Interaction strategy generation method based on user data, cloud equipment and robot equipment

Abstract

The invention discloses an interaction strategy generation method based on user data, cloud equipment and robot equipment, and relates to the technical field of robots, and the method comprises the steps: receiving target user data sent by a robot end, and determining the target user data of the target user data; performing emotion analysis on the target user data through a preset analysis model to obtain an emotion state of the target user; a target portrait is constructed according to historical user data of the target user, and the historical user data comprises historically received target user data; generating a target interaction strategy based on the target portrait and the emotional state; the target interaction strategy is sent to the robot end, and the robot end is used for receiving the target interaction strategy sent by the cloud end and executing the target interaction strategy. The emotional connection and accompanying effect between the user and the robot is enhanced.

Description

Technical Field

[0001] The present application relates to the field of robotics technology, and in particular to a method for generating an interaction strategy based on user data, a cloud device, and a robotic device. Background Art

[0002] With the development of artificial intelligence (AI), the demand for intelligent companion robots is growing. While existing companion robots possess basic interactive capabilities, they still lack emotional understanding and personalized companionship. Traditional robots are often limited to simple voice conversations and preset actions, unable to conduct in-depth analysis and respond based on long-term user behavior data. This results in a limited emotional connection between users and robots, making it difficult for them to truly become intimate companions. Summary of the Invention

[0003] The main purpose of this application is to provide a method for generating an interaction strategy based on user data, a cloud device and a robot device, aiming to solve the technical problem of imbalance in the generation of interaction strategies based on user data.

[0004] To achieve the above objectives, this application proposes a method for generating an interaction strategy based on user data, which is applied to the cloud. The cloud is connected to a preset robot terminal for communication. The method for generating an interaction strategy based on user data includes:

[0005] receiving target user data sent by the robot end, determining target user data of the target user data, wherein the robot end is used to collect the target user data of the target user; and sending the target user data to the cloud;

[0006] Perform sentiment analysis on target user data through a preset analysis model to obtain the target user's emotional state;

[0007] Building a target profile based on the target user's historical user data, wherein the historical user data includes the target user data received historically;

[0008] Generate target interaction strategies based on target profile and emotional state;

[0009] The target interaction strategy is sent to the robot side, where the robot side is used to receive the target interaction strategy sent by the cloud and execute the target interaction strategy.

[0010] In one embodiment, after receiving the target user data sent by the robot and determining the target user of the target user data, the process further includes:

[0011] Decrypting the target user data to obtain plaintext user data, wherein the target user data is encrypted data;

[0012] Removing personal identification information from plaintext user data to obtain anonymous user data, wherein the anonymous user data is used for sentiment analysis using a preset analysis model;

[0013] Update anonymous user data to historical user data.

[0014] In one embodiment, the step of performing sentiment analysis on target user data to obtain the target user's emotional state includes:

[0015] Extract features from target user data to obtain multiple sets of feature parameters;

[0016] The feature parameters are fused to obtain the joint sentiment feature vector;

[0017] Input the joint sentiment feature vector into the preset sentiment classification model;

[0018] The joint emotional feature vector is analyzed and processed through the emotion classification model to obtain the emotional state of the target user.

[0019] In one embodiment, the step of constructing a target profile based on the historical user data of the target user includes:

[0020] Classify the historical user data of the target user and extract feature vectors for different categories of historical user data;

[0021] Label the feature vector to obtain the portrait label;

[0022] Use the preset portrait construction algorithm to construct the portrait according to the portrait label to generate the target portrait.

[0023] In one embodiment, the step of generating a target interaction strategy based on the target profile and emotional state includes:

[0024] Retrieve the interaction strategy that matches the target profile and emotional state from the preset interaction strategy library;

[0025] If there is an interaction strategy retrieval result, then determine the target interaction strategy according to the interaction strategy retrieval result;

[0026] If there is no interaction strategy retrieval result, the target interaction strategy is generated based on the target profile and emotional state.

[0027] In one embodiment, the step of sending the target interaction strategy to the robot terminal includes:

[0028] Receive feedback data sent by the robot;

[0029] Categorize feedback data;

[0030] Compare and analyze the classified feedback data with the preset historical feedback data to determine the changing trend of the feedback data;

[0031] Adjust the weight parameters in the analysis model according to the changing trend.

[0032] A method for generating an interaction strategy based on user data, characterized in that it is applied to a robot end, the robot end is connected to a preset cloud communication, and the method for generating an interaction strategy based on user data includes:

[0033] Collect target user data of target users;

[0034] Sending target user data to the cloud, wherein the cloud is used to receive the target user data sent by the robot end and determine the target user data of the target user data; performing sentiment analysis on the target user data using a preset analysis model to obtain the target user's emotional state; building a target profile based on the target user's historical user data, wherein the historical user data includes historically received target user data; generating a target interaction strategy based on the target profile and emotional state, and sending the target interaction strategy to the robot end;

[0035] Receive the target interaction strategy sent by the cloud and execute the target interaction strategy.

[0036] In one embodiment, the target user data includes at least one of touch data, distance data, posture data, position data, motion data, voice data, and external user data. The step of collecting the target user data of the target user includes:

[0037] Controlling the touch sensor array to collect touch actions of a target user and generate touch data; and / or,

[0038] Controlling the infrared sensor to collect the relative distance between the target user and the robot and generate distance data; and / or,

[0039] Controlling the IMU sensor to collect posture change information of the robot and generate posture data; and / or,

[0040] Controlling the GPS locator to collect the location information of the robot and generate location data; and / or,

[0041] Controlling the camera to capture the target user's expressions and movements and generate movement data; and / or,

[0042] Controlling the microphone array to collect the target user's voice and generate voice data; and / or,

[0043] Obtain external device information bound to the target user, and obtain external user data corresponding to the external device information, wherein the external user data includes user social data and schedule data.

[0044] In addition, to achieve the above-mentioned purpose, the present application also proposes a cloud device, which includes: a memory, a processor, and a computer program stored in the memory and runnable on the processor, and the computer program is configured to implement the steps of the user data-based interaction strategy generation method applied to the cloud as described above.

[0045] In addition, to achieve the above-mentioned purpose, the present application also proposes a robot device, which includes: a memory, a processor, and a computer program stored in the memory and runnable on the processor, and the computer program is configured to implement the steps of the user data-based interaction strategy generation method applied to the robot side as described above.

[0046] In addition, to achieve the above-mentioned purpose, the present application also proposes a medium, which is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for generating an interaction strategy based on user data as described above are implemented.

[0047] In addition, to achieve the above-mentioned purpose, the present application also provides a product, which is a computer program product. The computer program product includes a computer program. When the computer program is executed by a processor, it implements the steps of the above-mentioned method for generating an interaction strategy based on user data.

[0048] One or more technical solutions proposed in this application have at least the following technical effects:

[0049] This application receives target user data sent by the robot side and determines the target user data of the target user data, wherein the robot side is used to collect the target user data of the target user; sends the target user data to the cloud, and comprehensively perceives the user's emotional state through multimodal fusion technology to provide rich data support for accurate emotional analysis; performs emotional analysis on the target user data through a preset analysis model to obtain the target user's emotional state, realizes real-time and accurate recognition of the user's emotions, and provides a basis for the formulation of personalized interaction strategies; constructs a target profile based on the target user's historical user data, wherein the historical user data includes historically received target user data, deeply mines the user's behavior patterns and emotional needs, and enables the robot to understand the user's long-term preferences and personalized characteristics; generates a target interaction strategy based on the target profile and emotional state to ensure that the interaction content is highly matched with the user's current emotions and historical preferences, thereby improving the fit and effectiveness of emotional companionship; sends the target interaction strategy to the robot side, wherein the robot side is used to receive the target interaction strategy sent by the cloud, execute the target interaction strategy, and provide the user with a real-time, natural, and warm emotional interaction experience through the robot's multimodal response methods such as voice, expression, and action, thereby enhancing the emotional connection and companionship effect between the user and the robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0052] Figure 1 This is a flowchart of the first embodiment of the method for generating an interaction strategy based on user data of the present application;

[0053] Figure 2 This is an interactive diagram of the method for generating an interactive strategy based on user data in this application;

[0054] Figure 3 This is a process diagram of Example 3 of the method for generating an interaction strategy based on user data of this application;

[0055] Figure 4 This is a schematic diagram of the device structure of the hardware operating environment involved in the cloud device in the embodiment of the present application;

[0056] Figure 5 This is a schematic diagram of the device structure of the hardware operating environment involved in the robot device in the embodiment of the present application.

[0057] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0058] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0059] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0060] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, mobile phone, etc., or an electronic device or terminal system capable of implementing the above functions. The following uses the system as an example to illustrate this embodiment and the following embodiments.

[0061] Based on this, this embodiment provides a method for generating an interaction strategy based on user data. Figure 1 , Figure 1This is a flow chart of a method for generating an interaction strategy based on user data applied to the cloud in this application. The method for generating an interaction strategy based on user data is applied to the cloud, and the cloud is connected to a preset robot terminal for communication. The method for generating an interaction strategy based on user data includes steps S10 to S50:

[0062] Step S10, receiving target user data sent by the robot end, determining target user data of the target user data, wherein the robot end is used to collect target user data of the target user; and sending the target user data to the cloud;

[0063] Step S20, performing sentiment analysis on the target user data using a preset analysis model to obtain the target user's emotional state;

[0064] Step S30, constructing a target profile based on the historical user data of the target user, wherein the historical user data includes the target user data received historically;

[0065] Step S40, generating a target interaction strategy based on the target portrait and emotional state;

[0066] Step S50: Send the target interaction strategy to the robot side, wherein the robot side is used to receive the target interaction strategy sent by the cloud and execute the target interaction strategy.

[0067] It should be noted that in this embodiment, the target user data refers to the voice, touch and behavior data collected by the robot side, including information such as the user's tone, speaking speed, volume, touch strength, position, duration and activity trajectory; the target portrait refers to a set of user features constructed by integrating the historical user data of the target user (such as voice, touch and behavior records), covering basic information, interests and hobbies, emotional tendencies and behavior patterns; the emotional state is the user's current emotional label obtained after processing the target user data through the emotion analysis model; the target interaction strategy is an instruction set generated based on the target portrait and emotional state, used to guide the robot side to execute voice, expression or action responses, including speech synthesis text, expression display rules and action control parameters; the cloud refers to a server cluster that deploys the emotion analysis model and user portrait construction algorithm, which is responsible for receiving data transmitted by the robot side and returning the interaction strategy to the robot side.

[0068] This embodiment receives target user data through step S10. The robot side collects the user's voice data (such as tone and speaking speed) through a microphone array, records the user's touch behavior (such as strength and position) through a flexible capacitive touch sensor, and monitors the user's activity trajectory (such as relative distance and posture change) through a distance sensor and an IMU (Inertial Measurement Unit) sensor. These data are transmitted to the cloud in real time via a 4G (4Generation, fourth generation mobile communication technology) communication module or Wi-Fi (Wireless Fidelity, wireless fidelity technology). Furthermore, in this application, communication between the robot side and the cloud can also be carried out through 5G (5Generation, fifth generation mobile communication technology) or 6G (6Generation, sixth generation mobile communication technology). After parsing the data packet, the cloud side extracts the voice signal, touch parameters and behavior characteristics to form structured target user data.

[0069] This embodiment performs sentiment analysis in step S20, calling a preset sentiment analysis model in the cloud and inputting the target user's data into the model. For example, the voice signal is converted into a vector using an acoustic feature extraction layer (such as a mel-spectrogram), and then the emotion category is identified using a convolutional neural network. Touch data is combined with time series analysis to determine user emotional fluctuations. The model outputs an emotional state label.

[0070] In this embodiment, a target profile is constructed in step S30, where the current target user data is compared with historical user data in the cloud. A clustering algorithm is used to identify the user's long-term behavior patterns, and a decision tree model is used to extract key features to ultimately generate a target profile containing the user's basic attributes, interest tags, and emotional thresholds.

[0071] This embodiment generates a target interaction strategy through step S40. The cloud combines the emotional state with the target portrait and calls the preset strategy generation module. For example, if the emotional state is "sad" and the target portrait shows that the user prefers a soft tone, a speech synthesis text "I'm here with you" is generated, and the OLED (Organic Light-Emitting Diode) on the robot side is triggered to display a crying expression, while the motor controls the arm to pat the user's shoulder. Furthermore, the display screen on the robot side can also be an LCD (Liquid Crystal Display) or AMOLED (Active-matrix organic light-emitting diode). Strategy generation relies on a rule engine and a natural language generation model to ensure that the response meets user needs.

[0072] In this embodiment, the interaction strategy is sent in step S50. The cloud encapsulates the generated voice text, expression instructions and action parameters into JSON (JavaScript Object Notation).

[0073] The text is transmitted to the robot via the 4G communication module. After the robot parses the data packet, the speech synthesis module converts the text into speech signals, the OLED display updates the expression image, and the actuator drives the robotic arm to complete the specified action, realizing multimodal interaction with the user.

[0074] Exemplary, reference Figure 2 , Figure 2 The workflow of the interactive scenario of this embodiment is demonstrated. Users interact with the toy through operations such as voice wake-up, voice interaction, touch and shaking. The input signal is transmitted to the MCU (Microcontroller Unit) on the PCB (Printed Circuit Board) through an external module (such as a touch panel, IMU). After the MCU processes the signal, it communicates with the cloud service through the 4G module. The cloud-based large model combines the user portrait and generates response content (such as image stream, user portrait) according to the data type (such as voice, picture / video). It is returned to the MCU via the 4G module, and the interaction result is finally displayed on the OLED screen.

[0075] Furthermore, this embodiment can monitor physiological data. A heart rate sensor and temperature detection module can be added to the robot to collect the user's physiological state. This data can be integrated with voice and touch data and then fed into a sentiment analysis model to improve the accuracy of emotion recognition. Furthermore, by integrating augmented reality technology, a micro-projector can be deployed on the robot to project virtual scenes based on interaction strategies, enhancing the sense of realism in emotional companionship.

[0076] Furthermore, this embodiment can be applied to multi-user scenarios. When the motion data collected by the camera includes motion data of multiple users, the cloud will use the target detection model to identify the number of users in the picture in real time and track each user independently. The target user is determined in combination with the user's historical user data. If the target user cannot be determined, an independent interaction strategy is generated for each user. If there is a conflict in the actions of multiple users, the system generates a final strategy through a preset priority rule (such as "the latest instruction overwrites the old instruction") or a voting mechanism (such as counting the intentions of the majority of users).

[0077] In a feasible implementation manner, step S10 further includes steps A10 to A30:

[0078] Step A10: decrypt the target user data to obtain plaintext user data, wherein the target user data is encrypted data;

[0079] Step A20: removing personal identification information from the plaintext user data to obtain anonymous user data, wherein the anonymous user data is used for sentiment analysis using a preset analysis model;

[0080] Step A30: Update the anonymous user data into the historical user data.

[0081] It should be noted that, in this embodiment, plaintext user data is the original data obtained after decrypting the target user data through a symmetric encryption algorithm or an asymmetric algorithm deployed in the cloud, retaining complete user behavior characteristics; anonymous user data is desensitized data formed by removing personal identity information (such as name, ID number, unique device ID (Identifier)) in plaintext user data, which is used for subsequent sentiment analysis and user portrait construction; historical user data is the accumulated collection of anonymous user data in the cloud database, which contains records of users' long-term behavior patterns and emotional tendencies.

[0082] In this embodiment, after receiving the encrypted target user data from the robot, the cloud invokes a pre-defined decryption module. This module matches the key identifier in the encrypted data packet, extracts the corresponding decryption key from the key management repository, and performs a byte-by-byte decryption operation on the encrypted data. For example, if the target user data is an encrypted voice signal, the decrypted plaintext user data contains the original voice's sampling rate, spectral characteristics, and timestamp information.

[0083] Next, a desensitizing algorithm module (such as a regular expression matching or natural language processing model) runs in the cloud to identify and replace sensitive fields in the plaintext user data. For example, if the plaintext data contains "User Name: Zhang San" or "Device ID: 123456," the algorithm will replace "Zhang San" with "User A" and "123456" with "Device X." After desensitization, the anonymous user data retains only behavioral characteristics and non-sensitive attributes.

[0084] The cloud then writes the anonymous user data to the cloud database in time series and associates it with historical user data. For example, if the anonymous user data includes "User A touched the robot's head three times at 10:00 AM on May 29, 2025," the system merges this record with the historical data, "User A touched the robot's head five times at 3:30 PM on May 28, 2025," to form a continuous behavioral pattern analysis sample. Updates are performed incrementally, ensuring the timeliness and integrity of historical data.

[0085] Furthermore, in medical rehabilitation scenarios, physiological status fields (such as heart rate and body temperature) can be added to historical user data, and a multi-dimensional health portrait can be constructed by combining anonymized behavioral data, so that the robot can adjust the intensity of rehabilitation training according to the user's physiological and emotional state.

[0086] This embodiment ensures the security of user data through the decryption, desensitization and data update processes, and provides a reliable data foundation for sentiment analysis and personalized interaction.

[0087] In a feasible implementation manner, step S20 further includes steps B10 to B40:

[0088] Step B10: extracting features from the target user data to obtain multiple sets of feature parameters;

[0089] Step B20, fusing the feature parameters to obtain a joint emotion feature vector;

[0090] Step B30, inputting the joint emotion feature vector into a preset emotion classification model;

[0091] Step B40: Analyze and process the joint emotion feature vector through the emotion classification model to obtain the emotional state of the target user.

[0092] It should be noted that, in this embodiment, feature parameters refer to quantitative indicators extracted from target user data through algorithms; the joint emotion feature vector is a high-dimensional vector obtained by fusing multiple sets of feature parameters through weighted averaging or neural networks, which integrates the emotion correlation information of various types of data bureaus; the emotion classification model is a multi-classification neural network based on deep learning training, and its output result is an emotion state label in the form of a probability distribution; the emotional state of the target user is the emotion label with the highest probability in the output result of the emotion classification model, which is used to guide the generation of subsequent interaction strategies.

[0093] First, the cloud performs feature extraction on the touch data, distance data, posture data, location data, motion data, voice data, and external user data in the anonymous user data, forming multiple independent sets of feature parameters. Specifically, the touch data type includes character position, force, duration, and sliding trajectory. The extraction methods include sliding window statistical analysis and trajectory feature extraction. The sliding window statistical analysis calculates the force value by sliding within a fixed time window, extracting the force mean, variance, and peak value. The trajectory feature extraction uses a Kalman filter to smooth the trajectory data and calculate the trajectory velocity, acceleration peak, and trajectory directional stability. The output feature vector contains force statistics, trajectory dynamic parameters, and character duration distribution. The distance data type includes the change in the physical distance between the user and the device. The methods include statistical description and time series pattern analysis. The statistical description calculates the mean, variance, and extreme value of the distance. The time series pattern analysis uses a sliding window to extract the distance change rate and periodic fluctuation characteristics. The output feature vector contains distance distribution statistics, dynamic change rate, and periodic indicators. Posture data types include the coordinates of key points on the human body and the trajectory of posture changes. Methods include key point coordinate analysis and posture dynamic feature extraction. Key point coordinate analysis uses a CNN (Convolutional Neural Network) model to extract key point coordinates and calculate the relative angles between adjacent key points. Posture dynamic feature extraction uses an LSTM (Long Short-Term Memory) network to analyze the rate of change of key points over time and extract the frequency and stability of posture transitions. The output feature vector contains the key point coordinate sequence, angle distribution, and dynamic change pattern. Position data types include the user's absolute or relative position in space. Methods include spatial distribution statistics and trajectory pattern recognition. Spatial distribution statistics calculate the mean, variance, and hotspot distribution of position coordinates. Trajectory pattern recognition uses the HOG (Histogram of Oriented Gradients) feature to extract the directional histogram of the position trajectory and combines it with a Kalman filter to extract the curvature of the smoothed trajectory. The output feature vector contains position distribution statistics, trajectory directional features, and spatial activity patterns. The action data type includes the specific actions performed by the user; the methods include action pattern recognition and time series modeling; action pattern recognition extracts action amplitude, speed and acceleration based on a sliding window; time series modeling uses an LSTM network to analyze the time series characteristics of the action, extract action periodicity and energy distribution; the output feature vector contains action dynamic parameters, time series pattern and energy distribution characteristics.Speech data types include speech signals; methods include frequency domain analysis, time domain feature extraction, and emotion-related feature analysis. Frequency domain analysis uses short-time Fourier transforms to calculate mel-frequency spectra and extract MFCC (Mel-Frequency Cepstral Coefficients). Time domain features calculate speech rate, mean volume, and fundamental frequency. Emotion-related features are combined with HMM (Hidden Markov Model) models to extract speech prosodic features. The output feature vector includes frequency domain features, time domain statistics, and emotion-related parameters. External data types include third-party system data; methods include behavioral pattern analysis. Behavioral pattern analysis uses clustering algorithms to identify the periodicity of social media behavior. The output feature vector includes physiological indicator statistics, environmental parameter distributions, and behavioral periodicity characteristics. After all these features are extracted, the feature vectors of each data set are normalized and concatenated into structured data in a unified format according to data type, forming multiple independent sets of feature parameters.

[0094] The multiple sets of feature parameters are then fed into the feature fusion module. For example, the module uses a multi-head attention network to perform cross-modal correlation on touch feature vectors, distance feature vectors, posture feature vectors, position feature vectors, motion feature vectors, speech feature vectors, and external feature vectors. Attention weights are calculated and weighted summed to generate a joint emotion feature vector. This vector contains cross-modal information such as touch force mean, distance change rate, posture transition frequency, position hotspot distribution, motion energy distribution, speech intonation fluctuations, and external behavior periodicity.

[0095] The specific fusion process is as follows: touch and distance data fusion: the touch force mean and distance change rate are merged into a low-dimensional feature vector through splicing operation, and then cross-modal features are extracted through 1×1 convolution layer to reduce computational complexity; posture and position data fusion: pyramid pooling is used to extract multi-scale features between the posture key point coordinate sequence and the position direction histogram; action and speech data fusion: the weights of action energy distribution and speech rhythm features are dynamically adjusted through the attention mechanism; external data and global feature fusion: the entropy method is used to calculate the correlation between external behavior periodicity and global features to generate a weighted fusion vector.

[0096] The joint sentiment feature vector is then used as input for nonlinear mapping through a pre-set sentiment classification model. The sentiment classification model uses a fully connected network with two hidden layers, a ReLU activation function, and a Softmax output layer that generates a probability distribution. For example, the network outputs the value [0.15, 0.65, 0.20] for the input joint feature vector, corresponding to the emotion categories "joy," "sadness," and "neutral."

[0097] Finally, in the probability distribution output by the model, the emotion category with the highest probability value is selected as the emotional state of the target user.

[0098] Furthermore, this embodiment supports a multimodal incremental learning mechanism, adding a dynamic weight adjustment module to the sentiment classification model. For example, if a user's emotional state deviates from the predicted result after three consecutive interactions, the system automatically triggers an incremental training process, retraining the model by weighting the latest data with historical data using a time decay factor (such as exponential decay).

[0099] This embodiment significantly improves the robustness and accuracy of sentiment analysis through hierarchical feature extraction and cross-modal fusion. The construction of a joint feature vector effectively captures the multi-dimensional correlations of user behavior, ultimately enabling more accurate emotion recognition and the generation of personalized interaction strategies, enhancing the continuity and emotional resonance of the user's companionship experience.

[0100] In a feasible implementation manner, step S30 further includes steps C10 to C30:

[0101] Step C10: classify the historical user data of the target user and extract feature vectors for different categories of historical user data;

[0102] Step C20, labeling the feature vector to obtain a portrait label;

[0103] Step C30: Use a preset portrait construction algorithm to construct a portrait based on the portrait label to generate a target portrait.

[0104] It should be noted that, in this embodiment, the historical user data of the target user refers to an anonymous user data set stored in the cloud, including voice, touch and behavior records; classification refers to dividing the historical user data according to the data types of touch data, distance data, posture data, position data, motion data, voice data and external user data to form independent data groups; the feature vector is a quantitative indicator extracted from the classified data by the algorithm, such as the Mel spectrum of voice data and the mean force of touch data; labeling processing is to convert the feature vector into an interpretable label to describe the user's behavior pattern; the portrait construction algorithm is a model based on machine learning, whose input is labeled data and output is the target portrait, that is, a structured data set that integrates the user's basic information, interest tags, emotional tendencies and behavior patterns.

[0105] This embodiment first classifies historical user data. The cloud retrieves the historical user data of the target user from the database, including all anonymous user data sets with a preset time threshold. The time threshold can be customized. The cloud divides the data into touch data, distance data group, posture data group, position data group, action data group, voice data group and external user data group through data type identifiers. For example, the voice data group contains multiple voice files and corresponding intonation and speech speed parameters, the touch data group contains touch position, force and duration records, and the posture data group contains posture activity trajectory and posture change data. The classification operation is completed through database classification query or regular expression matching data.

[0106] Next, a feature extraction algorithm is executed on each set of data. Refer to the description of step C10 above and will not be repeated here. The final output feature vector set will serve as input for subsequent labeling processing to generate user profile labels.

[0107] Next, the feature vector is labeled. The feature vector is input into a pre-trained labeling model to obtain the portrait label. The labeling model is trained based on the mapping rules between a large amount of multimodal data and labels.

[0108] Finally, the labeled data is fed into a profile-building algorithm. In this example, K-means (a clustering algorithm) is used. The algorithm clusters tags based on similarity. The profile-building algorithm calculates the co-occurrence frequency and weights of tags, ultimately outputting a structured target profile that includes the user's basic attributes, interest tags, sentiment thresholds, and behavioral patterns.

[0109] Furthermore, this embodiment can introduce an incremental learning mechanism, adding a dynamic weight adjustment module to the portrait construction algorithm. For example, if the labeling results change significantly after three consecutive user interactions, the system automatically triggers the incremental training process, retraining the model by weighting the latest label with the historical label according to the time decay factor.

[0110] This embodiment significantly improves the efficiency and accuracy of user portrait construction through hierarchical classification and labeling processing, enabling the robot to accurately match user needs and improve the personalization level and interaction quality of emotional companionship.

[0111] In a feasible implementation manner, step S40 further includes steps D10 to D30:

[0112] Step D10, searching a preset interaction strategy library for an interaction strategy that matches the target profile and emotional state;

[0113] Step D20: if there is an interaction strategy search result, determining a target interaction strategy based on the interaction strategy search result;

[0114] Step D30: If there is no interaction strategy search result, a target interaction strategy is generated according to the target portrait and emotional state.

[0115] It should be noted that, in this embodiment, the interaction strategy library refers to a set of preset interaction strategies stored in the cloud, which includes combinations of voice, expression and action instructions that match different user portraits and emotional states; the target portrait refers to a set of user features constructed through historical user data, covering basic information, interest tags, emotional tendencies and behavior patterns; the target interaction strategy is a specific response plan after matching the target portrait and emotional state, including speech synthesis text, expression display rules and action control parameters; generating a target interaction strategy means calling a rule engine or generating a model to create a new strategy based on the target portrait and emotional state when there is no matching strategy.

[0116] In this embodiment, the cloud first takes the target profile (e.g., "low stimulation preference user") and emotional state (e.g., "sadness") as input and calls the search module of the interaction strategy library. The search module selects pre-stored interaction strategies in the library by matching keywords (e.g., "low stimulation preference + sadness") or calculating semantic similarity.

[0117] If a matching strategy is retrieved, the system directly extracts it as the target interaction strategy. For example, if the library contains a strategy for "low-stimulus preference users playing soft music and displaying a crying expression when in a sad mood," this strategy is returned as a search result, encapsulated in a JSON format data packet (containing the voice text "I'm here with you," the expression instruction "cry," and the action parameter "arm pat") and sent to the robot for execution. If there are multiple interaction search results, the accuracy of the strategy is ensured by sorting the strategy priorities. The priority can be determined by calculating the match between the search results and the user's target profile and emotional state, with the highest match being given priority.

[0118] If no matching strategy exists, the system calls the generation module. This module combines the target profile and emotional state, generates speech text using a natural language generation model, and then calls expression generation rules and action control algorithms to ultimately create a new strategy.

[0119] This embodiment significantly improves the matching efficiency and personalization level of interactive strategies through hierarchical retrieval and dynamic generation mechanisms. The preset strategy library ensures rapid response to common scenarios, while the generation module fills the gaps in rare situations, ultimately achieving a more natural and emotionally relevant interactive experience.

[0120] In a feasible implementation manner, step S50 further includes steps E10 to E40:

[0121] Step E10, receiving feedback data sent by the robot end;

[0122] Step E20, classifying the feedback data;

[0123] Step E30: Compare and analyze the classified feedback data with preset historical feedback data to determine the change trend of the feedback data;

[0124] Step E40: Adjust the weight parameters in the analysis model according to the change trend.

[0125] It should be noted that in this embodiment, feedback data refers to the feedback provided by all sensors on the robot side after the robot side executes the target interaction strategy, such as voice evaluation collected by the microphone array, user touch force recorded by the flexible capacitive touch sensor (such as tapping to indicate satisfaction), and user activity trajectory monitored by the camera (such as actively approaching to indicate interest). These feedbacks include explicit feedback, that is, some relatively obvious feedback, such as the voice command "The answer just now was great", and implicit feedback, such as the interaction time. A short time indicates that the user is not interested, and a long time indicates that the user is interested. Classification refers to dividing the received feedback data into independent data groups according to the data type. Historical feedback data is divided into two types: one is immediate historical feedback data, that is, historical feedback data for the current interaction scenario, such as feedback data after the previous interaction command is executed; the other is long-term historical feedback data, such as historical feedback data in the last month or year. Therefore, there are also two types of analysis for historical feedback data: one is immediate historical feedback data analysis, and the other is long-term historical feedback data analysis. The changing trend is a statistical feature calculated by comparing the classified feedback data with the historical feedback data; the weight parameter is a numerical value used to adjust the importance of the feature parameter in the sentiment analysis model, and its adjustment is based on the changing trend of the feedback data.

[0126] Exemplarily, the process of analyzing instant historical feedback data specifically includes: dividing the current feedback data into feedback groups of different data types and comparing them with the instant historical feedback data, such as conducting a longitudinal year-on-year analysis of the current voice emotion score and the voice emotion score of the last interaction instruction, comparing the current average touch force with the average touch force of the last time, and calculating the rate of change; comparing the current interaction time with the last interaction time to determine the growth or decline trend; if the voice emotion score increases by more than 20%, mark it as "high satisfaction"; if the touch force decreases and the interaction time is extended by more than 30%, mark it as "increased interest"; dynamically adjust the voice feature weight, touch force weight and trajectory feature weight according to the change trend, for example, the voice feature weight is increased by 0.1, the touch force weight is reduced by 0.05, and the trajectory feature weight is increased by 0.05.

[0127] Exemplarily, the process of analyzing long-term historical feedback data specifically includes: dividing the long-term historical feedback data into feedback groups of different data types and comparing them with the long-term historical feedback data; calculating the difference between the current voice emotion score and the average voice emotion score of the past 30 days, if the current score is 15% higher than the average, marking it as "long-term satisfaction improvement"; calculating the difference between the current average touch force and the average touch force of the past 30 days, if the current force decreases by 10%, marking it as "touch preference change"; calculating the difference between the current interaction time and the average interaction time of the past 30 days, if the time increases by 20%, marking it as "long-term interest enhancement"; if the voice emotion score shows a continuous upward trend, marking it as "voice interaction optimization is effective"; if the touch force shows a downward trend and the interaction time shows an upward trend, marking it as "low-intervention interaction preference"; adjusting the model parameters according to the long-term trend, for example, fixing the voice feature weight to 0.4, gradually reducing the touch force weight to 0.1, and increasing the trajectory feature weight to 0.3.

[0128] This embodiment significantly improves the adaptive capabilities of the sentiment analysis model through layered feedback data processing and dynamic weight adjustment. The trend-based weight optimization mechanism enables the robot to accurately match changing user needs, ultimately achieving a more natural and emotionally engaging interactive experience.

[0129] Based on the embodiment 1 of the present application, in the embodiment 2 of the present application, the same or similar contents as those in the above embodiment 1 can be referred to the above introduction, and no further details will be given later. Figure 3 , Figure 3 This is a flow chart of a method for generating an interaction strategy based on user data applied to a robot side in this application. The method for generating an interaction strategy based on user data is applied to a robot side, and the robot side is connected to a preset cloud communication. The method for generating an interaction strategy based on user data includes steps S11 to S31:

[0130] Step S11, collecting target user data of the target user;

[0131] Step S21: Send the target user data to the cloud, wherein the cloud is used to receive the target user data sent by the robot end, determine the target user data of the target user data; perform sentiment analysis on the target user data using a preset analysis model to obtain the emotional state of the target user; construct a target profile based on the target user's historical user data, wherein the historical user data includes historically received target user data; generate a target interaction strategy based on the target profile and the emotional state, and send the target interaction strategy to the robot end;

[0132] Step S31: receiving the target interaction strategy sent by the cloud and executing the target interaction strategy.

[0133] It should be noted that collecting the target user data of the target user means that the robot side obtains the original information such as voice intonation, touch strength, activity trajectory, etc. through hardware such as microphone arrays, flexible capacitive touch sensors, and cameras; sending the target user data to the cloud means transmitting data to the cloud server through a 4G communication module or Wi-Fi. Furthermore, in this application, data can also be transmitted to the cloud server through a new generation of mobile networks such as 5G or 6G. After receiving the data, the cloud server filters out the valid data and identifies it as the target user data of the current interaction scenario; the preset analysis model is a pre-trained multimodal emotion recognition system, which compares the speech feature vector with the The tag library matches the emotional state; the historical user data of the target user is a collection of anonymous user data stored in the cloud, which includes voice, touch and behavior records collected in the past; the target portrait is a structured data set generated by classifying and extracting feature vectors from the historical user data, and then labeling and clustering it; the emotional state is the quantitative result output by the analysis model, which represents the user's immediate emotional tendency; the target interaction strategy is a specific behavioral instruction generated by combining the target portrait and emotional state, which is used to guide the interactive actions of the robot side; executing the target interaction strategy means that after the robot side receives the instructions sent by the cloud, it drives the voice module, touch feedback unit and motion controller to complete the corresponding operation.

[0134] The robot's microphone array captures the user's voice and intonation, a flexible capacitive touch sensor records touch force, and a camera tracks the user's movements. The sensors encapsulate the raw data in JSON format, including a timestamp, sensor type, and value. This JSON data is then transmitted to a cloud server.

[0135] After parsing the data packets, the cloud-based system filters the target user data for the current interaction scenario by time window, removing outliers. The cloud-based system then invokes a pre-configured analysis model, consisting of an image encoder and a text encoder based on the CLIP (Contrastive Language-Image Pre-training) architecture. The speech data is converted into a mel-spectrogram feature vector, and its cosine similarity is calculated with text vectors such as "satisfied" and "confused" in the tag library, outputting the emotional state as "satisfied." The cloud-based system retrieves the target user's historical user data from a database and categorizes it into touch and voice groups based on data type. For the touch group, it extracts the mean force feature and labels it to generate the label "prefers gentle touch." For the voice group, it extracts speech rate features and generates the label "standard speaking rate user." After clustering, the labels are integrated into a structured target profile, which includes fields such as interest tags and emotion thresholds. The cloud-based system combines the emotional state with the interest tags in the target profile, triggering rules in the pre-configured policy library to generate a target interaction strategy. This strategy is then sent to the robot via the MQTT (Message Queuing Telemetry Transport) protocol.

[0136] After receiving the target interaction strategy, the robot performs display interaction (which can display expressions, images, or videos), voice interaction, or action interaction according to the target interaction strategy.

[0137] Furthermore, this embodiment can also be applied to educational scenarios. When this embodiment is switched to the teaching question-and-answer mode, teaching questions and answers can be conducted based on the collected target user data. For example, when the target user data is the posture information of the robot, the teaching information with the highest correlation with the posture information can be searched in the external network, and voice interaction instructions can be generated and sent to the robot for playback. Specifically, when the robot's hand points to the sky, astronomical knowledge can be introduced to the user through dialogue. In addition, the target user data can also include collected voice data (such as questions raised by the user), etc. In addition, when the robot is conducting teaching questions and answers, it can use the posture movements of plush toys to cooperate with language teaching, demonstrate body language corresponding to different tones and intonations, assist children in learning language expression and communication skills, and organically integrate entertainment and education.

[0138] In a feasible implementation manner, step S11 further includes steps T10 to T70:

[0139] Step T10, controlling the touch sensor array to collect touch actions of the target user and generate touch data; and / or,

[0140] Step T20, controlling the infrared sensor to collect the relative distance between the target user and the robot and generate distance data; and / or,

[0141] Step T30, controlling the IMU sensor to collect posture change information of the robot end and generate posture data; and / or,

[0142] Step T40, controlling the GPS (Global Positioning System) locator to collect the position information of the robot end and generate position data; and / or,

[0143] Step T50, controlling the camera to capture the target user's expression and action, and generating action data; and / or,

[0144] Step T60, controlling the microphone array to collect the target user's voice and generate voice data; and / or,

[0145] Step T70: Obtain the external device information bound to the target user, and obtain the external user data corresponding to the external device information, wherein the external user data includes user social data and schedule data.

[0146] It should be noted that the target user data is a set of multimodal information related to user behavior and environment obtained by the robot side through various sensors and external devices; touch data is a numerical set of the user's touch force and position recorded by the flexible capacitive touch sensor array; distance data is a numerical set of the relative distance between the target user and the robot calculated by the infrared sensor by emitting and receiving reflected light waves. Furthermore, distance data can also be measured by radar, lidar or ultrasonic sensor to make the measurement data more accurate. This embodiment only uses infrared sensors as an example and does not limit the specific type of sensor; posture data is the dynamic state information generated by the inertial measurement unit sensor detecting the three-dimensional acceleration and angular velocity of the robot side; position data is a numerical set of the geographical coordinates of the robot side calculated by the GPS locator through satellite signals; motion data is the structured information of user expressions and limb movement trajectories extracted through AI (Artificial Intelligence) image recognition algorithm; voice data is the user voice signal generated by the microphone array through sound wave collection and noise reduction processing; external user data is a set of user social platform account data and schedule data obtained through the binding device interface.

[0147] Data collection is carried out simultaneously without any particular order. The order of introduction here does not represent the order of collection.

[0148] The flexible capacitive touch sensor array deployed on the robot side monitors the user's touch behavior in real time. When the user taps the robot's arm, the capacitance change between the sensor electrodes is quantified into touch force value and contact coordinates to form touch data.

[0149] The infrared sensor calculates distance by emitting a grid of infrared light and detecting the reflected signal after it is blocked. For example, when the user stands 1.5 meters in front of the robot, the sensor outputs a distance data of 1500mm. This data is calculated using the formula of the time difference of the reflected light wave (distance = speed of light × time difference / 2).

[0150] The IMU sensor (including a three-axis accelerometer and gyroscope) detects the motion state of the robot. When the pitch angle of the robot head changes due to user touch, the accelerometer outputs a linear acceleration value and the gyroscope outputs an angular velocity value, which are then fused through the Kalman filter algorithm to generate posture data.

[0151] The GPS locator receives signals from at least four satellites and uses triangulation to calculate the longitude and latitude coordinates of the robot. The positioning accuracy is affected by satellite distribution and atmospheric attenuation, and the error range is usually within 3 meters.

[0152] The camera captures user movements, and the AI ​​image recognition algorithm extracts the coordinates of key points and combines them with timestamps to generate motion data. For example, the trajectory of a user's waving movement can be expressed as [(t1,X1,Y1),(t2,X2,Y2),...,(tn,Xn,Yn)].

[0153] The microphone array uses beamforming technology to isolate ambient noise and convert the user's voice signal into a digital audio stream.

[0154] Connect to the user's mobile phone via Bluetooth or Wi-Fi. Further, it can also connect to the user's mobile phone through mobile networks such as 4G, 5G, or 6G to communicate with the user's mobile phone, read the bound social platform (such as the number of WeChat friends and like records) and calendar applications (such as meeting time and location), and generate an external user data packet containing structured fields such as {"social friends": 500,"recent meetings": [{"time":"2025-06-0110:00","location":"Conference Room A"}]}.

[0155] This embodiment can introduce a multimodal feature fusion mechanism in the data collection stage and improve information utilization by establishing a data association model between sensors.

[0156] This embodiment significantly improves the robot's recognition accuracy and response flexibility for user intent through multi-dimensional data collection and correlation analysis. Furthermore, by integrating external user data, the adaptability of service scenarios is improved, effectively enhancing the naturalness and practicality of human-computer interaction.

[0157] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the method of generating interactive strategies based on user data in the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0158] Reference below Figure 4 , which shows a schematic diagram of the structure of a cloud device suitable for implementing an embodiment of the present application. The cloud device in the embodiment of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., as well as fixed terminals such as digital TVs, desktop computer servers, and cloud terminals. Figure 4 The cloud device shown is only an example and should not limit the functions and scope of use of the embodiments of the present application.

[0159] like Figure 4 As shown, the cloud device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory 1002 or programs loaded from a storage device 1003 into a random access memory 1004. Random access memory 1004 also stores various programs and data required for the operation of the cloud device. The processing device 1001, the read-only memory 1002, and the random access memory 1004 are connected to each other via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the cloud device to communicate with other devices wirelessly or wired to exchange data. Although the figure shows a cloud device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0160] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are performed.

[0161] The cloud device provided by this application utilizes the user data-based interaction strategy generation method of the aforementioned embodiment to address the technical issue of imbalanced interaction strategy generation based on user data. Compared to the prior art, the beneficial effects of the cloud device provided by this application are the same as those of the user data-based interaction strategy generation method provided by the aforementioned embodiment. Other technical features of this cloud device are the same as those disclosed in the aforementioned embodiment and are not further elaborated here.

[0162] Reference below Figure 5, which shows a schematic diagram of the structure of a robot device suitable for implementing the embodiments of the present application. The robot device in the embodiments of the present application may include but is not limited to intelligent machines such as companion robots, smart plush toys, etc. that can accompany users through interaction. Figure 5 The robot device shown is only an example and should not limit the functions and scope of use of the embodiments of the present application.

[0163] like Figure 5 As shown, the robotic device may include a processing device 2001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory 2002 or programs loaded from a storage device 2003 into a random access memory 2004. The random access memory 2004 also stores various programs and data required for the operation of the robotic device. The processing device 2001, the read-only memory 2002, and the random access memory 2004 are interconnected via a bus 2005. An input / output interface 2006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 2006: input devices 2007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 2008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 2003 including, for example, a magnetic tape, hard disk, etc.; and communication devices 2009. The communication device 2009 can allow the robotic device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a robotic device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have instead.

[0164] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 2003, or installed from a read-only memory 2002. When the computer program is executed by the processing device 2001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are performed.

[0165] The robotic device provided in this application utilizes the user data-based interaction strategy generation method described in the aforementioned embodiment, resolving the technical issue of unbalanced user data-based interaction strategy generation. Compared to the prior art, the robotic device provided in this application offers the same beneficial effects as the user data-based interaction strategy generation method described in the aforementioned embodiment. Other technical features of the robotic device are the same as those disclosed in the aforementioned embodiment and are not further elaborated upon here.

[0166] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0167] The above are only specific embodiments of the present application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0168] The present application provides a medium, which is a computer-readable storage medium having computer-readable program instructions (ie, a computer program) stored thereon, and the computer-readable program instructions are used to execute the method for generating an interaction strategy based on user data in the above embodiment.

[0169] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0170] The computer-readable storage medium may be included in the cloud device, or may exist independently without being installed in the cloud device.

[0171] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the cloud device, the cloud device:

[0172] receiving target user data sent by the robot end, determining target user data of the target user data, wherein the robot end is used to collect the target user data of the target user; and sending the target user data to the cloud;

[0173] Perform sentiment analysis on target user data through a preset analysis model to obtain the target user's emotional state;

[0174] Building a target profile based on the target user's historical user data, wherein the historical user data includes the target user data received historically;

[0175] Generate target interaction strategies based on target profile and emotional state;

[0176] The target interaction strategy is sent to the robot side, where the robot side is used to receive the target interaction strategy sent by the cloud and execute the target interaction strategy.

[0177] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0178] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0179] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0180] The computer-readable storage medium provided in this application stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned method for generating interaction strategies based on user data. This computer-readable storage medium can address the technical issue of unbalanced generation of interaction strategies based on user data. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the method for generating interaction strategies based on user data provided in the aforementioned embodiments, and are not further elaborated here.

[0181] The present application also provides a product, which is a computer program product, including a computer program. When the computer program is executed by a processor, it implements the steps of the above-mentioned method for generating an interaction strategy based on user data.

[0182] The computer program product provided in this application can solve the technical problem of imbalanced generation of interaction strategies based on user data. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the method for generating interaction strategies based on user data provided in the above embodiment, and will not be elaborated here.

[0183] The above are only some embodiments of the present application and are not intended to limit the patent scope of the present application. All equivalent structural transformations made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A method for generating an interaction strategy based on user data, characterized in that: Applied to the cloud, the cloud is connected to the preset robot terminal for communication, and the method for generating an interaction strategy based on user data includes: receiving target user data sent by the robot end, and determining target user data of the target user data, wherein the robot end is used to collect the target user data of the target user and send the target user data to the cloud; Performing sentiment analysis on the target user data using a preset analysis model to obtain the target user's emotional state; Building a target profile based on historical user data of the target user, wherein the historical user data includes historically received target user data; generating a target interaction strategy based on the target portrait and the emotional state; The target interaction strategy is sent to the robot end, wherein the robot end is used to receive the target interaction strategy sent by the cloud and execute the target interaction strategy.

2. The method for generating an interaction strategy based on user data according to claim 1, wherein: After the steps of receiving the target user data sent by the robot end and determining the target user of the target user data, the method further includes: Decrypting the target user data to obtain plaintext user data, wherein the target user data is encrypted data; Removing personal identification information from the plaintext user data to obtain anonymous user data, wherein the anonymous user data is used for sentiment analysis using a preset analysis model; The anonymous user data is updated into the historical user data.

3. The method for generating an interaction strategy based on user data according to claim 1, wherein: The step of performing sentiment analysis on the target user data to obtain the target user's emotional state includes: Performing feature extraction on the target user data to obtain multiple sets of feature parameters; Fusing the feature parameters to obtain a joint emotion feature vector; The combined emotional feature vector is input into a preset emotional classification model, and the combined emotional feature vector is analyzed and processed by the emotional classification model to obtain the emotional state of the target user.

4. The method for generating an interaction strategy based on user data according to claim 1, wherein: The step of constructing a target profile based on the historical user data of the target user includes: Classifying the historical user data of the target user and extracting feature vectors for different categories of historical user data; Performing labeling processing on the feature vector to obtain a portrait label; A preset portrait construction algorithm is used to construct a portrait based on the portrait tags to generate a target portrait.

5. The method for generating an interaction strategy based on user data according to claim 1, wherein: The step of generating a target interaction strategy based on the target portrait and the emotional state includes: Retrieving an interaction strategy that matches the target profile and the emotional state from a preset interaction strategy library; If there is an interaction strategy retrieval result, determining a target interaction strategy according to the interaction strategy retrieval result; If the interaction strategy retrieval result does not exist, a target interaction strategy is generated according to the target portrait and the emotional state.

6. The method for generating an interaction strategy based on user data according to claim 1, wherein: After the step of sending the target interaction strategy to the robot side, the following steps are included: Receiving feedback data sent by the robot end; classifying the feedback data; Comparing and analyzing the classified feedback data with preset historical feedback data to determine the change trend of the feedback data; According to the change trend, the weight parameters in the analysis model are adjusted.

7. A method for generating an interaction strategy based on user data, characterized in that: Applied to a robot side, the robot side is connected to a preset cloud communication, and the method for generating an interaction strategy based on user data includes: Collect target user data of target users; The target user data is sent to the cloud, wherein the cloud is used to receive the target user data sent by the robot end, determine the target user data of the target user data, perform sentiment analysis on the target user data through a preset analysis model to obtain the emotional state of the target user, build a target profile based on the historical user data of the target user, wherein the historical user data includes historically received target user data, generate a target interaction strategy based on the target profile and the emotional state, and send the target interaction strategy to the robot end; Receive the target interaction strategy sent by the cloud, and execute the target interaction strategy.

8. The method for generating an interaction strategy based on user data according to claim 7, wherein: The target user data includes at least one of touch data, distance data, posture data, position data, action data, voice data, and external user data. The step of collecting the target user data of the target user includes: controlling the touch sensor array to collect the touch action of the target user and generate the touch data; and / or, controlling the infrared sensor to collect the relative distance between the target user and the robot and generate the distance data; and / or, Controlling the IMU sensor to collect the posture change information of the robot end and generate the posture data; and / or, Controlling the GPS locator to collect the position information of the robot terminal and generate the position data; and / or, controlling a camera to capture the target user's expression and action, and generating the action data; and / or, controlling the microphone array to collect the target user's voice and generate the voice data; and / or, Acquire external device information bound to the target user, and acquire external user data corresponding to the external device information, wherein the external user data includes user social data and schedule data.

9. A cloud device, characterized in that: The cloud device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the method for generating an interaction strategy based on user data according to any one of claims 1 to 6.

10. A robotic device, characterized in that: The robot device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the method for generating an interaction strategy based on user data according to any one of claims 7 to 8.

Images