A method and system for identifying public opinion information based on deep learning

Through the combination of deep learning and Transformer model and periodic analysis, the problem of semantic information capture and time point correlation in public opinion data is solved, and the accuracy and real-timeness of public opinion trend prediction are improved.

CN118152566BActive Publication Date: 2024-11-22WEIBOREACH COM
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202410240298.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-03-04
Publication Date
2024-11-22
Estimated Expiration
2044-03-04

AI Technical Summary

Technical Problem

The prior art is difficult to capture the rich semantic information in text data when processing unstructured and diversified large-scale public opinion data, and ignores the correlation between various time points, resulting in insufficient accuracy and real-time prediction of public opinion trends.

Method used

Using a deep learning-based method, the Transformer model is used to preprocess and feature extraction of public opinion data, combined with trend analysis and periodic analysis, comprehensive time series features are constructed, and the model is optimized through real-time update mechanism to improve the accuracy of public opinion trend prediction. and real-time.

Benefits of technology

Through the integration of trend and periodic analysis, the semantic understanding ability and timing correlation of public opinion data are improved, and the accuracy and real-timeness of public opinion trend prediction are significantly improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118152566B_ABST
    Figure CN118152566B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and system for identifying public opinion information based on deep learning, which relates to the field of data identification technology, including preprocessing based on collected public opinion data, using a transformer model to learn the public opinion data and extract public opinion features, and constructing a public opinion recognition model; fusing the public opinion features and obtaining comprehensive time series features; using the public opinion recognition model to predict the comprehensive time series features to obtain the future trend of the public opinion data; and updating the public opinion recognition model based on a real-time update mechanism. The present invention uses a transformer model to learn and extract features from public opinion data, fuses trend analysis and periodicity analysis to obtain comprehensive time series features, more accurately analyzes the development direction of public opinion, identifies periodic fluctuations of public opinion, has stronger semantic understanding capabilities and time series correlation, and improves the prediction accuracy and real-time performance of public opinion trends.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of data recognition technology, and in particular to a method and system for identifying public opinion information based on deep learning. Background Art

[0002] With the rapid development of the Internet, the collection and analysis of public opinion information has become increasingly important in the fields of social management and market forecasting. Traditional public opinion analysis methods are mainly based on rules and statistical models. However, with the rapid increase in the amount of information and the diversification of information forms, these methods have gradually revealed their limitations when processing large-scale heterogeneous data and complex text information. In recent years, deep learning technology has achieved remarkable results in the fields of natural language processing and data mining, and has become an effective means to solve the problem of complex text information processing. Among them, the Transformer model is a deep learning model based on the attention mechanism. It has attracted much attention due to its superior performance in processing sequence data. The attention mechanism of the Transformer model allows the model to focus on information at different positions when processing the input sequence, thereby better capturing the semantic relationship in the text. By introducing the Transformer model, it becomes more efficient to learn key information and features in massive public opinion data, so that public opinion analysis can more comprehensively understand the emotions, topics and events in the text, thereby improving the accuracy and depth of the analysis. In addition, the extraction of comprehensive time series features and the prediction of public opinion trends have also become research hotspots in the current field of public opinion information processing.

[0003] Although traditional public opinion analysis methods and some machine learning-based models have achieved certain results in certain scenarios, they still have some shortcomings when processing unstructured and diverse large-scale public opinion data. First, traditional methods often find it difficult to capture the rich semantic information in text data, and have limited processing effects on long texts. Second, these methods often ignore the correlation between various time points when processing time series information, which leads to certain difficulties in accurately predicting public opinion trends. Summary of the invention

[0004] In view of the problems that when predicting public opinion data, the existing technology is difficult to capture the rich semantic information in text data and lacks analysis of the correlation between various time points, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is how to provide a method for modeling the time series of public opinion data and improving the accuracy and real-time performance of predicting public opinion trends by utilizing comprehensive time series features.

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

[0007] In the first aspect, an embodiment of the present invention provides a method for identifying public opinion information based on deep learning, which includes preprocessing the collected public opinion data, using a transformer model to learn the public opinion data and extract public opinion features, and constructing a public opinion recognition model; fusing the public opinion features to obtain comprehensive time series features; using the public opinion recognition model to predict the comprehensive time series features to obtain future trends of the public opinion data; and updating the public opinion recognition model based on a real-time update mechanism.

[0008] As a preferred solution of the deep learning-based public opinion information identification method described in the present invention, wherein: the public opinion data is the time series data of public opinion; the time series data of public opinion includes the release time and update time of public opinion; the transformer model is used to learn the public opinion data and extract public opinion features, and feature extraction is performed on the time series data of public opinion by using trend analysis and periodicity analysis.

[0009] As a preferred solution of the method for identifying public opinion information based on deep learning described in the present invention, the trend analysis includes the following steps: using slope detection to analyze the changing trend of the time series data of public opinion; constructing a trend classifier to classify the detected changing trends; the calculation formula of the trend analysis is as follows:

[0010]

[0011] Among them, S(T) is the trend analysis score of the time series data of public opinion at time point T; D(T) is the value of the time series data of public opinion at time point T; k 1 is an adjustment parameter for trend analysis; the specific situation of the trend analysis is as follows: if the trend analysis score S(T) of the time series data of public opinion at time point T is less than or equal to the first threshold, then the trend of the time series data of public opinion is determined to be downward; if the trend analysis score S(T) of the time series data of public opinion at time point T is greater than the first threshold and less than the second threshold, then the trend of the time series data of public opinion is determined to be stable; if the trend analysis score S(T) of the time series data of public opinion at time point T is greater than the second threshold, then the trend of the time series data of public opinion is determined to be upward.

[0012] As a preferred solution of the method for identifying public opinion information based on deep learning described in the present invention, the periodic analysis includes the following steps: introducing Fourier transform to analyze the frequency domain information of the time series data of public opinion; constructing a periodicity analyzer to convert the frequency domain information obtained by Fourier transform into periodic information; classifying and identifying periodic information of different frequencies; the calculation formula of the periodic analysis is as follows:

[0013]

[0014] Where C(T) is the periodicity analysis score of the time series data of public opinion at time point T; N is the number of frequencies considered in Fourier transform; A n is the amplitude of the nth frequency; f n is the nth frequency; φ n is the phase of the nth frequency; if the periodic analysis score C(T) of the time series data of public opinion at time point T is greater than or equal to the third threshold, the periodic intensity is judged to be significant; if the periodic analysis score C(T) of the time series data of public opinion at time point T is less than the third threshold, the periodic intensity is judged to be insignificant.

[0015] As a preferred solution of the public opinion information identification method based on deep learning described in the present invention, the comprehensive time series feature is obtained by fusing the detected change trend and periodic information; the calculation formula of the comprehensive time series feature is as follows:

[0016] F(T)=α·S(T)+β·C(T)

[0017] Among them, F(T) is the score of the comprehensive time series characteristics; α is the weight parameter of trend analysis; β is the weight parameter of periodic analysis.

[0018] As a preferred solution of the method for identifying public opinion information based on deep learning described in the present invention, the future trend of the public opinion data is obtained by constructing the output layer of the public opinion identification model, and then predicting the future trend of the public opinion data; the calculation formula for predicting the future trend of the public opinion data is as follows:

[0019]

[0020] Among them, P(T) is the predicted value of future public opinion data; k 2 It is an adjustment parameter for predicting future trends of public opinion data.

[0021] As an optimal solution of the public opinion information identification method based on deep learning described in the present invention, the specific situation of predicting the future trend of the public opinion data is as follows: if the predicted value P(T) of the future public opinion data is less than the fourth threshold, then the probability of the prediction of the future trend of the public opinion data is low, that is, the future trend of the public opinion data is inconsistent with the past data; if the predicted value P(T) of the future public opinion data is equal to the fourth threshold, then the probability of the prediction of the future trend of the public opinion data is equal to the probability of not occurring, that is, there is a certain consistency between the future trend of the public opinion data and the past data; if the predicted value P(T) of the future public opinion data is greater than the fourth threshold, then the probability of the prediction of the future trend of the public opinion data is high, that is, the future trend of the public opinion data is consistent with the past data.

[0022] On the second aspect, in order to further solve the security problems existing in public opinion prediction, the embodiment of the present invention provides a public opinion information identification system based on deep learning, which includes: a model construction module, which is used to collect time series data of public opinion, and use the transformer Transformer model to perform trend analysis and periodicity analysis on the time series data of public opinion to construct a public opinion recognition model; a public opinion prediction module, which is used to fuse the public opinion features extracted by trend analysis and the public opinion features extracted by periodicity analysis to obtain comprehensive time series features, and predict the future trend of public opinion data; a model update module, which is used to update the parameters of the public opinion recognition model according to real-time public opinion data and user feedback.

[0023] 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 a public opinion information identification method based on deep learning as described in the first aspect of the present invention.

[0024] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of a method for identifying public opinion information based on deep learning as described in the first aspect of the present invention.

[0025] The beneficial effects of the present invention are as follows: the present invention utilizes the transformer model to learn and extract features of public opinion data, integrates trend analysis and periodic analysis to obtain comprehensive time series features, more accurately analyzes the development direction of public opinion, identifies the periodic fluctuations of public opinion, has stronger semantic understanding ability and time series correlation, and improves the prediction accuracy and real-time performance of public opinion trends. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0027] Figure 1 This is a flow chart for implementing the present invention in Example 1. DETAILED DESCRIPTION

[0028] 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.

[0029] 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.

[0030] 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.

[0031] Example 1

[0032] Reference Figure 1 , which is the first embodiment of the present invention, and provides a method for identifying public opinion information based on deep learning, comprising the following steps:

[0033] S1: Preprocess the collected public opinion data, use the Transformer model to learn the public opinion data and extract public opinion features, and build a public opinion recognition model.

[0034] Preferably, Figure 1 The figure shows the implementation process of the present invention, which collects time series data of public opinion, uses the transformer transform model to perform trend analysis and periodicity analysis on the public opinion data, obtains the trend characteristics and periodicity characteristics of the public opinion data and fuses them, and then obtains comprehensive time series characteristics. After that, the constructed public opinion recognition model is used to predict the public opinion data to obtain the corresponding future trend of the public opinion data. Finally, through the established real-time update mechanism, the parameters of the public opinion recognition model are updated using real-time public opinion data and user feedback.

[0035] Preferably, the public opinion data is time series data of public opinion.

[0036] Specifically, the time series data of public opinion includes the release time and update time of public opinion.

[0037] Preferably, the transformer model is used to learn the public opinion data and extract the public opinion features by extracting features from the time series data of the public opinion through trend analysis and periodic analysis.

[0038] Specifically, the trend analysis includes the following steps: using slope detection to analyze the changing trend of the time series data of public opinion.

[0039] Build a trend classifier to classify the detected change trends into rising, falling, and stable ones.

[0040] Furthermore, the calculation formula for trend analysis is as follows:

[0041]

[0042] Among them, S(T) is the trend analysis score of the time series data of public opinion at time point T, with a value range of [0,1], indicating the strength of the trend; D(T) is the value of the time series data of public opinion at time point T; k 1 It is the adjustment parameter for trend analysis.

[0043] Furthermore, the specific situation of the trend analysis is as follows: if the trend analysis score S(T) of the time series data of public opinion at the time point T is less than or equal to the first threshold, it is determined that the trend of the time series data of public opinion is downward.

[0044] If the trend analysis score S(T) of the time series data of public opinion at time point T is greater than the first threshold and less than the second threshold, it is determined that the trend of the time series data of public opinion is stable.

[0045] If the trend analysis score S(T) of the time series data of public opinion at the time point T is greater than the second threshold, it is determined that the trend of the time series data of public opinion is rising.

[0046] Specifically, the use of trend analysis helps capture the overall changing trends of public opinion information, helps identify factors that influence public sentiment, and provides a scientific basis for public opinion management.

[0047] Specifically, the periodicity analysis includes the following steps: introducing Fourier transform to analyze the frequency domain information of the time series data of public opinion.

[0048] Construct a periodicity analyzer to convert the frequency domain information obtained by Fourier transform into periodic information.

[0049] Classify and identify periodic information of different frequencies.

[0050] Furthermore, the calculation formula for periodic analysis is as follows:

[0051]

[0052] Where C(T) is the periodicity analysis score of the time series data of public opinion at time point T, indicating the intensity of periodicity; N is the number of frequencies considered in Fourier transform; A n is the amplitude of the nth frequency; f n is the nth frequency; φ n is the phase of the nth frequency.

[0053] Furthermore, if the periodic analysis score C(T) of the time series data of public opinion at time point T is greater than or equal to the third threshold, the periodicity intensity is judged to be significant; if the periodic analysis score C(T) of the time series data of public opinion at time point T is less than the third threshold, the periodicity intensity is judged to be insignificant.

[0054] Specifically, by utilizing periodic analysis, we can identify periodic fluctuations in public opinion information, help understand the cyclical changes in public opinion, and provide support for the formulation of targeted management strategies.

[0055] S2: Fusion of public opinion features to obtain comprehensive time series features.

[0056] Preferably, the comprehensive time series feature is obtained by fusing the detected change trend and periodic information.

[0057] Specifically, the calculation formula for the comprehensive time series characteristics is as follows:

[0058] F(T)=α·S(T)+β·C(T)

[0059] Among them, F(T) is the score of the comprehensive time series characteristics; α is the weight parameter of trend analysis; β is the weight parameter of periodic analysis.

[0060] Furthermore, by integrating trend analysis and periodic information to obtain comprehensive time series features, the characteristics of public opinion data can be more comprehensively reflected, thereby improving the robustness and accuracy of the public opinion recognition model.

[0061] S3: Use the public opinion recognition model to predict the comprehensive time series characteristics and obtain the future trend of public opinion data.

[0062] Preferably, the future trend of the public opinion data is obtained by constructing an output layer of a public opinion recognition model, thereby predicting the future trend of the public opinion data.

[0063] Specifically, the calculation formula for predicting the future trend of public opinion data is as follows:

[0064]

[0065] Among them, P(T) is the predicted value of future public opinion data, with a value range of [0,1], indicating the predicted probability of future public opinion; k 2 It is an adjustment parameter for predicting future trends of public opinion data.

[0066] Furthermore, the specific situation of predicting the future trend of public opinion data is as follows: if the predicted value P(T) of future public opinion data is less than the fourth threshold, it is determined that the probability of predicting the future trend of public opinion data is low, that is, the future trend of public opinion data is inconsistent with past data.

[0067] If the predicted value P(T) of future public opinion data is equal to the fourth threshold, it is determined that the probability of the prediction of the future trend of public opinion data occurring is equal to the probability of not occurring, that is, there is a certain consistency between the future trend of public opinion data and past data.

[0068] If the predicted value P(T) of the future public opinion data is greater than the fourth threshold, it is determined that the probability of the prediction of the future trend of the public opinion data is high, that is, the future trend of the public opinion data is consistent with the past data.

[0069] S4: Update the public opinion recognition model based on the real-time update mechanism.

[0070] Preferably, the real-time update mechanism includes the following steps: collecting the latest public opinion data in real time and performing preprocessing.

[0071] Formulate model parameter update strategies to update trend analysis parameters and periodic analysis parameters.

[0072] The output layer of the public opinion recognition model is adjusted according to the update of trend analysis parameters and periodic analysis parameters to obtain the future trend of new public opinion data.

[0073] Collect user feedback data including satisfaction with prediction results and prediction accuracy to adjust the parameters of the public opinion recognition model.

[0074] This embodiment also provides a public opinion information identification system based on deep learning, including: a model construction module, which is used to collect time series data of public opinion, use a transformer model to perform trend analysis and periodicity analysis on the time series data of public opinion, and build a public opinion identification model; a public opinion prediction module, which is used to fuse the public opinion features extracted by trend analysis and the public opinion features extracted by periodicity analysis to obtain comprehensive time series features, and predict the future trend of public opinion data; a model update module, which is used to update the parameters of the public opinion identification model according to real-time public opinion data and user feedback.

[0075] This embodiment also provides a computer device, which is suitable for a public opinion information identification method based on deep learning, 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 a public opinion information identification method based on deep learning as proposed in the above embodiment.

[0076] 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.

[0077] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, a method for identifying public opinion information based on deep learning is implemented as proposed in the above embodiment; 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, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, disk or optical disk.

[0078] In summary, the present invention utilizes the transformer model to learn and extract features of public opinion data, integrates trend analysis and periodic analysis to obtain comprehensive time series features, more accurately analyzes the development direction of public opinion, identifies the periodic fluctuations of public opinion, has stronger semantic understanding ability and time series correlation, and improves the prediction accuracy and real-time performance of public opinion trends.

[0079] Example 2

[0080] Referring to Tables 1 to 3, the second embodiment of the present invention is shown. This embodiment is different from the first embodiment in that, in order to verify its beneficial effects, operating data and related instructions of the present invention in an actual environment are provided.

[0081] This example collects public opinion time series data, including the release time and update time of public opinion, and then extracts features from the public opinion data through trend analysis and periodic analysis to predict future public opinion trends. Table 1 shows the trend analysis score and periodic analysis score of this example. It can be seen from the table that the present invention can more accurately capture the overall change trend and periodic fluctuations of public opinion information, and has obvious advantages in identifying factors that affect public sentiment and understanding the cyclical changes of public opinion.

[0082] Table 1 Public opinion feature extraction data table

[0083] Public opinion sample Trend analysis score Periodicity analysis score 1 0.82 0.76 2 0.64 0.72 3 0.88 0.82 ... ... ... n 0.78 0.68

[0084] As shown in Table 2, the comprehensive time series feature score of this example, the present invention more comprehensively reflects the characteristics of public opinion data by utilizing comprehensive time series features, improves the robustness and accuracy of the public opinion recognition model, and has better performance than single feature analysis.

[0085] Table 2 Comprehensive time series feature score data table

[0086] Public opinion sample Comprehensive time series feature score 1 0.79 2 0.68 3 0.85 ... ... n 0.73

[0087] Table 3 shows the prediction table of the future trend of public opinion data in this example.

[0088] Table 3 Future trend forecast of public opinion data

[0089] Public opinion sample Predicted value of future public opinion data 1 0.88 2 0.75 3 0.92 ... ... n 0.80

[0090] It can be seen from the above table that the present invention effectively predicts the future trend of public opinion data and improves the accuracy of public opinion prediction by utilizing the trend characteristics and periodic characteristics of public opinion data in combination with a real-time update mechanism.

[0091] 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 method for identifying public opinion information based on deep learning, characterized in that: include: Based on the collected public opinion data, the public opinion data is preprocessed, and the transformer model is used to learn and extract public opinion features from the public opinion data to build a public opinion recognition model; The public opinion features are integrated to obtain comprehensive time series features; Using the public opinion recognition model to predict the comprehensive time series features, and obtain the future trend of the public opinion data; Updating the public opinion recognition model based on a real-time update mechanism; Trend analysis includes the following steps: Use slope detection to analyze the changing trend of time series data of public opinion; Build a trend classifier to classify the detected change trends; The calculation formula for the trend analysis is as follows: Among them, S(T) is the trend analysis score of the time series data of public opinion at time point T; D(T) is the value of the time series data of public opinion at time point T; k1 is the adjustment parameter of trend analysis; The details of the trend analysis are as follows: If the trend analysis score S(T) of the time series data of public opinion at time point T is less than or equal to the first threshold, it is determined that the trend of the time series data of public opinion is downward; If the trend analysis score S(T) of the time series data of public opinion at time point T is greater than the first threshold and less than the second threshold, the trend of the time series data of public opinion is determined to be stable; If the trend analysis score S(T) of the time series data of public opinion at time point T is greater than the second threshold, it is determined that the trend of the time series data of public opinion is rising; Periodicity analysis includes the following steps: Introduce Fourier transform to analyze the frequency domain information of time series data of public opinion; Construct a periodicity analyzer to convert the frequency domain information obtained by Fourier transform into periodic information; Classify and identify periodic information of different frequencies; The calculation formula for the periodicity analysis is as follows: Where C(T) is the periodicity analysis score of the time series data of public opinion at time point T; N is the number of frequencies considered in Fourier transform; A n is the amplitude of the nth frequency; f n is the nth frequency; φ n is the phase of the nth frequency; If the periodicity analysis score C(T) of the time series data of public opinion at time point T is greater than or equal to the third threshold, it is determined that the periodicity intensity is significant; If the periodicity analysis score C(T) of the time series data of public opinion at time point T is less than the third threshold, it is determined that the periodicity intensity is not significant; The comprehensive time series feature is obtained by fusing the detected change trend and periodic information; The calculation formula of the comprehensive time series characteristics is as follows: F(T)=α·S(T)+β·C(T) Among them, F(T) is the score of the comprehensive time series characteristics; α is the weight parameter of trend analysis; β is the weight parameter of periodic analysis; The future trend of the public opinion data is obtained by constructing an output layer of the public opinion recognition model, thereby predicting the future trend of the public opinion data; The calculation formula for predicting the future trend of public opinion data is as follows: Among them, P(T) is the predicted value of future public opinion data; k2 is the adjustment parameter for predicting the future trend of public opinion data; The specific situation of predicting the future trend of public opinion data is as follows: If the predicted value P(T) of the future public opinion data is less than the fourth threshold, it is determined that the probability of the prediction of the future trend of the public opinion data is low, that is, the future trend of the public opinion data does not match the past data; If the predicted value P(T) of the future public opinion data is equal to the fourth threshold, it is determined that the probability of the prediction of the future trend of the public opinion data occurring is equal to the probability of not occurring, that is, there is consistency between the future trend of the public opinion data and the past data; If the predicted value P(T) of the future public opinion data is greater than the fourth threshold, it is determined that the probability of the prediction of the future trend of the public opinion data is high, that is, the future trend of the public opinion data is consistent with the past data.

2. The method for identifying public opinion information based on deep learning according to claim 1, characterized in that: The public opinion data is time series data of public opinion; The time series data of public opinion includes the release time and update time of public opinion; The transformer model is used to learn the public opinion data and extract public opinion features, and feature extraction is performed on the time series data of public opinion by using trend analysis and periodic analysis.

3. A public opinion information recognition system based on deep learning, based on a public opinion information recognition method based on deep learning according to any one of claims 1 to 2, characterized in that: include, The model building module is used to collect the time series data of public opinion, use the transformer model to perform trend analysis and periodicity analysis on the time series data of public opinion, and build a public opinion recognition model; The public opinion prediction module is used to integrate the public opinion features extracted by trend analysis and the public opinion features extracted by periodic analysis to obtain comprehensive time series features and predict the future trend of public opinion data; The model updating module is used to update the parameters of the public opinion recognition model based on real-time public opinion data and user feedback.

4. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a method for identifying public opinion information based on deep learning as described in any one of claims 1 to 2 are implemented.

5. 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 a method for identifying public opinion information based on deep learning as described in any one of claims 1 to 2 are implemented.

Citation Information

Cited By

  • Dynamic public opinion early warning method based on LSTM-XGB-RF

    CN120410288A

  • Public opinion event multi-mode semantic fusion modeling and abstract generation method and system

    CN121880552A