Keyboard area customization method, system and medium based on user behavior analysis

Abstract

The present application relates to the technical field of user behavior analysis, and discloses a keyboard area customization method, system and medium based on user behavior analysis. The method includes: real-time data collection of user keyboard operation behavior to obtain the original user input data set; preprocessing and feature extraction to obtain the user input feature vector; clustering and classification and building a user behavior pattern classification model; regional division of the keyboard to obtain multiple keyboard areas and calculate the importance weight of each keyboard area to obtain a keyboard area importance scoring matrix; keyboard layout optimization according to the keyboard area importance scoring matrix to obtain a personalized keyboard layout plan; continuous collection of usage data and layout optimization of the personalized keyboard layout plan to obtain layout optimization iteration parameters. The present application realizes highly personalized keyboard layout customization based on user behavior, thereby greatly improving the user's input efficiency and comfort.

Description

Technical Field

[0001] The present application relates to the technical field of user behavior analysis, and in particular to a keyboard area customization method, system and medium based on user behavior analysis. Background Art

[0002] With the rapid development of information technology, keyboards are an important input device for human-computer interaction, and their efficiency directly affects the user's work efficiency and experience. Traditional keyboard layout design is mainly based on a fixed arrangement, which is difficult to adapt to the personalized needs and usage habits of different users.

[0003] Although some customizable keyboard products have appeared on the market, most of them still require manual adjustment by users and lack intelligent and automated features. This not only increases the user's learning cost, but also fails to adapt to the dynamic changes in user habits in a timely manner. In addition, existing keyboard customization methods often only consider a single factor, such as keystroke frequency or finger movement distance, while ignoring the complexity and diversity of user behavior, resulting in limited optimization effects. Summary of the invention

[0004] The present application provides a keyboard area customization method, system and medium based on user behavior analysis, which are used to achieve highly personalized keyboard layout customization based on user behavior, thereby greatly improving the user's input efficiency and comfort.

[0005] In a first aspect, the present application provides a keyboard area customization method based on user behavior analysis, and the keyboard area customization method based on user behavior analysis includes:

[0006] Collect data of user keyboard operation behaviors in real time to obtain the original user input data set;

[0007] Preprocessing and feature extraction are performed on the original user input data set to obtain a user input feature vector;

[0008] Clustering and classifying the user input feature vectors to build a user behavior pattern classification model;

[0009] Dividing the keyboard into regions to obtain a plurality of keyboard regions, and calculating the importance weight of each keyboard region based on the user behavior pattern classification model to obtain a keyboard region importance scoring matrix;

[0010] Optimizing the layout of the keyboard according to the keyboard area importance scoring matrix to obtain a personalized keyboard layout solution;

[0011] The personalized keyboard layout scheme is subjected to continuous usage data collection and layout optimization to obtain layout optimization iteration parameters.

[0012] In a second aspect, the present application provides a keyboard area customization system based on user behavior analysis, and the keyboard area customization system based on user behavior analysis includes:

[0013] The acquisition module is used to collect real-time data of user keyboard operation behavior to obtain the original user input data set;

[0014] An extraction module, used for preprocessing and feature extraction of the original user input data set to obtain a user input feature vector;

[0015] A construction module, used to cluster and classify the user input feature vectors to construct a user behavior pattern classification model;

[0016] A calculation module, used to divide the keyboard into regions to obtain a plurality of keyboard regions, and calculate the importance weight of each keyboard region based on the user behavior pattern classification model to obtain a keyboard region importance scoring matrix;

[0017] An optimization module, used to optimize the layout of the keyboard according to the keyboard area importance scoring matrix to obtain a personalized keyboard layout solution;

[0018] The iteration module is used to continuously collect usage data and optimize the layout of the personalized keyboard layout solution to obtain layout optimization iteration parameters.

[0019] The third aspect of the present application provides a keyboard area customization device based on user behavior analysis, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the keyboard area customization device based on user behavior analysis executes the above-mentioned keyboard area customization method based on user behavior analysis.

[0020] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned keyboard area customization method based on user behavior analysis.

[0021] In the technical solution provided by the present application, a comprehensive original user input data set is constructed by real-time monitoring of user key operations, collecting multi-dimensional data, and combining device environment parameters. User input features are extracted from multiple angles such as frequency, speed, combination mode, rhythm, etc., and the user's input behavior pattern is fully characterized. A hierarchical user behavior pattern classification model is constructed in combination with K-means clustering, which can accurately identify and classify different user input patterns. Through grid processing and Voronoi diagram optimization, flexible division of keyboard areas is achieved, and combined with user behavior patterns and entropy analysis, regional importance scores that are more in line with user usage habits are obtained. Multiple candidate solutions are generated, and through efficiency scoring and local optimization, the optimal layout solution that takes into account efficiency and ergonomics is quickly obtained. Through continuous data collection and learning curve analysis, dynamic updating and optimization of the layout solution are achieved, so that the keyboard layout can continuously adapt to changes in user habits. The whole process fully considers factors such as the user's personal usage habits, device environment, and application scenarios, and realizes highly personalized keyboard layout customization. In the optimization process, input efficiency and ergonomic factors are considered at the same time, which can not only improve the user's input speed, but also reduce fatigue and improve the comfort of long-term use. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0023] Figure 1 A schematic diagram of an embodiment of a keyboard area customization method based on user behavior analysis in an embodiment of the present application;

[0024] Figure 2 A schematic diagram of an embodiment of a keyboard area customization system based on user behavior analysis in an embodiment of the present application. DETAILED DESCRIPTION

[0025] Embodiments of the present application provide a keyboard area customization method, system and medium based on user behavior analysis. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of the keyboard area customization method based on user behavior analysis includes:

[0027] Step S101: collect real-time data of user keyboard operation behavior to obtain an original user input data set;

[0028] It is understandable that the execution subject of the present application may be a keyboard area customization system based on user behavior analysis, or a terminal or a server, which is not specifically limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0029] Specifically, the user's key operation is monitored in real time, and the key event sequence is captured by each key operation of the user, and the key timestamp information is extracted according to the key event sequence to obtain the key time sequence. The key event sequence is analyzed to obtain the key position coordinate sequence, and the time difference between adjacent key events is calculated to obtain the user's key speed and form a key speed sequence. The key event sequence is statistically analyzed to obtain the key frequency distribution matrix, and the key events are marked in combination with the application information currently used by the user to generate a key event sequence with scene labels. At the same time, the user's device information is collected to obtain the device environment parameter set. The device environment parameters include screen resolution, operating system type, device model, etc., which may affect the user's key behavior. The device environment parameters are combined with the key event sequence to generate a more representative segmented input behavior feature sequence. The segmented input behavior feature sequence not only reflects the user's input behavior characteristics in different time periods, but also comprehensively considers the impact of the device environment. By combining the segmented input behavior feature sequence with the key event sequence with scene labels, a complete original user input data set is constructed. This dataset integrates the user's keystroke time series, keystroke speed series, keystroke frequency distribution matrix, scene labels, and device environment parameters, comprehensively reflecting the user's keystroke behavior and the environment in which he or she is located.

[0030] Step S102: preprocessing and feature extraction of the original user input data set to obtain a user input feature vector;

[0031] Specifically, outlier detection is performed on the original user input data set to identify and remove abnormal data to obtain a cleaned data set. The usage frequency of each key is calculated based on the cleaned data set to generate a key frequency vector, which reflects the usage frequency distribution of each key on the keyboard. A sliding time window analysis is performed on the cleaned data set to capture the time series feature matrix by analyzing the changes in user input behavior within a fixed time window. This matrix reflects the input characteristics of the user in different time periods. Combined with this matrix, the user's average input speed and speed change rate are calculated to obtain a speed feature vector. The average input speed reflects the overall rhythm of the user's typing, while the speed change rate reveals the fluctuation of the user's input speed. The cleaned data set is subjected to n-gram analysis. n-gram analysis is a commonly used natural language processing technology. By analyzing the key sequence input by the user, the user's commonly used key combination patterns can be identified to form a set of commonly used key combination patterns. Based on this pattern set, a key transfer probability matrix is ​​generated, which reflects the probability distribution of the user's transfer from one key to another during the input process. At the same time, the cleaned data set is subjected to Fourier transform to convert the time domain data into the frequency domain to obtain the user input rhythm spectrum. The user input rhythm spectrum reveals the periodic characteristics of the user input behavior. The rhythm feature vector is extracted through this spectrum, which reflects the rhythm pattern of the user input. The key frequency vector, speed feature vector, key transfer probability matrix and rhythm feature vector are reduced in dimension to simplify the data structure and reduce the computational complexity. The reduced-dimensional data form a reduced-dimensional feature matrix, which retains the main features of the original high-dimensional data while eliminating redundant information and noise. The user input feature vector is constructed based on the reduced-dimensional feature matrix and the scene labels in the original user input data set. The scene labels distinguish the input features under different usage environments, making the constructed user input feature vector more comprehensive and accurate.

[0032] Step S103: clustering and classifying the user input feature vectors to construct a user behavior pattern classification model;

[0033] Specifically, K-means clustering analysis is performed on the user input feature vector. K-means clustering distributes data points into k clusters through iterative optimization to generate initial clustering results. Gaussian mixture processing is performed on the initial clustering results. Through the Gaussian mixture model, the clustering results are represented as a mixture of multiple Gaussian distributions to obtain a probability distribution model. The sample similarity matrix is ​​calculated according to the probability distribution model. The similarity matrix reflects the degree of similarity between different samples. Spectral clustering is performed on the sample similarity matrix. Spectral clustering uses the spectral graph theory in graph theory to identify the natural clustering structure in the data by analyzing the eigenvalues ​​and eigenvectors of the sample similarity matrix to obtain the target clustering result. Spectral clustering can handle complex nonlinear structures, so that the clustering results are more in line with the actual user behavior pattern. The initial classification model is constructed according to the target clustering result. The initial classification model is based on the data distribution in the target clustering result and classifies different categories through a supervised learning algorithm. The initial classification model is integrated. By integrating multiple classification models to form an integrated classifier, the advantages of each model can be combined to improve the accuracy and robustness of classification. The feature weight vector of the user input feature vector is calculated based on the ensemble classifier. The feature weight vector reflects the importance of each feature in the classification process, which helps to identify and focus on key features. Based on the feature weight vector, a nonlinear classification boundary is generated. The nonlinear classification boundary can better adapt to complex data distribution and has higher classification accuracy and generalization ability than the linear classification boundary. Based on the nonlinear classification boundary and the ensemble classifier, a user behavior pattern classification model is constructed. The model can accurately classify and identify user behavior patterns and predict and analyze user behavior based on the user input feature vector.

[0034] Step S104: Divide the keyboard into regions to obtain a plurality of keyboard regions, and calculate the importance weight of each keyboard region based on the user behavior pattern classification model to obtain a keyboard region importance scoring matrix;

[0035] Specifically, the physical layout of the keyboard is gridded, and the keyboard surface is divided into multiple small grid units to form a keyboard grid coordinate system. The coordinate system helps to locate the position of each key, and divides the initial keyboard function area according to the range of user finger activity, ensuring that the division of each area is in line with the user's operating habits and comfort. The initial keyboard function area is optimized by Voronoi diagram. Voronoi diagram is a method of dividing space into several areas, each area corresponds to a seed point, and the distance from any point in the area to the seed point is the minimum. Through Voronoi diagram optimization, the initial keyboard function area is refined and adjusted to obtain multiple more reasonable and natural keyboard areas. According to the user behavior pattern classification model, a label is assigned to each keyboard area to obtain a regional function label set. Based on the user's actual usage habits and behavior patterns, a suitable function label is assigned to each area. A hierarchical analysis is performed on the regional function label set, and the similarity and correlation between the functions of each region are analyzed by constructing a regional function similarity tree. The regional function similarity tree can reveal the relationship between different functional areas. Based on the regional function similarity tree, the functional correlation between multiple keyboard areas is calculated to obtain a regional correlation matrix. The matrix reflects the functional relevance and mutual influence between different keyboard areas. At the same time, the distribution of user input data in each keyboard area is entropy analyzed to obtain the regional information entropy vector. The information entropy vector reflects the complexity and frequency of user operations in each area. High entropy areas usually indicate that user operations are concentrated and frequent, requiring more attention and optimization. Combining the regional information entropy vector and the regional association matrix, the regional importance initial scoring vector is constructed. The regional importance initial scoring vector is processed using the fuzzy comprehensive evaluation method to obtain the regional importance weight vector. The fuzzy comprehensive evaluation method can comprehensively consider the influence of multiple factors and obtain more accurate and reasonable weight values. According to the regional importance weight vector and multiple keyboard areas, a keyboard area importance scoring matrix is ​​constructed. The scoring matrix integrates multiple factors such as user behavior data, regional functional relevance and information entropy, and accurately reflects the importance and priority of each keyboard area.

[0036] Step S105: Optimizing the keyboard layout according to the keyboard area importance scoring matrix to obtain a personalized keyboard layout solution;

[0037] Specifically, the keyboard area importance score matrix is ​​processed by heat map visualization to generate a keyboard importance heat map. The importance of each keyboard area is displayed in a graphical manner. The changes in color depth in the heat map can clearly identify the high-frequency and low-frequency areas on the keyboard. According to the keyboard importance heat map, a key area set is identified, and the key area set includes keys and areas that users frequently operate during use. Genetic analysis is performed on the key area set, and multiple candidate key position arrangement schemes are generated by simulating the process of natural selection and evolution. The genetic analysis method can find the optimal solution in a wide solution space, and continuously optimize the key position arrangement through operations such as crossover, mutation and selection to form multiple candidate schemes with high potential. Based on multiple candidate key position arrangement schemes and user input feature vectors, the efficiency score of each candidate key position arrangement scheme is calculated. The efficiency score quantifies the advantages and disadvantages of each candidate scheme by comprehensively considering factors such as user input habits, key frequency, key combination, and finger movement distance. Multiple candidate key position arrangement schemes are sorted according to the efficiency score, and the top N candidate key position arrangement schemes are selected as the optimization basis, where N is a positive integer. The N selected solutions represent the best key arrangements in the current solution space. Local optimization and adjustment are performed on the top N candidate key arrangement solutions, and local details of each solution are fine-tuned to improve its efficiency and user experience. After completing the local optimization and adjustment, the local optimization adjustment results are obtained, and the optimal layout solution is generated based on these results. The optimal layout solution has the highest efficiency score and the best user applicability based on the combination of global and local optimization, ensuring the overall optimization effect of the keyboard layout. The key mapping relationship table is calculated based on the optimal layout solution and the original keyboard layout. The key mapping relationship table lists the key mapping rules from the original layout to the optimal layout, ensuring that users can smoothly transition to the new personalized keyboard layout, and finally obtain a personalized keyboard layout solution.

[0038] Step S106: continuously collect usage data and optimize the layout of the personalized keyboard layout solution to obtain layout optimization iteration parameters.

[0039] Specifically, the real-time input data of the user using the personalized keyboard layout scheme is collected to form a target user input data set. By continuously monitoring the user's keyboard operation, detailed input data including key frequency, key combination, typing speed, etc. are collected. The user's fitness index is calculated according to the target user input data set to generate a layout fitness score. The score reflects the user's adaptation to the new layout and measures the user's efficiency and comfort when using the personalized keyboard layout scheme. The layout fitness score is analyzed in time series, and the user learning curve model is obtained by analyzing the trend of score changes. The learning curve model reveals the process and speed of the user's adaptation to the new layout. According to the user learning curve model, the learning stage division result is generated to divide the user's learning process into different stages. The data of each stage reflects the characteristics of the user at different adaptation stages, and the features of the data of each stage are extracted to form a stage feature vector. These feature vectors contain the input behavior characteristics of the user at each stage, and reflect the user's operation habits and change trends at different learning stages. The user behavior pattern classification model is updated according to the stage feature vector to generate a dynamic behavior pattern classification model. The dynamic behavior pattern classification model can be adjusted as the user behavior changes, reflecting the user's operation characteristics and habits in different time periods. By dynamically updating, the classification model is ensured to always be consistent with the actual usage of users. The dynamic behavior pattern classification model and the keyboard area importance scoring matrix are compared and analyzed to obtain the importance change matrix. The importance change matrix reflects how the importance of each keyboard area changes with the changes in user fitness and behavior patterns. According to the importance change matrix, the personalized keyboard layout scheme is tested and evaluated to generate the layout scheme effect evaluation results. The evaluation results determine the advantages and disadvantages of the current layout scheme and the performance of users in the new layout by analyzing the actual effect of the layout. According to the layout scheme effect evaluation results, the layout optimization iteration parameters are constructed. The iteration parameters are used to guide the next round of layout optimization, so that the keyboard layout can be continuously improved and the user's input efficiency and experience can be gradually improved.

[0040] In the embodiment of the present application, by real-time monitoring of user key operations, collecting multi-dimensional data, and combining device environment parameters, a comprehensive original user input data set is constructed, and user input features are extracted from multiple angles such as frequency, speed, combination mode, rhythm, etc., to comprehensively characterize the user's input behavior pattern. A hierarchical user behavior pattern classification model is constructed in combination with K-means clustering, which can accurately identify and classify different user input patterns. Through grid processing and Voronoi diagram optimization, flexible division of keyboard areas is achieved, and combined with user behavior patterns and entropy analysis, regional importance scores that are more in line with user usage habits are obtained. Multiple candidate solutions are generated, and through efficiency scoring and local optimization, the optimal layout solution that takes into account efficiency and ergonomics is quickly obtained. Through continuous data collection and learning curve analysis, dynamic updating and optimization of the layout solution are achieved, so that the keyboard layout can continuously adapt to changes in user habits. The whole process fully considers factors such as the user's personal usage habits, device environment, and application scenarios, and realizes highly personalized keyboard layout customization. In the optimization process, input efficiency and ergonomic factors are considered at the same time, which can not only improve the user's input speed, but also reduce fatigue and improve the comfort of long-term use.

[0041] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0042] (1) Monitor the user's key operations in real time to obtain a key event sequence, and extract key timestamp information based on the key event sequence to obtain a key time sequence;

[0043] (2) Analyze the key event sequence to obtain the key position coordinate sequence, and calculate the user's key speed based on the time difference between adjacent key events to obtain the key speed sequence;

[0044] (3) Perform statistical analysis on the key event sequence to obtain the key frequency distribution matrix, and mark the key events according to the application information currently used by the user to obtain the key event sequence with scene labels;

[0045] (4) Collecting the user's device information to obtain a device environment parameter set, and generating a segmented input behavior feature sequence based on the key event sequence and the device environment parameter set;

[0046] (5) Construct the original user input dataset based on the segmented input behavior feature sequence and the key event sequence with scene labels.

[0047] Specifically, the user's key operations are monitored in real time to capture every key operation of the user. Each key operation generates a key event, which includes the key identifier (such as the key code), the key timestamp (i.e., the time when it was pressed), and the key position coordinates (i.e., the position of the key on the keyboard). These key events form a key event sequence in chronological order. The timestamp information can help the system construct a key time sequence, which represents the time interval and rhythm of the user's key presses. The key event sequence is analyzed, and the key position coordinate sequence is obtained by parsing the position coordinates in the key event. The sequence reflects the path and method of the user's movement on the keyboard. By calculating the time difference between adjacent key events, the formula is as follows:

[0048] ;

[0049] in, Representative The speed of keystrokes, Representative The distance between the keys (which can be calculated based on the coordinate difference on the keyboard), and It represents the The time interval between the last key press and the last key press. Through this formula, the speed of each key press is calculated to form a key speed sequence. The key event sequence is statistically analyzed to generate a key frequency distribution matrix, which records the number of times each key is pressed and its frequency distribution. Assuming that there are n keys on the keyboard, the key frequency distribution matrix can be expressed as an n×1 matrix, with each element corresponding to the frequency of a key. The key events are labeled according to the application information currently used by the user. These application information includes the name and type of the current active window, and a key event sequence with scene labels is generated. These labels can help the system understand the differences in user key behavior in different application scenarios. At the same time, the user's device information is collected, including device type, operating system version, screen resolution, etc. These device environment parameters can affect the user's key behavior. This information is combined with the key event sequence to generate a segmented input behavior feature sequence. In order to construct a complete original user input dataset, the segmented input behavior feature sequence and the key event sequence with scene labels are integrated to obtain a user input dataset containing rich information, which can reflect the user's key behavior, usage scenario, and device environment.

[0050] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0051] (1) Perform outlier detection on the original user input data set to obtain a cleaned data set, and calculate the usage frequency of each key based on the cleaned data set to obtain a key frequency vector;

[0052] (2) Perform sliding time window analysis on the cleaned data set to obtain a time series feature matrix, and calculate the user's average input speed and speed change rate based on the time series feature matrix to obtain a speed feature vector;

[0053] (3) Perform n-gram analysis on the cleaned data set to obtain a set of commonly used key combination patterns, and generate a key transfer probability matrix based on the set of commonly used key combination patterns;

[0054] (4) Performing Fourier transform on the cleaned data set to obtain the user input rhythm spectrum, and extracting the rhythm feature vector based on the user input rhythm spectrum;

[0055] (5) Performing dimensionality reduction processing on the key frequency vector, speed feature vector, key transfer probability matrix, and rhythm feature vector to obtain a reduced-dimensional feature matrix;

[0056] (6) Construct a user input feature vector based on the reduced dimension feature matrix and the scene labels in the original user input dataset.

[0057] Specifically, outlier detection is performed on the original user input data set to identify and remove unreasonable or noisy data to obtain a cleaned data set. For example, the triple standard deviation method is used to detect outliers in the key time series. If the time interval between the timestamp of a key event and the events before and after it is greater than three standard deviations, it can be regarded as an outlier. The usage frequency of each key is calculated based on the cleaned data set to obtain a key frequency vector. Assume that keys, the usage frequency vector can be expressed as ( Indicates The key frequency vector can be obtained by counting the number of occurrences of each key and dividing it by the total number of key presses. The formula is as follows:

[0058] ;

[0059] in, Indicates The number of times a key appears, Represents the total number of all keystrokes. Perform sliding time window analysis on the cleaned data set to capture the temporal dynamic characteristics of user input behavior. Divide the data set into time windows of fixed length and calculate the time series feature matrix in each window. For example, assuming the time window is 5 seconds, then keystroke events within every 5 seconds constitute a window. For each time window, calculate the user's average input speed and speed change rate. Assume that there are key events, average input speed and speed change rate It can be calculated by the following formula:

[0060] ;

[0061] ;

[0062] in, Indicates The second key and The distance between the keys. Indicates the time interval between two key events. Perform n-gram analysis to capture common key combination patterns. n-gram is a natural language processing technology that identifies high-frequency key sequences by analyzing n consecutive key combinations. Assuming n=3, analyze three-key combinations (trigram), such as "the", "and", etc. By counting these high-frequency combinations, a set of common key combination patterns is obtained. Based on the pattern set, a key transfer probability matrix is ​​generated. ,in Indicates that from the key Transfer to key The probability of the transition probability matrix is ​​calculated by the following formula:

[0063] ;

[0064] in, Indicates that from the key To button At the same time, the cleaned data set is transformed by Fourier transform to convert the input behavior in the time domain to the frequency domain to obtain the user input rhythm spectrum. Fourier transform can reveal the periodic characteristics of user input and extract the rhythm feature vector through the spectrum graph. Assume that the time series is , the Fourier transform formula is as follows:

[0065] ;

[0066] in, Representative frequency The amplitude on is time series data, is the kernel function of Fourier transform. Perform dimensionality reduction on key frequency vector, velocity feature vector, key transfer probability matrix and rhythm feature vector. Reduce the dimension of feature vector and remove redundant information through principal component analysis or other dimensionality reduction techniques to obtain a reduced dimensionality feature matrix. Construct a user input feature vector based on the reduced dimensionality feature matrix and the scene labels in the original user input dataset. For example, assuming that the user uses the keyboard in different applications, these scene labels can help the system understand the differences in user input behavior in different scenarios. The user input feature vector is obtained by combining the reduced dimensionality feature matrix with the scene labels.

[0067] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0068] (1) Perform K-means clustering analysis on the user input feature vector to obtain the initial clustering result, and perform Gaussian mixture processing on the initial clustering result to obtain a probability distribution model;

[0069] (2) Calculate the sample similarity matrix according to the probability distribution model, and perform spectral clustering on the sample similarity matrix to obtain the target clustering result;

[0070] (3) Construct an initial classification model based on the target clustering results, and integrate the initial classification model to obtain an integrated classifier;

[0071] (4) Calculating a feature weight vector of the user input feature vector according to the integrated classifier, and generating a nonlinear classification boundary according to the feature weight vector;

[0072] (5) Construct a user behavior pattern classification model based on nonlinear classification boundaries and integrated classifiers.

[0073] Specifically, K-means clustering analysis is performed on the user input feature vector. K-means clustering is an unsupervised learning algorithm that divides data into clusters by minimizing the distance between samples and the centers of their respective clusters. clusters. Each cluster is represented by a centroid, which is the average value of all samples in the cluster. Suppose the user input feature vector set is ,in It is The user input feature vector, the goal of the algorithm is to find Cluster Center , so that the following objective function Minimize:

[0074] ;

[0075] in, is a binary indicator variable, indicating that the sample Is it assigned to a cluster? . Through iterative optimization, the initial clustering results are obtained. The initial clustering results are processed by Gaussian mixture model. The Gaussian mixture model assumes that the data comes from a mixture of multiple Gaussian distributions, and describes the data distribution by estimating the parameters of each Gaussian distribution (mean, covariance matrix and mixture weight). Assumption The mixture model of Gaussian distributions is:

[0076] ;

[0077] in, Indicates Gaussian distribution weights satisfy N It means the mean , the covariance matrix is Gaussian distribution. Through the expectation maximization algorithm, these parameters are iteratively estimated to obtain the probability distribution model. The sample similarity matrix is ​​calculated based on the probability distribution model. Similarity matrix Elements Representation sample and samples The similarity between them is calculated using the Gaussian kernel function:

[0078] ;

[0079] in, It is a sample and The Euclidean distance between is the bandwidth parameter of the Gaussian kernel function. Perform spectral clustering on the sample similarity matrix. Spectral clustering is a clustering method based on graph theory. It identifies the clustering structure in the data by constructing a Laplacian matrix and calculating its eigenvalues ​​and eigenvectors. Assume that the similarity matrix is , define the degree matrix is a diagonal matrix with diagonal elements is the first The sum of the row elements. Laplacian matrix Defined as: . For the Laplacian matrix Perform feature decomposition and select The eigenvectors form the eigenmatrix , then Perform Kmeans clustering on the rows to obtain the target clustering result. Build the initial classification model based on the target clustering result. Assume that there are multiple initial classification models , each model is trained by different algorithms (such as decision tree, support vector machine, k nearest neighbor, etc.). In order to improve the classification performance, the initial classification models are integrated to construct an integrated classifier. The integrated classifier combines the prediction results of each model by voting or weighted average to obtain the final classification result. The feature weight vector of the user input feature vector is calculated according to the integrated classifier. The feature weight vector reflects the importance of each feature in the classification process and can be calculated by the feature importance measurement method in the integrated model (such as the feature importance of random forest). A nonlinear classification boundary is generated based on the feature weight vector. Assume that the function of the classification boundary is , which can be expressed as:

[0080] ;

[0081] in, is the feature weight, It is a feature The nonlinear transformation of is the dimension of the feature vector. Based on the nonlinear classification boundary and the integrated classifier, a user behavior pattern classification model is constructed. The model can classify according to the user input feature vector and identify the user's behavior pattern.

[0082] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0083] (1) Gridding the physical layout of the keyboard to obtain a keyboard grid coordinate system, and dividing the initial keyboard function area according to the keyboard grid coordinate system and the user's finger activity range;

[0084] (2) Optimize the Voronoi diagram of the initial keyboard function area to obtain multiple keyboard areas;

[0085] (3) Assign labels to each keyboard area according to the user behavior pattern classification model to obtain a regional function label set, and perform hierarchical analysis on the regional function label set to obtain a regional function similarity tree;

[0086] (4) Calculate the functional association between multiple keyboard regions based on the regional functional similarity tree to obtain the regional association matrix, and perform entropy analysis on the distribution of user input data in each keyboard region to obtain the regional information entropy vector;

[0087] (5) Based on the regional information entropy vector and the regional association matrix, the initial scoring vector of regional importance is constructed, and the initial scoring vector of regional importance is subjected to fuzzy comprehensive evaluation to obtain the regional importance weight vector;

[0088] (6) Construct a keyboard region importance scoring matrix based on the region importance weight vector and multiple keyboard regions.

[0089] Specifically, the physical layout of the keyboard is gridded. By mapping the physical position of the keyboard into a two-dimensional grid coordinate system, the position of each key can be represented as a coordinate point. The initial keyboard function area is divided according to the keyboard grid coordinate system and the user's finger activity range. The user's finger activity range refers to the activity area of ​​each finger when the user is typing normally. For example, the little finger of the left hand is mainly active on the left side of the keyboard, and the index finger of the right hand is active on the middle right side of the keyboard. According to these activity ranges, the keyboard is divided into several initial function areas, each area corresponding to a specific finger activity range. The initial keyboard function area is optimized by Voronoi diagram. Voronoi diagram is a space segmentation method, which divides the space into several polygonal areas, each polygonal area corresponds to a seed point, and the distance from any point in the area to the seed point is the minimum. By optimizing the initial keyboard function area by Voronoi diagram, multiple more accurate and natural keyboard areas are obtained. Labels are assigned to each keyboard area according to the user behavior pattern classification model to obtain a set of regional function labels. For example, a certain area may be mainly used to input letters, and another area may be mainly used to input numbers or special symbols. Through this label assignment, the main function of each area is clarified. A hierarchical analysis is performed on the regional function label set to obtain the regional function similarity tree. The hierarchical analysis method is a method of decomposing complex problems into multiple levels for analysis. By building a hierarchical model, the relative importance of each factor is calculated. The regional function similarity tree can reveal the similarity and correlation between different functional areas. Based on the regional function similarity tree, the functional correlation between multiple keyboard areas is calculated to obtain the regional correlation matrix. The elements of the regional correlation matrix are Indicates area and Region The functional association between the two regions is assumed to be , then the correlation can be expressed as:

[0090] ;

[0091] in, is the total number of regions. Through standardization, the correlation between each region and other regions is obtained. Entropy analysis is performed on the distribution of user input data in each keyboard region to obtain the regional information entropy vector. Information entropy is an indicator of the degree of information confusion. By calculating the entropy value of the input data distribution in each region, the complexity and frequency of user operations in each region can be understood. Assuming the region The information entropy is , which can be expressed as:

[0092] ;

[0093] in, Is the user in the area Input button The probability of It is a region According to the regional information entropy vector and the regional association matrix, the regional importance initial scoring vector is constructed. The regional importance initial scoring vector can be expressed as:

[0094] ;

[0095] in, and are weight parameters, which represent the importance of information entropy and correlation respectively. The initial scoring vector of regional importance is subjected to fuzzy comprehensive evaluation to obtain the regional importance weight vector. The fuzzy comprehensive evaluation method introduces fuzzy logic and comprehensively considers multiple factors to obtain a more accurate and reasonable weight value. Assume that the region The importance weight is , calculated by the fuzzy comprehensive evaluation formula:

[0096] ;

[0097] in, It is a region Initial scoring The membership degree of the keyboard region is obtained by calculating the regional importance weight vector and the multiple keyboard regions, and constructing a keyboard region importance scoring matrix. The keyboard region importance scoring matrix indicates the importance of each region in the keyboard layout.

[0098] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0099] (1) Perform heat map visualization on the keyboard area importance score matrix to obtain the keyboard importance heat distribution map;

[0100] (2) Identify the key area set according to the keyboard importance heat distribution map, and perform genetic analysis on the key area set to generate multiple candidate key position arrangement schemes;

[0101] (3) Calculating the efficiency score of each candidate key position arrangement scheme based on multiple candidate key position arrangement schemes and the user input feature vector;

[0102] (4) Sort multiple candidate key position arrangement schemes according to the efficiency scores, and select the top N candidate key position arrangement schemes as the optimization basis, where N is a positive integer;

[0103] (5) Locally optimize and adjust the top N candidate key position arrangement schemes to obtain local optimization adjustment results, and generate an optimal layout scheme based on the local optimization adjustment results;

[0104] (6) Calculate the key mapping relationship table based on the optimal layout plan and the original keyboard layout to obtain a personalized keyboard layout plan.

[0105] Specifically, the keyboard area importance score matrix is ​​visualized using a heat map. A heat map is an intuitive visualization method that uses the depth of color to represent the size of data. For the keyboard area importance score matrix, a keyboard importance heat map is generated by mapping the importance score of each area to a color value. Assume that the score matrix is , each element in the matrix Represents the keyboard area Importance score. A heat map is obtained by normalizing the values ​​in the scoring matrix to between 0 and 1 and mapping them to a color gradient, where the darker the color, the higher the importance of the area. A set of key areas is identified based on the keyboard importance heat map. These key areas are the areas with the darkest colors, i.e., the areas with the highest importance scores. After identifying these key areas, these areas are analyzed in detail to optimize the keyboard layout. In order to generate multiple candidate key arrangement schemes, a genetic algorithm is used to find the optimal solution by simulating natural selection and genetic mutation. An initial population is generated, and each individual represents a key arrangement scheme. Through operations such as selection, crossover, and mutation, a new population is iteratively generated, gradually approaching the optimal solution. For each candidate key arrangement scheme, its efficiency score is calculated based on the user input feature vector. Assume that the candidate key arrangement scheme is , the user input feature vector is The efficiency score can be evaluated by calculating the typing efficiency of the user under this arrangement. The specific formula is as follows:

[0106] ;

[0107] in, Represents the arrangement scheme The efficiency score, represents the dimension of the user input feature vector, Representation characteristics The weight of Indicates that the user is arranging the plan Down button The efficiency of each candidate key arrangement scheme is calculated by calculating the efficiency score of each scheme, and the score value of each scheme is obtained. Multiple candidate key arrangement schemes are sorted according to the efficiency score, and the top N candidate key arrangement schemes are selected as the optimization basis. N is a positive integer, which represents the number of optimal schemes selected. These schemes represent the best key arrangement schemes under the current conditions. The top N candidate key arrangement schemes are locally optimized and adjusted. Based on the candidate schemes, the efficiency is improved by fine-tuning the key positions. After the local optimization adjustment is completed, the local optimization adjustment results are obtained. Based on these results, the optimal layout scheme is generated. The optimal layout scheme refers to the scheme with the highest overall efficiency and the best user experience among all candidate schemes. The key mapping relationship table is calculated based on the optimal layout scheme and the original keyboard layout to obtain a personalized keyboard layout scheme. The key mapping relationship table lists the key mapping rules from the original layout to the optimal layout to ensure that users can clearly understand the changes in key positions.

[0108] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0109] (1) Collecting real-time input data of users using the personalized keyboard layout solution to obtain a target user input data set;

[0110] (2) Calculate the user's fitness index based on the target user input data set to obtain the layout fitness score, and perform time series analysis on the layout fitness score to obtain the user learning curve model;

[0111] (3) Generate a learning stage division result based on the user learning curve model, and extract features from each stage data in the learning stage division result to obtain a stage feature vector;

[0112] (4) Update the user behavior pattern classification model according to the phased feature vector to obtain a dynamic behavior pattern classification model;

[0113] (5) Compare and analyze the dynamic behavior pattern classification model and the keyboard area importance scoring matrix to obtain the importance change matrix;

[0114] (6) Test and evaluate the personalized keyboard layout scheme according to the importance change matrix to obtain the layout scheme effect evaluation result;

[0115] (7) Based on the layout scheme effect evaluation results, construct layout optimization iteration parameters, and the layout optimization iteration parameters are used for the next round of layout optimization of the keyboard.

[0116] Specifically, the real-time input data of users using the personalized keyboard layout solution is collected to form a target user input data set. The keyboard monitoring software runs in the background to record the timestamp, key position, key type (such as letters, numbers or symbols) of each key press, and obtain the detailed input behavior data of the user when using the personalized keyboard layout solution. The user's fitness index is calculated based on the target user input data set to evaluate the user's adaptation to the new layout. The fitness index can be calculated by measuring the user's typing speed, accuracy, and comfort. Assume that the user's input data over a period of time is , then the fitness index It can be expressed as the following function:

[0117] ;

[0118] in, Represents typing speed (such as words per minute), represents the accuracy of the input (such as error rate), Represents the operating comfort (such as whether the key distribution conforms to the law of natural finger movement). is the coefficient used to balance the weights of each indicator. Through this formula, the user's layout fitness score is obtained. The layout fitness score is analyzed in time series to obtain the user's learning curve model. The learning curve model reflects the user's adaptation process to the new layout over a period of time. Assume that the fitness score sequence is , through time series analysis methods (such as moving average, exponential smoothing, etc.), fit the user's learning curve model , indicating that the user is in The user learning curve model is used to generate the learning stage division results, and the user's learning process is divided into different stages. For example, the process of users adapting to the new layout can be divided into the initial adaptation stage, the stable improvement stage, and the efficient use stage. The data of each learning stage is extracted to form a stage feature vector. The user behavior pattern classification model is updated according to the stage feature vector to obtain a dynamic behavior pattern classification model. The dynamic behavior pattern classification model can be adjusted as the user behavior changes, and more accurately reflect the user's operating characteristics and habits in different time periods. Assume that the original user behavior pattern classification model is , through the new phase feature vector Train to obtain an updated dynamic behavior pattern classification model The dynamic behavior pattern classification model and the keyboard area importance scoring matrix are compared and analyzed to evaluate the impact of user behavior changes on the importance of keyboard layout. and the original rating matrix , calculate the importance change matrix , whose elements Indicates area The importance change of . Assume that the calculation formula of importance change is:

[0119] ;

[0120] in, Represents the dynamic model for the region The importance rating of Represents the region in the original score matrix According to the importance change matrix, the personalized keyboard layout scheme is tested and evaluated to obtain the layout scheme effect evaluation results. The effect evaluation results can be evaluated by the user's fitness score change, input efficiency change and other aspects. According to the layout scheme effect evaluation results, the layout optimization iteration parameters are constructed. The layout optimization iteration parameters are used to guide the next round of layout optimization.

[0121] The above describes the keyboard area customization method based on user behavior analysis in the embodiment of the present application. The following describes the keyboard area customization system based on user behavior analysis in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the keyboard area customization system based on user behavior analysis includes:

[0122] The collection module 201 is used to collect data of user keyboard operation behavior in real time to obtain an original user input data set;

[0123] Extraction module 202, used for preprocessing and feature extraction of the original user input data set to obtain a user input feature vector;

[0124] A construction module 203 is used to cluster and classify the user input feature vectors and construct a user behavior pattern classification model;

[0125] The calculation module 204 is used to divide the keyboard into regions to obtain multiple keyboard regions, and calculate the importance weight of each keyboard region based on the user behavior pattern classification model to obtain a keyboard region importance scoring matrix;

[0126] An optimization module 205 is used to optimize the keyboard layout according to the keyboard area importance scoring matrix to obtain a personalized keyboard layout solution;

[0127] The iteration module 206 is used to continuously collect usage data and optimize the layout of the personalized keyboard layout solution to obtain layout optimization iteration parameters.

[0128] Through the collaboration of the above components, by real-time monitoring of user keystrokes, collecting multi-dimensional data, and combining device environment parameters, a comprehensive original user input data set is constructed, and user input features are extracted from multiple angles such as frequency, speed, combination mode, rhythm, etc., which comprehensively characterizes the user's input behavior pattern. A hierarchical user behavior pattern classification model is constructed in combination with K-means clustering, which can accurately identify and classify different user input patterns. Through grid processing and Voronoi diagram optimization, flexible division of keyboard areas is achieved, and combined with user behavior patterns and entropy analysis, regional importance scores that are more in line with user usage habits are obtained. Multiple candidate solutions are generated, and through efficiency scoring and local optimization, the optimal layout solution that takes into account efficiency and ergonomics is quickly obtained. Through continuous data collection and learning curve analysis, the layout solution is dynamically updated and optimized, so that the keyboard layout can continuously adapt to changes in user habits. The entire process fully considers factors such as the user's personal usage habits, device environment, and application scenarios, and realizes highly personalized keyboard layout customization. In the optimization process, input efficiency and ergonomic factors are considered at the same time, which can not only improve the user's input speed, but also reduce fatigue and improve the comfort of long-term use.

[0129] The present application also provides a keyboard area customization device based on user behavior analysis, and the keyboard area customization device based on user behavior analysis includes a memory and a processor, and the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes the steps of the keyboard area customization method based on user behavior analysis in the above-mentioned embodiments.

[0130] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the keyboard area customization method based on user behavior analysis.

[0131] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0132] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0133] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A keyboard area customization method based on user behavior analysis, characterized in that: The method comprises: Real-time data collection of user keyboard operation behavior is performed to obtain an original user input data set; specifically, the method includes: real-time monitoring of the user's key operation to obtain a key event sequence, and extracting key timestamp information according to the key event sequence to obtain a key time sequence; analyzing the key event sequence to obtain a key position coordinate sequence, and calculating the user's key speed according to the time difference between adjacent key events to obtain a key speed sequence; statistically analyzing the key event sequence to obtain a key frequency distribution matrix, and marking the key events according to the application information currently used by the user to obtain a key event sequence with scene labels; collecting the user's device information to obtain a device environment parameter set, and generating a segmented input behavior feature sequence according to the key event sequence and the device environment parameter set; constructing an original user input data set according to the segmented input behavior feature sequence and the key event sequence with scene labels; The original user input data set is preprocessed and feature extracted to obtain a user input feature vector; specifically comprising: performing outlier detection on the original user input data set to obtain a cleaned data set, and calculating the usage frequency of each key based on the cleaned data set to obtain a key frequency vector; performing sliding time window analysis on the cleaned data set to obtain a time series feature matrix, and calculating the user's average input speed and speed change rate based on the time series feature matrix to obtain a speed feature vector; performing n-gram analysis on the cleaned data set to obtain a set of commonly used key combination patterns, and generating a key transfer probability matrix based on the set of commonly used key combination patterns; performing Fourier transform on the cleaned data set to obtain a user input rhythm spectrum, and extracting a rhythm feature vector based on the user input rhythm spectrum; performing dimensionality reduction processing on the key frequency vector, the speed feature vector, the key transfer probability matrix and the rhythm feature vector to obtain a reduced dimensionality feature matrix; constructing a user input feature vector based on the reduced dimensionality feature matrix and the scene labels in the original user input data set; Clustering and classifying the user input feature vector to construct a user behavior pattern classification model; specifically comprising: performing K-means clustering analysis on the user input feature vector to obtain an initial clustering result, and performing Gaussian mixture processing on the initial clustering result to obtain a probability distribution model; calculating a sample similarity matrix according to the probability distribution model, and performing spectral clustering on the sample similarity matrix to obtain a target clustering result; constructing an initial classification model according to the target clustering result, and integrating the initial classification model to obtain an integrated classifier; calculating a feature weight vector of the user input feature vector according to the integrated classifier, and generating a nonlinear classification boundary according to the feature weight vector; constructing a user behavior pattern classification model according to the nonlinear classification boundary and the integrated classifier; The keyboard is divided into regions to obtain a plurality of keyboard regions, and the importance weight of each keyboard region is calculated based on the user behavior pattern classification model to obtain a keyboard region importance scoring matrix; specifically, the method includes: gridding the physical layout of the keyboard to obtain a keyboard grid coordinate system, and dividing the initial keyboard function region according to the keyboard grid coordinate system and the user's finger activity range; performing Voronoi diagram optimization on the initial keyboard function region to obtain a plurality of keyboard regions; assigning labels to each keyboard region according to the user behavior pattern classification model to obtain a regional function label set, and performing hierarchical analysis on the regional function label set to obtain a regional function similarity tree; calculating the functional association between the plurality of keyboard regions according to the regional function similarity tree to obtain a regional association matrix, and performing entropy analysis on the distribution of user input data in each keyboard region to obtain a regional information entropy vector; constructing a regional importance initial scoring vector according to the regional information entropy vector and the regional association matrix, and performing fuzzy comprehensive evaluation on the regional importance initial scoring vector to obtain a regional importance weight vector; constructing a keyboard region importance scoring matrix according to the regional importance weight vector and the plurality of keyboard regions; The keyboard layout is optimized according to the keyboard area importance scoring matrix to obtain a personalized keyboard layout plan; specifically comprising: performing heat map visualization processing on the keyboard area importance scoring matrix to obtain a keyboard importance heat distribution map; identifying a key area set according to the keyboard importance heat distribution map, and performing genetic analysis on the key area set to generate multiple candidate key position arrangement schemes; calculating the efficiency score of each candidate key position arrangement scheme according to the multiple candidate key position arrangement schemes and the user input feature vector; sorting the multiple candidate key position arrangement schemes according to the efficiency score, selecting the first N candidate key position arrangement schemes as the optimization basis, where N is a positive integer; performing local optimization adjustment on the first N candidate key position arrangement schemes to obtain a local optimization adjustment result, and generating an optimal layout plan according to the local optimization adjustment result; calculating a key position mapping relationship table according to the optimal layout plan and the original keyboard layout to obtain a personalized keyboard layout plan; The usage data of the personalized keyboard layout scheme is continuously collected and the layout is optimized to obtain layout optimization iteration parameters; specifically, the method includes: collecting real-time input data of users using the personalized keyboard layout scheme to obtain a target user input data set; calculating the user's fitness index based on the target user input data set to obtain a layout fitness score, and performing time series analysis on the layout fitness score to obtain a user learning curve model; generating a learning stage division result based on the user learning curve model, and extracting features from each stage data in the learning stage division result to obtain a stage feature vector; updating the user behavior pattern classification model based on the stage feature vector to obtain a dynamic behavior pattern classification model; comparing and analyzing the dynamic behavior pattern classification model and the keyboard area importance score matrix to obtain an importance change matrix; testing and evaluating the personalized keyboard layout scheme based on the importance change matrix to obtain a layout scheme effect evaluation result; constructing layout optimization iteration parameters based on the layout scheme effect evaluation result, and the layout optimization iteration parameters are used for the next round of layout optimization of the keyboard.

2. A keyboard area customization system based on user behavior analysis, characterized in that: For executing the keyboard area customization method based on user behavior analysis as claimed in claim 1, the system comprises: The acquisition module is used to collect real-time data of user keyboard operation behavior to obtain the original user input data set; An extraction module, used for preprocessing and feature extraction of the original user input data set to obtain a user input feature vector; A construction module, used to cluster and classify the user input feature vectors to construct a user behavior pattern classification model; A calculation module, used to divide the keyboard into regions to obtain a plurality of keyboard regions, and calculate the importance weight of each keyboard region based on the user behavior pattern classification model to obtain a keyboard region importance scoring matrix; An optimization module, used to optimize the layout of the keyboard according to the keyboard area importance scoring matrix to obtain a personalized keyboard layout solution; The iteration module is used to continuously collect usage data and optimize the layout of the personalized keyboard layout solution to obtain layout optimization iteration parameters.

3. A keyboard area customization device based on user behavior analysis, characterized in that: The keyboard area customization device based on user behavior analysis includes: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the keyboard area customization device based on user behavior analysis to execute the keyboard area customization method based on user behavior analysis as claimed in claim 1.

4. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the keyboard area customization method based on user behavior analysis as described in claim 1 is implemented.

Images