflect the actual needs of holidays.

[0068] Time dimension: For example, the weight of the most recent ±30 minutes is set to 3; the weight of ±90~180 minutes is set to 2; and the weight of other times is set to 1. Date dimension: The weight of the previous day is set to 3; the weight of the previous two days or the same day of last week is set to 2; and the weight of earlier times is set to 1. The larger the weight, the greater the impact on the final score.

[0069] Context layer: If continuous switching or frequent exits are detected, increase the weight of the two most recent applications (i.e., when continuous switching or frequent exits are detected as current behavioral characteristics, increase the score of these apps in the short-term weight S_context(App)); TV status correction: Power-on status - applications before last shutdown; Playback status - similar category applications; Voice interaction - based on semantic keyword matching.

[0070] The top three apps are ranked based on their overall scores (i.e., the Top-3 recommended apps). If the top three apps meet the active recommendation trigger conditions, the recommended apps will be displayed. The active recommendation trigger conditions are: the cumulative usage time of the app in the current time period is greater than the first time period (e.g., 60 minutes), the usage time span is greater than the second time period (e.g., 30 minutes), and the current time is between the third time period before the start of the historical interval (e.g., 5 minutes) and the fourth time period before the end (e.g., 15 minutes). All three conditions must be met simultaneously for a recommendation to be made, ensuring that the user's current behavior is a stable behavior and avoiding accidental triggering of active recommendations.

[0071] In addition, self-learning and dynamic weight updates are set up. For example, if the recommendation hits, the feature weight is increased; if the recommendation fails, the feature impact is reduced; if the user manually cancels the recommendation, the weight is temporarily reduced (shifted 3-6 positions to the right).

[0072] This invention addresses identity recognition and user behavior prediction in television systems, enabling more accurate user identification based on multiple physiological signals and operational behaviors. It emphasizes predicting the user's next action through a comprehensive assessment of "time period, behavioral characteristics, and scene state," and provides personalized, scene-adaptive proactive recommendation services. It offers a technical solution integrating multiple data sources, dynamically adjusting the recommendation system, and optimizing the user experience.

[0073] The beneficial effects of this invention are:

[0074] (1) Accurate multi-source fusion identification: Combining physiological, behavioral and time signals to achieve efficient and accurate user identification.

[0075] (2) Periodic prediction recommendation: Based on time patterns, proactively predict the applications that users are about to open.

[0076] (3) Strong context adaptability: The algorithm can automatically adjust the weights according to the TV status, environment and operation mode.

[0077] (4) Privacy protection and low latency: Recognition and reasoning are performed locally on the TV without cloud transmission.

[0078] Furthermore, such as Figure 7 As shown, based on the above-mentioned television application recommendation method based on multi-physiological signal fusion, the present invention also provides a television application recommendation system based on multi-physiological signal fusion, wherein the television application recommendation system based on multi-physiological signal fusion includes:

[0079] Physiological signal acquisition module 51 is used to acquire physiological signals of the user collected by the remote control through multiple sensor units, and generate a standardized physiological feature vector based on the physiological signals;

[0080] The multimodal fusion recognition module 52 is used to acquire behavioral features, time features and scene features, and input the standardized physiological feature vector, the behavioral features, the time features and the scene features into the lightweight edge computing model for fusion calculation to obtain user identity data;

[0081] User profile modeling module 53 is used to acquire multimodal features and user history usage records, construct user profiles based on the user identity data, the multimodal features and the user history usage records, and obtain user profile data;

[0082] The application prediction and recommendation module 54 is used to predict and recommend applications based on the user profile data, and to display recommended applications when the active recommendation triggering conditions are met.

[0083] Furthermore, such as Figure 8 As shown, based on the above-mentioned method and system for recommending television applications based on multi-physiological signal fusion, the present invention also provides a television set, which includes a processor 10, a memory 20 and a display 30. Figure 8 Only some of the components of the television are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0084] In some embodiments, the memory 20 may be an internal storage unit of the television, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the television, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., provided on the television. Furthermore, the memory 20 may include both internal and external storage units of the television. The memory 20 is used to store application software and various types of data installed on the television, such as the program code installed on the television. The memory 20 may also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a television application recommendation program 40 based on multi-physiological signal fusion, which can be executed by the processor 10 to implement the television application recommendation method based on multi-physiological signal fusion in this application.

[0085] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the television application recommendation method based on multi-physiological signal fusion.

[0086] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-screen liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the television and to display a visual user interface. The processor 10, memory 20, and display 30 of the television communicate with each other via a system bus.

[0087] In one embodiment, when the processor 10 executes the television application recommendation program 40 based on multi-physiological signal fusion in the memory 20, the following steps are performed:

[0088] The remote control acquires the user's physiological signals through multiple sensor units and generates a standardized physiological feature vector based on the physiological signals.

[0089] Behavioral features, temporal features, and scene features are acquired. The standardized physiological feature vector, behavioral features, temporal features, and scene features are then input into a lightweight edge computing model for fusion calculation to obtain user identity data.

[0090] Obtain multimodal features and user history records, construct a user profile based on the user identity data, the multimodal features, and the user history records, and obtain user profile data;

[0091] Application prediction and recommendation are performed based on the user profile data, and recommended applications are displayed when the active recommendation trigger conditions are met.

[0092] The physiological signals include skin conductance data, skin temperature data, grip pressure data, and bioimpedance data.

[0093] The process of acquiring the remote control involves collecting the user's physiological signals through multiple sensor units and generating a standardized physiological feature vector based on the physiological signals. Specifically, this includes:

[0094] When a user holds the remote control, the skin point sensing unit collects skin conductance data reflecting the differences in individual skin conductivity characteristics, the temperature sensor collects skin temperature data reflecting individual temperature, the pressure array unit collects grip pressure data regarding differences in hand shape, grip strength, and grip posture, and the bioimpedance detection unit collects bioimpedance data regarding the measurement of skin resistance and phase difference.

[0095] After the data collection time has been preset, the average value, variance, rate of change, and spectral features are extracted from the skin conductance data, skin temperature data, grip pressure data, and bioimpedance data to generate a standardized physiological feature vector.

[0096] Specifically, the acquisition of behavioral features, time features, and scene features involves inputting the standardized physiological feature vector, the behavioral features, the time features, and the scene features into a lightweight edge computing model for fusion calculation to obtain user identity data, which includes:

[0097] Behavioral characteristics are collected based on key press rhythm, focus movement speed, and distribution of frequently used applications; time characteristics are collected based on usage period; and scene characteristics are collected based on TV status.

[0098] The standardized physiological feature vector, the behavioral feature, the time feature, and the scene feature are input into a lightweight edge computing model. The lightweight edge computing model performs fusion computing to identify the current operator's identity and obtain user identity data.

[0099] The step of acquiring multimodal features and user history records, constructing a user profile based on the user identity data, the multimodal features, and the user history records, and obtaining user profile data specifically includes:

[0100] Multimodal features are obtained by combining features from various types of data, and user history records are obtained based on the user's operation and application history data on the TV system.

[0101] A user profile is constructed based on the user identity data, the multimodal features, and the user's historical usage records, and user profile data is generated. The user profile includes behavioral patterns, application usage preferences, time cycle patterns, and scene adaptation characteristics.

[0102] Specifically, the step of predicting and recommending applications based on the user profile data, and displaying recommended applications when the active recommendation trigger condition is met, includes:

[0103] Based on the user profile data, time period features, user behavior features, scene state features, physiological signal features, and application usage features are obtained and input into the hierarchical fusion model, which includes a period layer and a context layer.

[0104] The periodic layer calculates long-term weights based on time patterns, and the context layer calculates short-term weights based on the current operating state. The application's overall score (Score) is then calculated based on the long-term and short-term weights.

[0105] Score(App)= α×S_cycle(App)+ β×S_context(App);

[0106] Where S_cycle(App) represents long-term weight, S_context(App) represents short-term weight, and α and β represent two dynamically changing weight coefficients. α is used to control the influence of long-term habits, and β is used to control the influence of current behavior and current TV state.

[0107] The top three apps are ranked based on their overall scores. If the top three apps meet the conditions for proactive recommendation, the recommended apps will be displayed.

[0108] The long-term weight is calculated as follows:

[0109] S_cycle(App)=λ×S_time +γ×S_similar+δ×S_holiday;

[0110] Where S_time represents the weight of the time dimension, S_similar represents the compensation weight for similar time periods, S_holiday represents the weight of holidays, λ represents the exponential decay factor used to adjust the weight of long-term data, γ represents the similarity compensation factor used to compensate for the similarity between different time periods, and δ represents the holiday correction factor used to adjust the weight of special dates.

[0111] The active recommendation triggering conditions are as follows: the cumulative usage time of the application in the current period is greater than the first time, the usage time span is greater than the second time, and the current time is between the third time before the start of the historical interval and the fourth time before the end.

[0112] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a television application recommendation program based on multi-physiological signal fusion, and the television application recommendation program based on multi-physiological signal fusion, when executed by a processor, implements the steps of the television application recommendation method based on multi-physiological signal fusion as described above.

[0113] In summary, this invention provides a television application recommendation method, system, television set, and computer-readable storage medium based on multi-physiological signal fusion. The method includes: acquiring physiological signals collected by a remote control through multiple sensor units from a user, and generating a standardized physiological feature vector based on the physiological signals; acquiring behavioral features, temporal features, and scene features, and inputting the standardized physiological feature vector, behavioral features, temporal features, and scene features into a lightweight edge computing model for fusion calculation to obtain user identity data; acquiring multimodal features and user historical usage records, constructing a user profile based on the user identity data, multimodal features, and user historical usage records, and obtaining user profile data; and performing application prediction and recommendation based on the user profile data, displaying recommended applications when proactive recommendation trigger conditions are met. This invention can automatically identify user identity without login and predict the application the user may use next, providing personalized services based on the characteristics of different users, improving the prediction accuracy of recommended content, and thus achieving true seamless identification and proactive prediction.

[0114] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or television set that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or television set. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or television set that includes that element.

[0115] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The computer-readable storage medium can be a memory, magnetic disk, optical disk, etc.

[0116] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A television application recommendation method based on multi-physiological signal fusion, characterized in that, The television application recommendation method based on multi-physiological signal fusion includes: The remote control acquires the user's physiological signals through multiple sensor units and generates a standardized physiological feature vector based on the physiological signals. Behavioral features are collected based on key rhythm, focus movement speed, and distribution of frequently used applications. Time features are collected based on usage time periods. Scene features are collected based on TV status. The standardized physiological feature vector, behavioral features, time features, and scene features are input into a lightweight terminal computing model for fusion calculation to obtain user identity data. Acquire multimodal features and user history records, construct a user profile based on the user identity data, the multimodal features, and the user history records, and obtain user profile data. The multimodal features refer to feature combinations derived from multiple different types of data, and the multimodal features are independently collected by different modules, different hardware, and different data sources. Based on the user profile data, application prediction and recommendation are performed, and recommended applications are displayed when the active recommendation trigger conditions are met. The step of predicting and recommending applications based on the user profile data, and displaying recommended applications when the active recommendation trigger condition is met, specifically includes: Based on the user profile data, time period features, user behavior features, scene state features, physiological signal features, and application usage features are obtained and input into the hierarchical fusion model, which includes a period layer and a context layer. The periodic layer calculates long-term weights based on time patterns, and the context layer calculates short-term weights based on the current operating state. The application's overall score (Score) is then calculated based on the long-term and short-term weights. Score(App)= α×S_cycle(App)+ β×S_context(App); Where S_cycle(App) represents long-term weight, S_context(App) represents short-term weight, and α and β represent two dynamically changing weight coefficients. α is used to control the influence of long-term habits, and β is used to control the influence of current behavior and current TV state. The top three apps are ranked based on their overall scores. If the top three apps meet the active recommendation trigger conditions, the recommended apps will be displayed. The long-term weight is calculated as follows: S_cycle(App)=λ×S_time +γ×S_similar+δ×S_holiday; Where S_time represents the weight of the time dimension, S_similar represents the compensation weight for similar time periods, S_holiday represents the weight of holidays, λ represents the exponential decay factor used to adjust the weight of long-term data, γ represents the similarity compensation factor used to compensate for the similarity between different time periods, and δ represents the holiday correction factor used to adjust the weight of special dates.

2. The television application recommendation method based on multi-physiological signal fusion according to claim 1, characterized in that, The physiological signals include skin conductance data, skin temperature data, grip pressure data, and bioimpedance data; The process of acquiring the remote control involves collecting the user's physiological signals through multiple sensor units and generating a standardized physiological feature vector based on the physiological signals. Specifically, this includes: When a user holds the remote control, the skin point sensing unit collects skin conductance data reflecting the differences in individual skin conductivity characteristics, the temperature sensor collects skin temperature data reflecting individual temperature, the pressure array unit collects grip pressure data regarding differences in hand shape, grip strength, and grip posture, and the bioimpedance detection unit collects bioimpedance data regarding the measurement of skin resistance and phase difference. After the data collection time has been preset, the average value, variance, rate of change, and spectral features are extracted from the skin conductance data, skin temperature data, grip pressure data, and bioimpedance data to generate a standardized physiological feature vector.

3. The television application recommendation method based on multi-physiological signal fusion according to claim 1, characterized in that, The process of inputting the standardized physiological feature vector, the behavioral feature, the temporal feature, and the scene feature into a lightweight edge computing model for fusion calculation to obtain user identity data specifically includes: The standardized physiological feature vector, the behavioral feature, the time feature, and the scene feature are input into a lightweight edge computing model. The lightweight edge computing model performs fusion computing to identify the current operator's identity and obtain user identity data.

4. The television application recommendation method based on multi-physiological signal fusion according to claim 1, characterized in that, The process of acquiring multimodal features and user history records, constructing a user profile based on the user identity data, the multimodal features, and the user history records, and obtaining user profile data specifically includes: Multimodal features are obtained by combining features from various types of data, and user history records are obtained based on the user's operation and application history data on the TV system. A user profile is constructed based on the user identity data, the multimodal features, and the user's historical usage records, and user profile data is generated. The user profile includes behavioral patterns, application usage preferences, time cycle patterns, and scene adaptation characteristics.

5. The television application recommendation method based on multi-physiological signal fusion according to claim 1, characterized in that, The active recommendation triggering conditions are: the cumulative usage time of the application in the current period is greater than the first time, the usage time span is greater than the second time, and the current time is between the third time before the start of the historical interval and the fourth time before the end.

6. A television application recommendation system based on multi-physiological signal fusion, characterized in that, The television application recommendation system based on multi-physiological signal fusion is used to implement the television application recommendation method based on multi-physiological signal fusion as described in any one of claims 1-5, wherein the television application recommendation system based on multi-physiological signal fusion comprises: The physiological signal acquisition module is used to acquire the user's physiological signals collected by the remote control through multiple sensor units, and generate a standardized physiological feature vector based on the physiological signals. The multimodal fusion recognition module is used to acquire behavioral features, time features, and scene features. The standardized physiological feature vector, the behavioral features, the time features, and the scene features are input into a lightweight edge computing model for fusion calculation to obtain user identity data. The user profile modeling module is used to acquire multimodal features and user history usage records, construct user profiles based on the user identity data, the multimodal features, and the user history usage records, and obtain user profile data. The application prediction and recommendation module is used to predict and recommend applications based on the user profile data, and to display recommended applications when the active recommendation trigger conditions are met.

7. A television set, characterized in that, The television set includes: a memory, a processor, and a television application recommendation program based on multi-physiological signal fusion stored in the memory and executable on the processor. When the multi-physiological signal fusion-based television application recommendation program is executed by the processor, it implements the steps of the television application recommendation method based on multi-physiological signal fusion as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a television application recommendation program based on multi-physiological signal fusion, which, when executed by a processor, implements the steps of the television application recommendation method based on multi-physiological signal fusion as described in any one of claims 1-5.

Images