Metacosm user behavior and data analysis method and system based on digital twinning

Abstract

The invention discloses a meta cosmic user behavior and data analysis method and system based on digital twinning, and relates to the technical field of data processing, and the method comprises the steps: obtaining to-be-analyzed target meta cosmic space activity data; preprocessing the acquired activity data; the preprocessed activity data are classified for feature extraction, and data features of different sources are fused; and based on the analysis of the fusion features, adjusting a target user activity big data analysis algorithm in real time. Data are collected and encrypted in real time through an API interface, data safety and user privacy are ensured, data quality and consistency are improved through a fine preprocessing step, multi-dimensional behavior features of a user are comprehensively captured through feature extraction and fusion of multi-modal data, and user experience is improved. Algorithm parameters are adjusted in real time through an adaptive learning rate method, and the adaptability and accuracy of the algorithm are improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a metaverse user behavior and data analysis method and system based on digital twins. Background Art

[0002] As a new digital ecosystem, the Metaverse has gradually become a hot area of ​​research and application. The Metaverse not only provides a rich virtual experience, but also promotes users' social interaction, entertainment consumption, work and study in the virtual space. In order to better understand and predict the user's behavior patterns in the Metaverse, data analysis technology has become particularly important.

[0003] Traditional data analysis methods mainly rely on single-modal data, such as text or images, while ignoring the comprehensive analysis of multimodal data. However, user activity data in the metaverse is highly diverse and complex, and single-modal data analysis is difficult to fully capture the multi-dimensional behavioral characteristics of users. Most existing algorithm training and parameter adjustment methods use offline learning methods and lack the ability to adapt to changes in user behavior in real time. As user behavior continues to change, fixed algorithm parameters may lead to a decrease in prediction accuracy, affecting system performance and user experience. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a metaverse user behavior and data analysis method and system based on digital twins to solve the shortcomings of the prior art in multimodal data processing and real-time user behavior analysis.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a metaverse user behavior and data analysis method based on digital twins, which includes:

[0008] Obtain the target metaverse space activity data to be analyzed;

[0009] Preprocessing the acquired activity data;

[0010] Classify the preprocessed activity data for feature extraction and fuse the data features from different sources;

[0011] Based on the analysis of fusion features, the big data analysis algorithm of target user activities is adjusted in real time.

[0012] As a preferred solution of the digital twin-based metaverse user behavior and data analysis method of the present invention, the specific steps of obtaining the target metaverse space activity data to be analyzed are as follows:

[0013] Through the API interface, real-time monitoring and collection of user-generated data, including text data, audio data, video data, and image data;

[0014] The data is encrypted using the AES-256 algorithm and uploaded to the server;

[0015] After the encrypted data is uploaded to the server, it is anonymized to remove the user's identity information, including user name and registered account;

[0016] Use geocoding to blur the precise location of the data to the province or city level.

[0017] As a preferred solution of the digital twin-based metaverse user behavior and data analysis method of the present invention, the specific steps of preprocessing the acquired activity data are as follows:

[0018] Use regular expressions to remove non-alphanumeric characters from text data, retain English, numeric, and Chinese characters, and remove URL links from text;

[0019] All images were resized to the same size (224 × 224 pixels) to meet the input requirements of the deep learning algorithm, and the image colors were normalized by subtracting the mean of each color channel;

[0020] When the image size is larger than the target size, the central area is cropped from it as the final input;

[0021] Convert the video file into a series of static images, extract a fixed number of frames per second, and process each frame in the same way as image data;

[0022] Unify all audio files to the same sampling rate and use spectral gating to remove background noise from the audio.

[0023] As a preferred solution of the digital twin-based metaverse user behavior and data analysis method of the present invention, the pre-processed activity data is classified for feature extraction, and data features from different sources are fused. The specific steps are as follows:

[0024] Use the TF-IDF algorithm to extract features from the preprocessed text data. The expression is:

[0025]

[0026] Among them, TF-IDF(t,d,D) is the extracted text data feature vector, t is a word, d is a document, D is a document set, count(t i ) is the word t i The number of times it appears in document d, max j (count(t j )) is the word t that appears most frequently in document d j , N is the total number of documents, n is the total number of words in document d, docFreq(t) is the number of words containing word t i The number of documents, i and j are index variables;

[0027] Using the trained deep learning algorithm, the image is input into ResNet-50 to obtain the feature vector, which is expressed as follows:

[0028] V = ResNet-50(I);

[0029] Among them, V is the output feature vector, and I is the input image;

[0030] The video data is preprocessed to obtain frame images, and the features of the video frame images are extracted according to the image feature extraction method. The expression is as follows:

[0031] V f =ResNet-50(I f );

[0032] f=1,2,3....F;

[0033] Among them, V f is the feature vector of the f-th frame image, I f is the f-th frame image of the video, and F is the total number of frame images extracted;

[0034] The Mel frequency cepstral coefficient is used to extract the characteristics of audio data. The expression is as follows:

[0035] M = MFCC(A);

[0036] Where A is the audio signal, M is the extracted audio signal feature vector, and MFCC is the Mel frequency cepstrum coefficient;

[0037] The feature vectors of different types of data are merged into a unified feature vector.

[0038] As a preferred solution of the digital twin-based metaverse user behavior and data analysis method of the present invention, the feature vectors of different types of data are fused into a unified feature vector. The specific steps are as follows:

[0039] The feature vectors of the extracted data of different types are fused into a unified feature vector and normalized. The expression is as follows:

[0040]

[0041] Among them, S is the fusion feature vector, ‖T‖ 2 , ‖V‖ 2 , ‖V f ‖ 2 , ‖M‖ 2 is the text feature vector T, image feature vector V, video feature vector V f , the L2 norm of the audio feature vector M, x is the kth element in the vector X, and L is the length of the vector.

[0042] As a preferred solution of the digital twin-based metaverse user behavior and data analysis method of the present invention, the analysis based on fusion features adjusts the target user activity big data analysis algorithm in real time. The specific steps are as follows:

[0043] Input the fused feature vector into the target user activity big data analysis algorithm and output the result. The expression is as follows:

[0044]

[0045] in, is the predicted value of the algorithm, h is the prediction function of the algorithm, and θ is the algorithm parameter;

[0046] Use the chain rule to calculate the gradient of the loss function with respect to the algorithm parameters, and update the algorithm parameters by gradient descent. The expression is as follows:

[0047]

[0048] Among them, Q is the number of fusion features, is the i-th predicted value, y i is the i-th true value, S i is the i-th fused feature vector, i is the index variable, θ ′ is the updated algorithm parameter, α is the learning rate, which controls the step size of parameter update, is the gradient of the loss function with respect to the algorithm parameters;

[0049] The Adam optimizer is used as the adaptive learning rate method to adaptively adjust the learning rate to minimize the error and obtain the final algorithm parameters. The algorithm finally reaches the optimal solution.

[0050] As a preferred solution of the digital twin-based metaverse user behavior and data analysis method described in the present invention, the Adam optimizer is used as an adaptive learning rate method to adaptively adjust the learning rate. The specific steps are as follows:

[0051] Through the Adam optimizer, the learning rate is adaptively updated to further improve the algorithm parameter update and obtain the final algorithm parameters. The expression is as follows:

[0052]

[0053] Among them, θ″ is the final algorithm parameter, is the bias-corrected second-order moment estimate, is the first-order moment estimate after bias correction, ∈ is a constant;

[0054] Regularly evaluate the performance of the algorithm. Evaluation indicators include accuracy, recall rate, F1 score, and AUC-ROC curve. Use the evaluation results as feedback. If the algorithm performance deteriorates or does not meet expectations, readjust the learning rate to improve the algorithm performance.

[0055] In a second aspect, the present invention provides a metaverse user behavior and data analysis system based on digital twins, including a data acquisition module, a data preprocessing module, a feature extraction and fusion module, and a parameter update module;

[0056] The data acquisition module is used to obtain the target metaverse space activity data to be analyzed;

[0057] The data preprocessing module is used to preprocess the acquired activity data;

[0058] The feature extraction and fusion module is used to classify the pre-processed activity data for feature extraction and fuse the data features from different sources;

[0059] The parameter updating module is used to adjust the target user activity big data analysis algorithm in real time based on the analysis of fusion features.

[0060] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the metaverse user behavior and data analysis method based on digital twins as described in the first aspect of the present invention.

[0061] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored, wherein: when the computer program is executed by a processor, it implements any step of the metaverse user behavior and data analysis method based on digital twins as described in the first aspect of the present invention.

[0062] The beneficial effects of the present invention are as follows: the present invention collects and encrypts data in real time through an API interface, thereby ensuring data security and user privacy; improves data quality and consistency through sophisticated preprocessing steps; comprehensively captures multi-dimensional behavioral characteristics of users through feature extraction and fusion of multimodal data; and improves algorithm adaptability and accuracy by adjusting algorithm parameters in real time through an adaptive learning rate method. These improvements not only improve the accuracy and comprehensiveness of user behavior analysis, but also ensure the real-time performance of the system, providing strong technical support for personalized services and user experience optimization in the metaverse. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] 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 only 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.

[0064] Figure 1 This is a flow chart of the metaverse user behavior and data analysis method based on digital twins in Example 1.

[0065] Figure 2 This is a module diagram of the metaverse user behavior and data analysis system based on digital twins in Example 1. DETAILED DESCRIPTION

[0066] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0067] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0068] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0069] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a metaverse user behavior and data analysis method based on digital twins, comprising the following steps:

[0070] S1. Obtain the target metaverse space activity data to be analyzed.

[0071] Furthermore, through the API interface, the data generated by users can be monitored and collected in real time, including text data, audio data, video data, and image data;

[0072] The data is encrypted using the AES-256 algorithm and uploaded to the server;

[0073] After the encrypted data is uploaded to the server, it is anonymized to remove the user's identity information, including user name and registered account;

[0074] Use geocoding to blur the precise location of data to the provincial and municipal level;

[0075] S2. Preprocess the acquired activity data.

[0076] Furthermore, regular expressions are used to remove non-alphanumeric characters from text data, retaining English, numeric, and Chinese characters, and removing URL links from the text;

[0077] Furthermore, all images were resized to the same size (224 × 224 pixels) to accommodate the input requirements of the deep learning algorithm, and the image colors were normalized by subtracting the mean of each color channel;

[0078] When the image size is larger than the target size, the central area is cropped from it as the final input;

[0079] Furthermore, the video file is converted into a series of static images, a fixed number of frames are extracted per second, and each frame is processed in the same way as image data;

[0080] Furthermore, all audio files are unified to the same sampling rate and spectral gating method is used to remove background noise from the audio.

[0081] S3. Classify the preprocessed activity data for feature extraction and fuse the data features from different sources.

[0082] Furthermore, the TF-IDF (Term Frequency-Inverse Document Frequency) algorithm is used to extract features from the preprocessed text data. The expression is:

[0083]

[0084] Among them, TF-IDF(t,d,D) is the extracted text data feature vector, t is a word, d is a document, D is a document set, count(t i) is the word t i The number of times it appears in document d, max j (count(t j )) is the word t that appears most frequently in document d j , N is the total number of documents, n is the total number of words in document d, docFreq(t) is the number of words containing word t i The number of documents, i and j are index variables;

[0085] Using the trained deep learning algorithm, the image is input into ResNet-50 to obtain the feature vector, which is expressed as follows:

[0086] V = ResNet-50(I);

[0087] Among them, V is the output feature vector, and I is the input image;

[0088] The video data is preprocessed to obtain frame images, and the features of the video frame images are extracted according to the image feature extraction method. The expression is as follows:

[0089] V f =ResNet-50(I f );

[0090] f=1,2,3....F;

[0091] Among them, V f is the feature vector of the f-th frame image, I f is the f-th frame image of the video, and F is the total number of frame images extracted;

[0092] The Mel frequency cepstral coefficient is used to extract the characteristics of audio data. The expression is as follows:

[0093] M = MFCC(A);

[0094] Where A is the audio signal, M is the extracted audio signal feature vector, and MFCC is the Mel frequency cepstrum coefficient;

[0095] Furthermore, the feature vectors of the extracted data of different types are fused into a unified feature vector and normalized. The expression is as follows:

[0096]

[0097] Among them, x is the kth element in vector X, L is the length of the vector, S is the fused feature vector, ‖T‖ 2 , ‖V‖ 2 , ‖V f ‖ 2 , ‖M‖ 2is the text feature vector T, image feature vector V, video feature vector V f , the L2 norm of the audio feature vector M;

[0098] The sum of the four normalized eigenvectors is multiplied by Ensure that the length of the fused feature vector S is 1;

[0099] It should be noted that through multimodal data fusion, text data, image data, video data, and audio data are synthesized into a fused feature vector with a length of 1. Normalization is used during feature fusion to ensure the comparability of features of different modalities and the consistency of algorithm input, and to avoid the situation where certain features have a dominant effect on the algorithm due to their large numerical range.

[0100] S4. Based on the analysis of fusion features, adjust the big data analysis algorithm of target user activities in real time.

[0101] Furthermore, the fused feature vector is input into the target user activity big data analysis algorithm, and the output result is expressed as follows:

[0102]

[0103] in, is the predicted value of the algorithm, h is the prediction function of the algorithm, and θ is the algorithm parameter;

[0104] Furthermore, the chain rule is used to calculate the gradient of the loss function with respect to the algorithm parameters, and the algorithm parameters are updated by the gradient descent method. The expression is as follows:

[0105]

[0106] Among them, Q is the number of fusion features, is the i-th predicted value, y i is the i-th true value, S i is the i-th fused feature vector, i is the index variable, θ ′ is the updated algorithm parameter, α is the learning rate, which controls the step size of parameter update, is the gradient of the loss function with respect to the algorithm parameters;

[0107] Furthermore, the Adam optimizer is used as an adaptive learning rate method to adaptively adjust the learning rate in real time to minimize the error and obtain the final algorithm parameters. The algorithm finally reaches the optimal solution, which is expressed as follows:

[0108]

[0109] Among them, θ″ is the final algorithm parameter, is the bias-corrected second-order moment estimate, is the bias-corrected first-order moment estimate, ∈ is a constant, 1e-8, for numerical stability to ensure that the denominator is not zero, m t is the first-order moment estimate of the gradient, that is, the exponentially weighted average of the gradient, with an initial value of 0, v t is the second-order moment estimate of the gradient, that is, the exponentially weighted average of the square of the gradient, with an initial value of 0, v t-1 is the second-order moment estimate of the previous step, m t-1 is the first-order moment estimate of the previous step, β 1 , β 2 They are the decay rate of the first-order moment estimate, usually 0.9, and the decay rate of the second-order moment estimate, usually 0.999.

[0110] Furthermore, the performance of the algorithm is evaluated regularly. The evaluation indicators include accuracy, recall rate, F1 score, and AUC-ROC curve. The evaluation results are used as feedback. If the algorithm performance deteriorates or does not meet expectations, the learning rate is readjusted to improve the algorithm performance.

[0111] It should be noted that by adjusting the algorithm parameters in real time, the algorithm can be made more in line with the actual data distribution, thereby improving the accuracy of the algorithm's prediction of target user activities. By regularly evaluating the algorithm performance and making adjustments based on feedback, unnecessary waste of computing resources can be avoided and effective use of resources can be ensured.

[0112] This embodiment also provides a metaverse user behavior and data analysis system based on digital twins, including: a data acquisition module, a data preprocessing module, a feature extraction and fusion module, and a parameter update module;

[0113] The data acquisition module is used to obtain the target metaverse space activity data to be analyzed;

[0114] The data preprocessing module is used to preprocess the acquired activity data;

[0115] The feature extraction and fusion module is used to classify the pre-processed activity data for feature extraction and to fuse the data features from different sources;

[0116] The parameter update module is used to adjust the target user activity big data analysis algorithm in real time based on the analysis of fusion features.

[0117] This embodiment also provides a computer device, which is suitable for the case of a metaverse user behavior and data analysis method based on digital twins, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the metaverse user behavior and data analysis method based on digital twins as proposed in the above embodiment.

[0118] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0119] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by the processor, the method for realizing the metaverse user behavior and data analysis based on digital twins as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, SRAM for short), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, EEPROM for short), erasable programmable read-only memory (Erasable Programmable Read Only Memory, EPROM for short), programmable read-only memory (Programmable Red-Only Memory, PROM for short), read-only memory (Read-Only Memory, ROM for short), magnetic storage, flash memory, disk or optical disk.

[0120] In summary, the present invention ensures data security and user privacy through: real-time data collection and encryption through API interface, improves data quality and consistency through sophisticated preprocessing steps, comprehensively captures multi-dimensional behavioral characteristics of users through feature extraction and fusion of multimodal data, and improves the adaptability and accuracy of the algorithm by adjusting algorithm parameters in real time through adaptive learning rate method. These improvements not only improve the accuracy and comprehensiveness of user behavior analysis, but also ensure the real-time performance of the system, providing strong technical support for personalized services and user experience optimization in the metaverse.

[0121] Example 2, referring to Table 1, is the second example of the present invention. In order to further verify the technical solution of the present invention, experimental simulation data of the metaverse user behavior and data analysis method based on digital twins are provided.

[0122] First, through the API interface, the data generated by users is monitored and collected in real time, including text data, audio data, video data and image data.

[0123] All collected data is encrypted using the AES-256 algorithm and uploaded to the server. After the encrypted data is uploaded to the server, it is anonymized to remove the user's identity information, including the user name and registered account. Geocoding is used to blur the precise location generated by the data to the provincial and municipal levels, ensuring user privacy while retaining the necessary location information.

[0124] Next, the collected text data is used to remove non-alphanumeric characters in the text data through regular expressions, retaining English, numeric and Chinese characters, and removing URL links in the text.

[0125] All images are resized to the same size (224×224 pixels) to meet the input requirements of the deep learning model, and the image colors are normalized by subtracting the mean of each color channel. When the image size is larger than the target size, the central area is cropped from it as the final input.

[0126] Convert a video file into a series of static images, extract a fixed number of frames per second, and process each frame in the same way as image data.

[0127] Unify all audio files to the same sampling rate and use spectral gating to remove background noise from the audio.

[0128] Then, the TF-IDF algorithm is used to extract features from the preprocessed text data, and a trained deep learning model (such as ResNet-50) is used to extract image features. After preprocessing, frame images are obtained, and feature extraction is performed on the video frame images according to the image feature extraction method. The Mel frequency cepstral coefficients are used to extract the features of the audio data.

[0129] The extracted feature vectors of different types of data are fused into a unified feature vector and normalized.

[0130] Finally, the fused feature vector is input into the target user activity big data analysis algorithm and the result is output. The gradient of the loss function with respect to the algorithm parameters is calculated using the chain rule, and the algorithm parameters are updated using the gradient descent method. The Adam optimizer is used as an adaptive learning rate method to adaptively adjust the learning rate to minimize the error and obtain the final algorithm parameters. The performance of the algorithm is regularly evaluated, and the evaluation indicators include accuracy, recall, F1 score, and AUC-ROC curve. The evaluation results are used as feedback. If the algorithm performance degrades or does not meet expectations, the learning rate is readjusted to improve the algorithm performance.

[0131] The details are shown in Table 1 below:

[0132] Table 1 Performance comparison between traditional method and the method of the present invention

[0133] Test subjects Accuracy (%) Recall rate (%) F1 score (%) AUC-ROC curve Traditional methods 80 75 77.5 0.82 Method of the present invention 92 88 90 0.94

[0134] Through comparative analysis of the above experimental data, the method of the present invention extracts text features through the TF-IDF algorithm, extracts image and video frame features using ResNet-50, extracts audio features using Mel frequency cepstral coefficients, and fuses these feature vectors into a unified feature vector. This multimodal data fusion method fully utilizes the complementary information of different modal data, comprehensively captures the multi-dimensional behavioral characteristics of users, and improves the richness and representativeness of the features.

[0135] The method of the present invention adjusts the model parameters in real time through an adaptive learning rate method (such as Adam optimizer), thereby improving the adaptability and accuracy of the model. This method can dynamically adjust the learning rate, making the model more stable and efficient during the training process, and ensuring the real-time adaptability of the model to changes in user behavior.

[0136] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A metaverse user behavior and data analysis method based on digital twins, characterized by: include, Obtain the target metaverse space activity data to be analyzed; Preprocessing the acquired activity data; Classify the preprocessed activity data for feature extraction and fuse the data features from different sources; Based on the analysis of fusion features, the big data analysis algorithm of target user activities is adjusted in real time.

2. The digital twin-based metaverse user behavior and data analysis method according to claim 1, characterized in that: The specific steps of obtaining the target metaverse space activity data to be analyzed are as follows: Through the API interface, real-time monitoring and collection of user-generated data, including text data, audio data, video data, and image data; The data is encrypted using the AES-256 algorithm and uploaded to the server; After the encrypted data is uploaded to the server, it is anonymized to remove the user's identity information, including user name and registered account; Use geocoding to blur the precise location of the data to the province or city level.

3. The digital twin-based metaverse user behavior and data analysis method according to claim 2, characterized in that: The specific steps of preprocessing the acquired activity data are as follows: Use regular expressions to remove non-alphanumeric characters from text data, retain English, numeric, and Chinese characters, and remove URL links from text; All images were resized to the same size (224 × 224 pixels) to meet the input requirements of the deep learning algorithm, and the image colors were normalized by subtracting the mean of each color channel; When the image size is larger than the target size, the central area is cropped from it as the final input; Convert the video file into a series of static images, extract a fixed number of frames per second, and process each frame in the same way as image data; Unify all audio files to the same sampling rate and use spectral gating to remove background noise from the audio.

4. The digital twin-based metaverse user behavior and data analysis method according to claim 3, characterized in that: The pre-processed activity data is classified for feature extraction, and data features from different sources are fused. The specific steps are as follows: Use the TF-IDF algorithm to extract features from the preprocessed text data. The expression is: Among them, TF-IDF(t,d,D) is the extracted text data feature vector, t is a word, d is a document, D is a document set, count(t i ) is the word t i The number of times it appears in document d, max j (count(t j )) is the word t that appears most frequently in document d j , N is the total number of documents, n is the total number of words in document d, docFreq(t) is the number of words containing word t i The number of documents, i and j are index variables; Using the trained deep learning algorithm, the image is input into ResNet-50 to obtain the feature vector, which is expressed as follows: V = ResNet-50(I); Among them, V is the output feature vector, and I is the input image; The video data is preprocessed to obtain frame images, and the features of the video frame images are extracted according to the image feature extraction method. The expression is as follows: V f =ResNet-50(I f ); f=1,2,3....F; Among them, V f is the feature vector of the f-th frame image, I f is the f-th frame image of the video, and F is the total number of frame images extracted; The Mel frequency cepstral coefficient is used to extract the characteristics of audio data. The expression is as follows: M = MFCC(A); Where A is the audio signal, M is the extracted audio signal feature vector, and MFCC is the Mel frequency cepstrum coefficient; The feature vectors of different types of data are merged into a unified feature vector.

5. The digital twin-based metaverse user behavior and data analysis method according to claim 4, characterized in that: The specific steps of fusing the feature vectors of different types of data into a unified feature vector are as follows: The feature vectors of the extracted data of different types are fused into a unified feature vector and normalized. The expression is as follows: Among them, S is the fusion feature vector, ‖T‖2, ‖V‖2, ‖V f ‖2, ‖M‖2 are the text feature vector T, image feature vector V, and video feature vector V f , the L2 norm of the audio feature vector M, x is the kth element in the vector X, and L is the length of the vector.

6. The digital twin-based metaverse user behavior and data analysis method according to claim 5, characterized in that: The analysis based on fusion features adjusts the target user activity big data analysis algorithm in real time. The specific steps are as follows: Input the fused feature vector into the target user activity big data analysis algorithm and output the result. The expression is as follows: in, is the predicted value of the algorithm, h is the prediction function of the algorithm, and θ is the algorithm parameter; Use the chain rule to calculate the gradient of the loss function with respect to the algorithm parameters, and update the algorithm parameters by gradient descent. The expression is as follows: Among them, Q is the number of fusion features, is the i-th predicted value, y i is the i-th true value, S i is the i-th fused feature vector, i is the index variable, θ′ is the updated algorithm parameter, α is the learning rate, and controls the step size of parameter update. is the gradient of the loss function with respect to the algorithm parameters; The Adam optimizer is used as the adaptive learning rate method to adaptively adjust the learning rate to minimize the error and obtain the final algorithm parameters. The algorithm finally reaches the optimal solution.

7. The digital twin-based metaverse user behavior and data analysis method according to claim 6, characterized in that: Use the Adam optimizer as the adaptive learning rate method to adaptively adjust the learning rate. The specific steps are as follows: Through the Adam optimizer, the learning rate is adaptively updated to further improve the algorithm parameter update and obtain the final algorithm parameters. The expression is as follows: Among them, θ″ is the final algorithm parameter, is the bias-corrected second-order moment estimate, is the first-order moment estimate after bias correction, ∈ is a constant; Regularly evaluate the performance of the algorithm. Evaluation indicators include accuracy, recall rate, F1 score, and AUC-ROC curve. Use the evaluation results as feedback. If the algorithm performance deteriorates or does not meet expectations, readjust the learning rate to improve the algorithm performance.

8. A digital twin-based metaverse user behavior and data analysis system, based on the digital twin-based metaverse user behavior and data analysis method according to any one of claims 1 to 7, characterized in that: Including data acquisition module, data preprocessing module, feature extraction and fusion module and parameter updating module; The data acquisition module is used to obtain the target metaverse space activity data to be analyzed; The data preprocessing module is used to preprocess the acquired activity data; The feature extraction and fusion module is used to classify the pre-processed activity data for feature extraction and fuse the data features from different sources; The parameter updating module is used to adjust the target user activity big data analysis algorithm in real time based on the analysis of fusion features.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, it implements the steps of the metaverse user behavior and data analysis method based on digital twins as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the metaverse user behavior and data analysis method based on digital twins are implemented in any one of claims 1 to 7.

Images