An artificial intelligence-based 5G message processing method, system and storage medium

Abstract

The application discloses a 5G message processing method and system based on artificial intelligence and a storage medium, relates to the technical field of data processing, and comprises the following steps: analyzing modal type message data to set a time interception window, comprehensively setting a time interception window for previous message conditions of a user and modal type message data conditions, the time window can preliminarily intercept and divide the messages that are relatively associated and continuous in time, and provides a reliable basis for subsequent message unification and association. A cross-modal alignment model is constructed, features under all modal types are uniformly aligned through the cross-modal alignment model, the features are subjected to correlation analysis, value information is extracted on the basis of the correlation information, the reliability and depth of message processing are improved, deep mining and analysis of the messages are completed, and the optimized development of the message service is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and particularly relates to a 5G message processing method and system based on artificial intelligence and a storage medium. BACKGROUND

[0002] As an upgraded form of short messages, 5G messages support various modal formats such as text, pictures, audio, and video, and have the characteristics of rich media, no need to install APP, strong interaction, high security, and wide coverage. These characteristics enable 5G messages to provide richer information display and interaction experience and meet the diversified communication needs of users. At the same time, the application of artificial intelligence technology further improves the processing efficiency and intelligent level of 5G messages, and realizes functions such as automatic classification, intelligent reply, and emotion perception of messages through natural language processing and machine learning, thereby bringing users a more convenient, efficient, and intelligent information service experience.

[0003] In the prior art, current message processing often only stays at the surface information level, such as simple transmission of text, picture, audio, and video content, and lacks deep mining of message data. This results in poor reliability and low depth of message processing, which cannot meet the growing demand for message business analysis.

[0004] Therefore, how to improve the reliability and depth of message processing is a technical problem to be solved at present. SUMMARY

[0005] The present application relates to the technical field of data processing, and particularly relates to a 5G message processing method based on artificial intelligence, which comprises,

[0006] Collecting message data generated by users on a message platform, distinguishing modal types of the message data, and analyzing the modal type message data to set a time interception window;

[0007] Intercepting the message data according to the time interception window, generating a message data segment, and extracting features under each modal type in the message data segment;

[0008] Constructing a cross-modal alignment model, and uniformly aligning the features under all modal types through the cross-modal alignment model;

[0009] Correlation analysis is performed on the unified features to obtain a correlated message data set, value information is extracted from the correlated message data set, and the value information is visually displayed.

[0010] In some embodiments of the present application, analyzing the modal type message data to set the time interception window comprises,

[0011] Establish a message data timeline based on the timestamps of user message data, and determine the relative activity of users in different time periods;

[0012] The time truncation window length is set based on the relative activity of users at different times and the frequency of actual modal type message data at different times.

[0013] In some embodiments of this application, a message data timeline is established based on the timestamps of user message data, and the relative activity of users in different time periods is determined, including,

[0014] Establish a message data timeline based on the timestamps of user message data, mark the message interaction parameters at the corresponding positions on the message data timeline, and split the overall message data timeline into multiple message data timeline segments according to the total length of the message data timeline;

[0015] A standard time interval is set by the mean and standard deviation of the time intervals between previous message data in each message data time segment. The standard time interval is used to distinguish between two interaction types: continuous message interaction and non-continuous message interaction in each message data time segment.

[0016] For both continuous and non-continuous message interaction, the activity level of each type is determined based on the message interaction parameters, and the relative activity level of users in different time periods is determined.

[0017] ;

[0018] in, For users Relative activity levels over time periods , 2 represents the number of messages in both continuous and non-continuous message interaction types. , The first The first consecutive message interaction and the first The combined weight of non-continuous message interactions, , The first The first consecutive message interaction and the first The activity level of non-continuous message interactions , These are the conversion coefficients for the number of consecutive messages in a message exchange and the average interval between messages, respectively. , They are respectively in The next period The number of consecutive messages in a message interaction and the average interval between messages. a first constant under a time period. a first constant under a time period.

[0019] In some embodiments of the present application, the length of the time interception window is set based on the relative activity of the user under different time periods and the frequency of the actual modal type message data under different time periods, including,

[0020] The frequency of occurrence of each modal type message data under different time periods is calculated, and the frequency of occurrence of all modal type message data is integrated to obtain the frequency of modal type message data.

[0021] The length of the time interception window is determined according to the relative activity and the frequency of modal type message data under the same time period.

[0022] In some embodiments of the present application, a cross-modal alignment model is constructed, including,

[0023] Collecting features under all modal types, determining the semantic content of the features under each modal type, and dividing the features under the modal types into positive sample pairs and negative sample pairs through the semantic content of the features, the positive sample pairs and the negative sample pairs each including a combination of features under different modal types.

[0024] According to the positive sample pairs and the negative sample pairs, a training set and a verification set are constructed, the parameters of the model are initialized through the training set, the loss between the model output and the real label is described through the contrast loss function, the parameters of the initialized model are updated through the gradient descent algorithm and the evaluation index of the initialized model on the verification set, to minimize the contrast loss function, to realize the cross-modal alignment model.

[0025] In the process of updating the parameters of the initialized model, the learning rate in the gradient descent algorithm is dynamically adjusted through the learning rate scheduler to improve the verification effect.

[0026] In some embodiments of the present application, in the process of updating the parameters of the initialized model, the learning rate in the gradient descent algorithm is dynamically adjusted through the learning rate scheduler, including,

[0027] The target value of the evaluation index is determined by analyzing the positive sample pairs and the negative sample pairs in the verification set, the deviation amount is obtained according to the actual value and the target value of the evaluation index, the learning rate is adjusted through the deviation amount, and the evaluation index includes the loss value and the accuracy rate.

[0028] ;

[0029] wherein, is the adjusted learning rate, is the initial learning rate, , and are the influence weights of the loss value and the accuracy rate respectively, , respectively are loss value and accuracy, , respectively are loss value and accuracy, is a second constant.

[0030] In some embodiments of the application, the unified features are analyzed for association to obtain an associated message data set, and value information is extracted from the associated message data set, including,

[0031] The Apriori algorithm is used for association analysis to obtain an associated message data set, and the value information is extracted from the associated message data set by using a data processing tool.

[0032] Correspondingly, the application also provides a 5G message processing system based on artificial intelligence, comprising,

[0033] The first module is used for collecting message data generated by users on a message platform, distinguishing modal types of the message data, and setting a time interception window by analyzing the modal type message data.

[0034] The second module is used for intercepting the message data according to the time interception window to generate message data segments, and extracting features under each modal type in the message data segments.

[0035] The third module is used for constructing a cross-modal alignment model, and aligning the features under all modal types by using the cross-modal alignment model.

[0036] The fourth module is used for analyzing the unified features for association to obtain an associated message data set, extracting value information from the associated message data set, and visually displaying the value information.

[0037] The application further provides a computer readable storage medium storing one or more programs, which, when executed by an electronic device comprising a plurality of application programs, cause the electronic device to execute the above method.

[0038] Compared with the prior art, the application has the following beneficial effects:

[0039] 1. The time interception window is set by analyzing the modal type message data, the past message situation of the user and the modal type message data situation are comprehensively set, the time window can preliminarily intercept and divide the time-continuous and relatively associated messages, and a reliable basis is provided for subsequent message unification and association.

[0040] 2. The cross-modal alignment model is constructed, the features of all modal types are uniformly aligned through the cross-modal alignment model, the features are analyzed in association, the valuable information is extracted based on the association information, the reliability and depth of message processing are improved, the deep mining and analysis of the message are completed, and the optimized development of the message service is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 A flowchart of a 5G message processing method based on artificial intelligence is provided.

[0042] Figure 2 A structural diagram of a 5G message processing system based on artificial intelligence is provided. DETAILED DESCRIPTION

[0043] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only some of the embodiments of the application, not all the embodiments.

[0044] REFERENCE Figure 1 A 5G message processing method based on artificial intelligence, comprising,

[0045] Step S101: Collecting message data generated by users on a message platform, distinguishing the modal types of the message data, and analyzing the modal type message data to set a time interception window.

[0046] In this embodiment, a big data collection platform is used to obtain and integrate various types of data resources from the Internet through network crawler technology and other means. These data resources can include text, images, audio, video and other modalities. Data from different message platforms is integrated to form a unified data set. During the integration process, attention should be paid to the problems of data format unification, data deduplication and outlier processing. Modality refers to the expression or perception mode of data, such as text, image, audio, etc. In multi-modal data, each modality has its unique characteristics and representation. The time interception window is set in combination with the user's message interaction habits and the modal type message data, and by analyzing the characteristics of different modal type message data and considering user behavior, we can set a suitable time interception window to capture messages that are relatively continuous in time and may be associated. This will provide a good foundation for subsequent message association analysis.

[0047] It should be noted that the present solution mainly solves how to determine the association between messages, so as to more accurately extract valuable information and mine the deep meaning embodied by the messages.

[0048] In some embodiments of the present application, analyzing the modal type message data to set a time interception window comprises,

[0049] establish a message data timeline according to the timestamps of the user message data, and determine the relative activity of the user in different time periods;

[0050] set the time interception window length based on the relative activity of the user in different time periods and the frequency of actual modal type message data in different time periods.

[0051] In this embodiment, the message data timeline is established according to the past message situation of the user, and the activity of the user's message in different time periods is analyzed. The time interception window length is set in combination with the relative activity and the frequency of actual modal type message data in different time periods.

[0052] In some embodiments of the present application, the message data timeline is established according to the timestamps of the user message data, and the relative activity of the user in different time periods is determined, including,

[0053] The message data timeline is established according to the timestamps of the user message data, and the message interaction parameters are marked on the corresponding positions on the message data timeline. The overall message data timeline is split into multiple message data timeline segments according to the total length of the message data timeline;

[0054] The standard time interval is set through the mean and standard deviation of the time interval between the past message data of each message data timeline segment, and the standard time interval is used to distinguish the two interaction types of continuous message interaction and discontinuous message interaction on each message data timeline segment;

[0055] The activity of each type is determined according to the message interaction parameters for the two interaction types of continuous message interaction and discontinuous message interaction, and the relative activity of the user in different time periods is determined;

[0056] ;

[0057] wherein, is the relative activity of the user in the time period, , and 2 are the respective numbers of the two interaction types of continuous message interaction and discontinuous message interaction, , and are the combined weights of the first continuous message interaction and the first discontinuous message interaction, respectively, , and are the activities of the first continuous message interaction and the first discontinuous message interaction, respectively, , and​ These are the conversion coefficients for the number of consecutive messages in a message exchange and the average interval between messages, respectively. , They are respectively in The next period The number of consecutive messages in a message interaction and the average interval between messages. In order to be in The first constant over a given time period.

[0058] In this embodiment, message interaction parameters include the number of times messages are sent, received, and replied, as well as related parameters such as likes and comments. The mean and standard deviation of the time intervals are calculated. A suitable multiple (such as 1x, 2x, etc.) is selected, and the mean plus or minus the standard deviation of that multiple is used as the time interval threshold (standard time interval). Continuous message interaction refers to multiple messages whose time intervals are within the standard time interval, while non-continuous message interaction refers to a single message whose time intervals are outside the standard time interval. The activity level of each type is determined based on the message interaction parameters (using methods such as weighted summation). This indicates how the number of messages and the average interval between messages affect activity levels in continuous message interactions. Both the number of messages and the average interval reflect activity levels. This is to balance the size of the correction function.

[0059] In some embodiments of this application, the time truncation window length is set based on the user's relative activity level at different time periods and the frequency of actual modal type message data at different time periods, including:

[0060] Calculate the frequency of each modality message data occurrence in different time periods, and combine the occurrence frequencies of all modality message data to obtain the frequency of modality message data;

[0061] The time truncation window length is determined based on the relative activity and frequency of modal type message data within the same time period.

[0062] In this embodiment, the length of the time capture window is set by combining relative activity and frequency to improve the accuracy of the capture.

[0063] Step S102: Extract message data according to the time truncation window to generate message data segments, and extract features for each modality type in the message data segments.

[0064] In this embodiment, the message data is intercepted by setting a time window and using window subclassing and other methods to intercept the message data. The message data in each data segment is ensured to be continuous in time. The length of the time interception window may vary in different time periods. The generated message data segments are numbered, stored and managed for subsequent feature extraction and correlation analysis.

[0065] Text modality: TF-IDF (Term Frequency-Inverse Document Frequency) feature extraction method can be used to convert text data into feature vectors. In addition, word embedding (such as Word2Vec, BERT, etc.) technology can be used to extract deeper text features.

[0066] Image modality: HOG (Histogram of Oriented Gradients) feature extraction method can be used to capture local shape and texture information of images. In addition, convolutional neural networks (CNN) can be used to extract deeper image features.

[0067] Audio modality: MFCC (Mel Frequency Cepstral Coefficient) feature extraction method can be used to convert audio signals into feature vectors. In addition, recurrent neural networks (RNN) or long short-term memory networks (LSTM) can be used to extract deeper audio features.

[0068] Step S103, constructing a cross-modal alignment model, and aligning all modal type features through the cross-modal alignment model.

[0069] In this embodiment, because the message involves multiple modal content, cross-modal alignment of features of different modalities is needed to establish a unified standard and space for subsequent comparison and correlation analysis. Cross-modal contrast learning method, such as UNIMO framework, is used to construct a cross-modal alignment model. The model obtains context-based feature representation through self-attention mechanism, and realizes alignment between features of different modalities. Features of different modal types are input into the cross-modal alignment model, and through learning and optimization of the model, features of different modalities are mapped into a unified semantic space to realize unified alignment of features.

[0070] In some embodiments of the present application, the cross-modal alignment model is constructed, including,

[0071] Collecting features of all modal types, determining semantic content of features of each modal type, dividing features of modal types into positive sample pairs and negative sample pairs through semantic content of features, and positive sample pairs and negative sample pairs both include feature combinations of different modal types;

[0072] The training set and the validation set are respectively constructed according to the positive sample pairs and the negative sample pairs, the parameters of the initialized model are initialized through the training set, the loss between the model output and the real label is described through the contrast loss function, the parameters of the initialized model are updated through the gradient descent algorithm and the evaluation index of the initialized model on the validation set, the contrast loss function is minimized to realize the cross-modal alignment model.

[0073] In the process of updating the parameters of the initialized model, the learning rate in the gradient descent algorithm is dynamically adjusted through the learning rate scheduler to improve the validation effect.

[0074] In this embodiment, the positive sample pairs are feature pairs from different modalities but representing the same semantic information. The model should make the distance between these sample pairs in the feature space smaller to strengthen their similarity. The negative sample pairs are feature pairs from different modalities and representing different semantic information. The model should make the distance between these sample pairs in the feature space larger to distinguish the difference between them. The contrast loss function is a function that measures the difference between the model's predicted value and the actual value, which quantifies the performance of the model by calculating the distance between these sample pairs in the feature space. The goal of the contrast loss function is to find the model parameters that minimize the loss function value. This is usually achieved through optimization algorithms such as gradient descent. During the training process, the model will continuously adjust the parameters according to the gradient information of the loss function to minimize the distance between the positive sample pairs and maximize the distance between the negative sample pairs.

[0075] It can be understood that the contrast loss function and the gradient descent algorithm are relatively common methods in the field, and the specific content will not be repeated here, and other algorithms that can be completed also belong to the protection scope of the present application.

[0076] In some embodiments of the present application, in the process of updating the parameters of the initialized model, the learning rate in the gradient descent algorithm is dynamically adjusted through the learning rate scheduler, including,

[0077] The target value of the evaluation index is determined by analyzing the positive sample pairs and the negative sample pairs in the validation set, the deviation amount is obtained according to the actual value and the target value of the evaluation index, the learning rate is adjusted through the deviation amount, and the evaluation index includes the loss value and the accuracy rate;

[0078] ;

[0079] wherein, is the adjusted learning rate, is the initial learning rate, , are the influence weights of the loss value and the accuracy rate respectively, , are the loss value and the accuracy rate respectively, is the second constant.

[0080] In this embodiment, modern deep learning frameworks (such as TensorFlow, PyTorch, etc.) provide various learning rate schedulers (such as ReduceLROnPlateau, etc.). These schedulers can dynamically adjust the learning rate according to the changes in the indicators on the validation set, thereby optimizing the training process. The loss value is an important indicator to measure the difference between the predicted value and the actual value of the model. During the training process, if the loss value on the validation set remains high or no longer decreases, it may mean that the model has fallen into a local optimum or the learning rate is too large, causing oscillation. At this time, reducing the learning rate may help the model to jump out of the local optimum or stabilize the training process. The accuracy rate is an important indicator to measure the classification performance of the model. During the training process, if the accuracy rate on the validation set no longer improves, it may mean that the model has converged or the learning rate is too small, causing the training speed to be too slow. At this time, adjusting the learning rate may help the model to continue optimization or speed up the convergence speed.

[0081] In this embodiment, according to the characteristics of the specific task and the validation data set, reasonable target values or target change degrees are set for the loss value and the accuracy rate. The target values here include parameter values and parameter change rates. The correction of the initial learning rate by the deviation of the loss value and the accuracy rate, is to balance the size of the correction function.

[0082] In step S104, the unified features are subjected to association analysis to obtain an associated message data set, value information is extracted from the associated message data set, and the value information is visually displayed.

[0083] In this embodiment, association analysis algorithms such as Apriori algorithm and FP-growth algorithm are used to perform association analysis on the unified features. These algorithms can find frequent item sets and association rules in the data and reveal the relationships between different modal features. The results of the association analysis are stored and managed for subsequent value information extraction and visual display. Value information extraction: Through data mining techniques such as machine learning algorithms, the associated message data set is analyzed to extract valuable information. These information can include trend prediction, anomaly detection, user behavior analysis, etc. Visual display: Use data visualization technology such as Smartbi platform to display the extracted value information in graphical form. Through charts, graphs, etc. to present the relationship and trend between data, helping users better understand and utilize data.

[0084] In this embodiment, the surface information refers to the information directly presented by the message content itself, such as text description, image display, etc., which is intuitive and easy to understand, but often lacks deep correlation and interpretation. The value information is hidden under the surface information and needs to be excavated and analyzed by artificial intelligence technology to reveal it. These value information may include user's consumption habits, interest preferences, behavior patterns, etc., which have high commercial value and social value for enterprises and governments.

[0085] In some embodiments of the application, the unified features are analyzed for association to obtain an associated message data set, and value information is extracted from the associated message data set, including,

[0086] The Apriori algorithm is used for association analysis to obtain an associated message data set, and the value information is extracted from the associated message data set by using a data processing tool.

[0087] Correspondingly, the application also provides a 5G message processing system based on artificial intelligence, as shown in Figure 2 The system includes,

[0088] The first module is configured to collect message data generated by users on a message platform, distinguish the message data by modal type, and set a time cutting window by analyzing the modal type message data.

[0089] The second module is configured to cut the message data according to the time cutting window to generate message data segments, and extract features under each modal type in the message data segments.

[0090] The third module is configured to construct a cross-modal alignment model, and unify and align the features under all modal types by using the cross-modal alignment model.

[0091] The fourth module is configured to analyze the unified features for association to obtain an associated message data set, extract value information from the associated message data set, and visually display the value information.

[0092] The application further provides a computer readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of application programs, cause the electronic device to perform the above method.

[0093] Compared with the prior art, the application has the following advantages:

[0094] 1. Analyzing the modal type message data to set a time interception window, setting a time interception window according to the past message situation of the user and the modal type message data situation, the time window can preliminarily intercept and divide the messages which are relatively associated in time, and provides a reliable basis for subsequent message unification and association.

[0095] 2. Building a cross-modal alignment model, aligning the features under all modal types through the cross-modal alignment model, performing correlation analysis on the features, extracting valuable information on the basis of the correlation information, improving the reliability and depth of message processing, completing deep mining and analysis of the message, and ensuring the optimized development of the message service.

[0096] Through the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by hardware, or by means of software and necessary general hardware platform. Based on this understanding, the technical solutions of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a U disk, a mobile hard disk, etc.), and includes a plurality of instructions for making a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in various embodiments of the present application.

[0097] Those skilled in the art can understand that the accompanying drawings are only a schematic diagram of a preferred embodiment, and the modules or processes in the drawings are not necessarily required for implementing the present application.

[0098] Those skilled in the art can understand that the modules in the system in the embodiments can be distributed in the system according to the description of the embodiments, or can be changed and located in one or more systems different from the embodiments. The modules in the above embodiments can be combined into one module, or can be further split into multiple sub-modules.

[0099] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art can make equivalent replacements or changes to the technical solutions and inventive concepts of the present application within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A 5G message processing method based on artificial intelligence, characterized in that, include, Collect message data generated by users on the messaging platform, distinguish message data by modality, and analyze the modality-type message data to set the time capture window; The message data is truncated according to the time truncation window to generate message data segments, and features under each modality type in the message data segments are extracted; Construct a cross-modal alignment model to uniformly align features across all modalities; The unified features are subjected to correlation analysis to obtain a set of correlated message data. Valuable information is extracted from the correlated message data set and then visualized. in, Analyze modal message data to set the time capture window, including, Establish a message data timeline based on the timestamps of user message data, and determine the relative activity of users in different time periods; The time truncation window length is set based on the relative activity of users at different times and the frequency of actual modal type message data at different times.

2. The 5G message processing method based on artificial intelligence according to claim 1, characterized in that, Establish a message data timeline based on the timestamps of user message data, and determine the relative activity of users in different time periods. include, Establish a message data timeline based on the timestamps of user message data, mark the message interaction parameters at the corresponding positions on the message data timeline, and split the overall message data timeline into multiple message data timeline segments according to the total length of the message data timeline; A standard time interval is set by the mean and standard deviation of the time intervals between previous message data in each message data time segment. The standard time interval is used to distinguish between two interaction types: continuous message interaction and non-continuous message interaction in each message data time segment. For both continuous and non-continuous message interaction, the activity level of each type is determined based on the message interaction parameters, and the relative activity level of users in different time periods is determined. ; in, For users Relative activity levels over time periods , 2 represents the number of messages in both continuous and non-continuous message interaction types. , The first The first consecutive message interaction and the first The combined weight of non-continuous message interactions, , The first The first consecutive message interaction and the first The activity level of non-continuous message interactions , These are the conversion coefficients for the number of consecutive messages in a message exchange and the average interval between messages, respectively. , They are respectively in The next period The number of consecutive messages in a message interaction and the average interval between messages. In order to be in The first constant over a given time period.

3. The 5G message processing method based on artificial intelligence according to claim 1, characterized in that, The time truncation window length is set based on the user's relative activity level at different times and the frequency of actual modal type message data at different times. include, Calculate the frequency of each modality message data occurrence in different time periods, and combine the occurrence frequencies of all modality message data to obtain the frequency of modality message data; The time truncation window length is determined based on the relative activity and frequency of modal type message data within the same time period.

4. The 5G message processing method based on artificial intelligence according to claim 1, characterized in that, Constructing a cross-modal alignment model, including, Collect features from all modalities, determine the semantic content of features in each modality, and divide features in each modality into positive and negative sample pairs based on the semantic content of the features. Both positive and negative sample pairs include combinations of features from different modalities. Training and validation sets are constructed based on positive and negative sample pairs, respectively. The parameters of the model are initialized using the training set. The loss function is used to describe the loss between the model output and the true label. The parameters of the initialized model are updated using the gradient descent algorithm and the evaluation metric of the initialized model parameters on the validation set to minimize the contrastive loss function and achieve a cross-modal alignment model. During the process of updating the parameters of the initial model, the learning rate in the gradient descent algorithm is dynamically adjusted through the learning rate scheduler to improve the validation effect.

5. The 5G message processing method based on artificial intelligence according to claim 4, characterized in that, During the process of updating the parameters of the initial model, the learning rate in the gradient descent algorithm is dynamically adjusted through the learning rate scheduler. include, The target value of the evaluation index is determined by analyzing positive and negative sample pairs in the validation set. The deviation is obtained based on the actual value and the target value of the evaluation index. The learning rate is adjusted by the deviation. The evaluation index includes the loss value and the accuracy. ； in, The adjusted learning rate, The initial learning rate, , These are the respective weights of the impact on loss and accuracy. , These are the loss value and accuracy, respectively. It is the second constant.

6. The 5G message processing method based on artificial intelligence according to claim 1, characterized in that, Correlation analysis is performed on the unified features to obtain a set of associated message data. Valuable information is then extracted from this set of associated message data, including... The Apriori algorithm is used to perform association analysis to obtain a set of associated message data. Valuable information is then extracted from this set of associated message data using data processing tools.

7. A 5G message processing system based on artificial intelligence, characterized in that, The system is used to implement the AI-based 5G message processing method as described in any one of claims 1-6, the system comprising: The first module is used to collect message data generated by users on the messaging platform, distinguish the message data by modal type, and analyze the modal type message data to set the time capture window; The second module is used to extract message data according to the time truncation window, generate message data segments, and extract features under each modality type in the message data segments; The third module is used to build a cross-modal alignment model, which performs unified alignment of features across all modalities. The fourth module is used to perform correlation analysis on the unified features to obtain a set of associated message data, extract valuable information from the set of associated message data, and visualize the valuable information.

8. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of applications, cause the electronic device to perform the method of any one of claims 1-6.

Images