Intelligent information extraction system and method for computer network

Through deep learning technology, the feature vectors of food safety news and public opinion data are extracted to judge user emotions, and the problem of difficult monitoring of public opinion dynamics in food safety events is solved, real-time monitoring and risk control in the field of food safety is achieved.

CN119003764AInactive Publication Date: 2024-11-22JIANGSU FENGSAI INTELLIGENT TECHNOLOGY CO LTD

Patent Information

Application Number
CN202410900156.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-05
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Public opinion dynamics caused by food safety incidents in the network environment are difficult to capture in a timely and accurate manner, affecting public opinion management and risk control.

Method used

A computer network information intelligent extraction system was designed, using deep learning technology to obtain food safety-related news and public opinion data, extract information text to understand feature vectors and associated feature vectors, judge user emotions through classifiers, and provide real-time monitoring and decision makers. analyze.

Benefits of technology

Real-time monitoring of dynamics in the food safety field and emotional attitude judgment have been achieved, helping decision makers take timely measures to avoid potential harm, and improving the effectiveness of public opinion management and the guarantee of food safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119003764A_ABST
    Figure CN119003764A_ABST
Patent Text Reader

Abstract

The invention relates to the field of information extraction, and particularly discloses an intelligent information extraction system and method for a computer network, and the method comprises the steps: firstly obtaining news and public opinion data related to food safety, then carrying out the feature extraction and correlation analysis of the news and public opinion data through a deep learning technology, and finally carrying out the classification of the news and public opinion data through a classifier. According to the method, positive and negative emotion analysis is carried out on news, whether the emotion of a user is positive or negative is judged, so that decision makers are helped to better know the attitude and emotion of the public to food safety events, the dynamic state of the food safety field is monitored in real time, industry development and event changes are known in time, and then corresponding measures are taken to avoid harm.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of information extraction, and more specifically, to a system and method for intelligent information extraction in a computer network. Background Art

[0002] Food refers to various substances that are processed or directly collected and can be eaten by humans, including animal and plant foods, processed foods and beverages, etc., which are the necessary sources of nutrition for human life. There are many kinds of food, covering the diversity of human food culture, and also carrying multiple meanings such as health, safety and culture.

[0003] In recent years, food incidents have become hot topics in the online environment, attracting widespread attention, and food safety issues have gradually become a social focus. Timely capture of accurate and complete data information is crucial for analyzing public opinion dynamics. Obtaining comprehensive, accurate, interesting and effective public opinion information through in-depth analysis plays an important role in formulating effective public opinion management measures, such as importance assessment, handling suggestions, early warning release, post-event summary, risk control and scenario analysis.

[0004] Therefore, a system and method for intelligently extracting information from a computer network is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides a computer network intelligent information extraction system and method, which first obtains news and public opinion data related to food safety, then uses deep learning technology to perform feature extraction and association analysis, and finally uses a classifier to perform positive and negative sentiment analysis on the news to determine whether the user's emotions are positive or negative, so as to help decision makers better understand the public's attitude and emotions towards food safety incidents, thereby monitoring the dynamics of the food safety field in real time, timely understanding industry development and event changes, and then taking corresponding measures to avoid harm.

[0006] According to one aspect of the present application, a system for intelligently extracting information from a computer network is provided, comprising:

[0007] Food safety review data acquisition module, used to obtain news and public opinion data related to food safety;

[0008] A food safety comment data extraction module is used to extract food safety information text understanding feature vectors and food safety information association feature vectors from the food safety related news and public opinion data;

[0009] The user emotion judgment module is used to judge whether the user emotion is positive or negative based on the food safety information text understanding feature vector and the food safety information association feature vector.

[0010] According to another aspect of the present application, a method for intelligently extracting information from a computer network is provided, comprising:

[0011] Obtain news and public opinion data related to food safety;

[0012] Extracting food safety information text understanding feature vectors and food safety information association feature vectors from the food safety related news and public opinion data;

[0013] Based on the food safety information text understanding feature vector and the food safety information association feature vector, it is determined whether the user emotion is positive or negative.

[0014] Compared with the prior art, the present application provides a computer network intelligent information extraction system and method, which first obtains news and public opinion data related to food safety, then uses deep learning technology to perform feature extraction and association analysis, and finally uses a classifier to perform positive and negative sentiment analysis on the news to determine whether the user's emotions are positive or negative, so as to help decision makers better understand the public's attitude and emotions towards food safety incidents, thereby monitoring the dynamics of the food safety field in real time, promptly understanding industry development and event changes, and then taking corresponding measures to avoid harm. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram of a system for intelligent information extraction of a computer network according to an embodiment of the present application.

[0017] Figure 2 It is a block diagram of a food safety review data extraction module in the information intelligent extraction system of a computer network according to an embodiment of the present application.

[0018] Figure 3 It is a block diagram of a food safety review text convolution unit in an intelligent information extraction system of a computer network according to an embodiment of the present application.

[0019] Figure 4 It is a block diagram of a user emotion judgment module in the information intelligent extraction system of a computer network according to an embodiment of the present application.

[0020] Figure 5 The present invention is a flowchart of a method for intelligently extracting information from a computer network according to an embodiment of the present application.

[0021] Figure 6 is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0023] Figure 1 Schematic diagram of the information intelligent extraction system of the computer network of the embodiment of the present application. Figure 1 As shown, according to the computer network intelligent information extraction system 100 of the embodiment of the present application, it includes: a food safety comment data acquisition module 110, which is used to obtain news and public opinion data related to food safety; a food safety comment data extraction module 120, which is used to extract food safety information text understanding feature vectors and food safety information association feature vectors from the food safety related news and public opinion data; a user emotion judgment module 130, which is used to judge whether the user emotion is positive or negative based on the food safety information text understanding feature vector and the food safety information association feature vector.

[0024] In the above-mentioned computer network information intelligent extraction system 100, the food safety comment data acquisition module 110 is used to obtain news and public opinion data related to food safety. It should be understood that food, as an indispensable part of human daily life, includes various foods and beverages derived from animals and plants and processed, and is an important source of nutrition and life maintenance for people. There are many kinds of food, which represent the richness and diversity of human food culture, and also carry multiple meanings such as health, safety and culture. In recent years, food-related events have frequently become hot topics on the Internet, attracting widespread attention, and food safety issues have gradually become the focus of social attention. In this context, timely and accurate acquisition of complete data information is crucial for analyzing public opinion dynamics. Obtaining comprehensive, accurate, interesting and effective public opinion information through in-depth analysis is helpful to formulate effective public opinion management measures. The acquisition and analysis of such information can help managers better understand the trend of public opinion, make decisions and countermeasures in a timely manner, and thus better maintain food safety and social stability. Therefore, in the technical solution of this application, by obtaining news and public opinion data related to food safety, and combining deep learning technology, it is possible to judge whether the user's emotions are positive or negative, so as to help managers better grasp the public's attitudes and emotions in food safety incidents, so as to monitor the dynamics of the food safety field in real time and keep abreast of industry development and event changes. In this way, decision makers can obtain key information in a timely manner so as to take corresponding measures to prevent potential hazards and protect public safety and interests.

[0025] Specifically, by obtaining news and public opinion data related to food safety, the system can keep abreast of the latest developments and public opinion trends in the field of food safety, so as to better respond to possible food safety issues. Among them, news data is an important source of information that can provide reports and analysis of various food safety incidents. By analyzing news data, the system can understand the frequency of food safety incidents, the scope of impact, and the reactions of relevant parties, so as to help predict possible food safety issues and take timely countermeasures. Public opinion data reflects the public's attention and attitude towards food safety issues. By analyzing public opinion data, the system can understand the public's emotional tendencies, focus, and public opinion trends towards different food safety incidents, so as to better grasp the social public opinion dynamics and timely discover and respond to hot issues of public concern. By obtaining news and public opinion data related to food safety, the system can fully understand the dynamics and public opinion trends in the field of food safety, improve the perception and analysis capabilities of food safety issues, and provide important support and reference for protecting public health and food safety.

[0026] In the above-mentioned computer network information intelligent extraction system 100, the food safety comment data extraction module 120 is used to extract food safety information text understanding feature vectors and food safety information association feature vectors from the food safety-related news and public opinion data. It should be understood that the process of extracting food safety information text understanding feature vectors and food safety information association feature vectors can help the system deeply understand the content and semantics of food safety data, while capturing the association relationship between food safety information, and providing a more accurate and comprehensive basis for subsequent user emotion judgment.

[0027] Figure 2 FIG. 1 is a block diagram of a food safety review data extraction module in an information intelligent extraction system of a computer network according to an embodiment of the present application. Figure 2 As shown, in a specific embodiment of the present application, the food safety comment data extraction module 120 includes: a food safety comment semantic encoding unit 121, used to perform semantic encoding on the food safety-related news and public opinion data to obtain the food safety information text understanding feature vector; a food safety comment text convolution unit 122, used to perform text convolution encoding on the food safety-related news and public opinion data to obtain the food safety information association feature vector.

[0028] It should be understandable that semantic coding can help the system better understand the meaning of text information. Food safety-related news and public opinion data often contain a lot of information and details, and this information often needs to be extracted through semantic coding to extract key features and content. Through semantic coding, the system can convert text information into numerical representations, capture the semantics and context, and thus achieve a deep understanding of text information. Among them, semantic coding can help the system mine hidden information and correlations in text information. The field of food safety involves many aspects of information and knowledge, and sometimes the text contains important correlations and hidden information. Through semantic coding, the system can convert text information into feature vectors, thereby realizing the analysis of correlations and similarities between text information, and helping the system to mine potential laws and patterns in text data.

[0029] Furthermore, the process of text convolutional coding of food safety-related news and public opinion data is to capture the correlation and connection in the text data, so as to construct the correlation feature vector between food safety information. This step is crucial for the system's emotional judgment and user feedback analysis. Among them, text convolutional coding can help the system identify the correlation in text data. In the field of food safety, there may be complex associations and connections between different information. For example, a food safety incident may trigger users' attention or feedback on other related issues. Through text convolutional coding, the system can capture the correlation features between these information, helping the system to more comprehensively understand the information interaction in user feedback and public opinion data. Text convolutional coding helps the system understand the contextual information of text data. In the field of food safety, understanding the contextual background of food safety information is crucial for accurately judging user emotions and feedback. Through text convolutional coding, the system can analyze the contextual relationship in text data, integrate the correlation between different information into the feature vector, and provide a more comprehensive and accurate information basis for the system.

[0030] In a specific embodiment of the present application, the food safety review semantic encoding unit 121 includes: segmenting the food safety-related news and public opinion data to obtain a sequence of food safety information vectors; and passing the sequence of food safety information vectors through a bidirectional LSTM model to obtain the food safety information text understanding feature vector.

[0031] It should be understood that word segmentation can divide continuous text information into meaningful word units. In the field of food safety, news and public opinion data often contain a large amount of text information. Word segmentation can be used to segment this information according to word units, and complex text information can be converted into a clearer and more structured form, providing a basis for subsequent information processing and analysis. Here, word segmentation can help the system better understand the meaning of text information. Each word carries certain semantic information. Word segmentation can convert text information into word sequences. The system can more accurately capture key words and information in the text information, thereby achieving a deep understanding and analysis of the text information.

[0032] Furthermore, by passing the sequence of food safety information vectors through the bidirectional LSTM model, the semantic information and contextual relationship of the text data can be deeply understood, so as to more accurately extract and represent the food safety information. The bidirectional LSTM model can effectively capture the long-term dependencies and contextual information in the text data. In the field of food safety, understanding the semantic information and contextual relationship of text data is crucial for accurately judging user emotions and feedback. Through the bidirectional LSTM model, the system can simultaneously consider the previous and next information in the text data, so as to more comprehensively understand the semantics and relevance in the text data and extract more representative feature vectors. In addition, the bidirectional LSTM model can help the system to have a deep semantic understanding of the text data. In the field of food safety, text data may contain rich semantic information and implicit meanings, and the system needs to be able to deeply understand the content. Through the bidirectional LSTM model, the system can process the text data at the word-by-word level, so as to better capture the semantic information in the text data and generate more representative text understanding feature vectors. Specifically, the converter-based Bert model of the bidirectional LSTM model is used to perform global context semantic encoding on the sequence of the food safety information vectors to obtain multiple feature vectors; and the multiple feature vectors are cascaded to obtain the food safety information text understanding feature vector.

[0033] In a specific embodiment of the present application, the food safety comment text convolution unit 122 includes: a text semantic feature encoding subunit 1221, which is used to encode the text semantic features of the food safety-related news and public opinion data to obtain a food safety information associated feature map; an associated feature pooling subunit 1222, which is used to pool the food safety information associated feature map to obtain the food safety information associated feature vector.

[0034] It should be understood that text semantic feature encoding can convert text data into high-dimensional semantic feature vectors. In the field of food safety, news and public opinion data often contain rich semantic information and implicit meanings. Through text semantic feature encoding, the system can convert this information into feature vectors with semantic meanings to better represent the information and relevance in text data. Text semantic feature encoding can achieve a deep understanding and representation of text data. In the process of food safety information processing, deep learning models can generate more representative and rich feature vectors by learning the semantic information and association rules in text data. These feature vectors can help the system better understand the relevance between text data and reveal the hidden features and rules in text data.

[0035] Furthermore, the process of pooling the food safety information association feature graph is to simplify and refine the complex graph structure information, so as to obtain a more representative and compact feature vector. Among them, the pooling operation can help the system extract the most important feature information from the complex association feature graph. In food safety information processing, the association feature graph often contains a large number of nodes and edges, representing the complex relationship between text data. Through the pooling operation, the system can simplify and refine these complex graph structure information, retain the most critical feature information, and thus generate a more representative and compact feature vector.

[0036] In a specific embodiment of the present application, the text semantic feature encoding subunit 1221 includes: segmenting and arranging the food safety-related news and public opinion data to obtain a food safety word feature matrix; passing the food safety word feature matrix through a food safety text convolutional neural network as a feature extractor to obtain the food safety information association feature map.

[0037] It should be understood that word segmentation and arrangement can convert text data into discrete word representations. In the field of food safety, news and public opinion data often contain a large amount of word information. By performing word segmentation on text data, the system can divide continuous text information into discrete word units and convert text information into a structured form that can be processed by computers. Through word segmentation, the system can identify key words and phrases in text data, helping the system to better understand the subject and content of text data. By arranging and combining these key words according to certain rules, a word feature matrix can be formed to convert the information in the text data into a feature representation.

[0038] Furthermore, the food safety text convolutional neural network can effectively capture the local features and semantic information between words. In food safety information processing, the order and combination of words are crucial to understanding the meaning of text data. Through the operation of the convolutional neural network, the system can perform a sliding window convolution operation on the word feature matrix to effectively capture the local features and semantic information between words, thereby extracting richer and more specific feature representations. Specifically, each layer of the food safety text convolutional neural network used as a feature extractor performs convolution processing, mean pooling processing based on the local feature matrix, and nonlinear activation processing on the input data in the forward pass of the layer to output the food safety information association feature map by the last layer of the food safety text convolutional neural network used as a feature extractor, wherein the input of the food safety text convolutional neural network used as a feature extractor is the food safety word feature matrix.

[0039] In the above-mentioned computer network information intelligent extraction system 100, the user emotion judgment module 130 is used to judge whether the user emotion is positive or negative based on the food safety information text understanding feature vector and the food safety information association feature vector. It should be understood that through the food safety information text understanding feature vector, the system can extract emotion-related features from text data. In the field of food safety, user comments and public opinion data often contain rich emotional information. Through text understanding technology, the system can identify and extract these emotional features to help the system understand the user's emotional tendencies. By using the food safety information association feature vector, the system can analyze user emotions more comprehensively. Food safety information often involves many aspects of content and factors. By analyzing the association features between food safety information, the system can more accurately grasp the reasons and motivations behind user emotions, thereby more comprehensively evaluating the user's emotional state. By judging whether the user emotion is positive or negative based on the food safety information text understanding feature vector and the food safety information association feature vector, the intelligent recognition and analysis of user emotions is achieved through the combination of deep learning technology and natural language processing methods. This process is of great significance for understanding the user's attitude and emotional tendency towards food safety issues, and can provide strong support for related decisions and actions.

[0040] In a specific embodiment of the present application, the user emotion judgment module 130 includes: a multi-scale food safety information feature encoding unit 131, which is used to perform feature encoding on the food safety information text understanding feature vector and the food safety information association feature vector to obtain a multi-scale food safety information association feature matrix; a food safety information feature optimization unit 132, which is used to perform data distribution dependency optimization on the multi-scale food safety information association feature matrix based on the decoding parameter space to obtain an optimized multi-scale food safety information association feature matrix; a user emotion judgment generation unit 133, which is used to pass the optimized multi-scale food safety information association feature matrix through a classifier to obtain a classification result, and the classification result is used to indicate whether the user emotion is positive or negative.

[0041] It should be understood that by obtaining a multi-scale food safety information correlation feature matrix through feature encoding, the feature information of different scales can be comprehensively utilized to improve the system's ability to understand and characterize the correlation of food safety information. Features of different scales often contain information of different levels and granularity. By encoding and fusing this feature information, the system can capture the correlation and complexity of food safety information at multiple scales, thereby achieving more comprehensive and accurate information processing and analysis.

[0042] Furthermore, through optimization based on the decoding parameter space, the distribution of the feature space can be effectively adjusted to make the feature representation more compact and effective. In food safety information processing, features of different scales often have different importance and contribution. By optimizing the data distribution in the parameter space, the relationship between features of each scale can be better balanced, and the consistency and stability of feature representation can be improved. Among them, the optimization of data distribution dependency based on the decoding parameter space can help the model better understand and utilize the intrinsic correlation between data. In food safety information processing, there may be complex dependencies and interactions between related features of different scales. By optimizing the feature matrix, these correlation information can be better mined and utilized, and the model's ability to understand and represent food safety information can be improved.

[0043] Furthermore, by inputting the optimized multi-scale food safety information correlation feature matrix into the classifier, the classification ability of the machine learning algorithm can be used to classify the sentiment of food safety information and divide user comments into positive or negative emotions. Such classification results can help the system better understand the user's attitude and emotional tendency towards food safety, and provide an important reference for further decision-making and analysis. The positive or negative labels of the classification results can intuitively reflect the user's views and feelings on food safety, which helps companies and related institutions understand consumer needs and preferences. By classifying user emotions, the system can quickly and accurately understand users' evaluation of food safety, help companies adjust product strategies and services in a timely manner, and improve user satisfaction and brand image.

[0044] In a specific embodiment of the present application, the multi-scale food safety information feature encoding unit 131 includes: fusing the food safety information text understanding feature vector and the food safety information association feature vector to obtain a multi-scale food safety information text understanding feature matrix; passing the multi-scale food safety information text understanding feature matrix through a convolutional neural network model of multiple mixed convolutional layers to obtain the multi-scale food safety information association feature matrix.

[0045] It should be understandable that by fusing the feature vectors at these two different levels, the system can obtain a multi-scale food safety information text understanding feature matrix. This matrix combines the deep understanding features of a single text and the correlation features between different texts, providing a more comprehensive and multi-dimensional information representation. Based on this multi-scale feature matrix, the system can more accurately analyze and judge user emotions and attitudes, thereby providing more effective support and guidance for public opinion management and food safety monitoring. This fusion method can help the system better grasp the complexity and diversity of food safety information, and provide decision makers with more comprehensive information support to meet the challenges and changes in the field of food safety. Specifically, a joint encoder is used to jointly encode the food safety information text understanding feature vector and the food safety information correlation feature vector using the following fusion formula to generate the multi-scale food safety information text understanding feature matrix;

[0046] Among them, the fusion formula is:

[0047]

[0048] in represents vector multiplication, M represents the multi-scale food safety information text understanding feature matrix, V1 represents the food safety information text understanding feature vector, V2 represents the food safety information text understanding feature vector, Represents the transpose of the food safety information associated feature vector.

[0049] Furthermore, the multi-scale food safety information text understanding feature matrix is ​​processed through a convolutional neural network model with multiple mixed convolutional layers, aiming to better capture the information association features at different scales, thereby improving the understanding and analysis capabilities of food safety information. This process involves extracting and fusing text features layer by layer through multiple mixed convolutional layers to generate a multi-scale food safety information association feature matrix with more representation and generalization capabilities. Among them, the multi-scale food safety information text understanding feature matrix contains information features of different levels and scales, which may include local features, global features, and semantic features at different levels of abstraction. Through the convolutional neural network model with multiple mixed convolutional layers, these multi-scale features can be effectively extracted and fused. In the convolutional neural network model, each mixed convolutional layer can capture feature information at different scales, from local to global, from low level to high level. Through the stacking and cross-connection of multiple mixed convolutional layers, these multi-scale features can be gradually extracted and integrated, making the final food safety information association feature matrix more rich and representational. Through this multi-level, multi-scale feature extraction and fusion process, the convolutional neural network model can better understand and characterize the correlation between food safety information, thereby improving the accuracy and effect of analyzing user emotions and attitudes. This method can not only help decision makers better grasp the public's attitude and emotions towards food safety incidents, but also provide more effective tools and support for real-time monitoring of the dynamics in the field of food safety. Specifically, each layer of the convolutional neural network model using the multiple mixed convolutional layers performs the following on the input data in the forward pass of the layer: the convolution units of each layer of the convolutional neural network model using the multiple mixed convolutional layers perform convolution processing on the input data based on a two-dimensional convolution kernel to obtain a convolution feature map; the pooling units of each layer of the convolutional neural network model using the multiple mixed convolutional layers perform pooling processing on the convolution feature map along the channel dimension to obtain a pooling feature map; and the activation units of each layer of the convolutional neural network model using the multiple mixed convolutional layers perform nonlinear activation on the feature values ​​of each position in the pooling feature map to obtain an activation feature map; wherein the output of the last layer of the convolutional neural network model of the multiple mixed convolutional layers is the multi-scale food safety information association feature matrix.

[0050] In a specific embodiment of the present application, the food safety information feature optimization unit 132 includes: multiplying the multi-scale food safety information associated feature matrix with the classification weight matrix of the classifier to obtain an intermediate feature vector; splicing the multi-scale food safety information associated feature matrix with the intermediate feature vector to obtain a spliced ​​feature vector; multiplying the first weight matrix by the spliced ​​feature vector and then adding the first bias vector to obtain a coded spliced ​​feature vector; activating the coded spliced ​​feature vector through a sigmoid function to obtain an activation value; subtracting the activation value from the intermediate feature vector by position to obtain a weighted intermediate feature vector; multiplying the activation value by position with the multi-scale food safety information associated feature matrix to obtain a weighted multi-scale food safety information associated feature matrix; subtracting the intermediate feature vector from the multi-scale food safety information associated feature matrix by position to obtain a difference feature vector; adding the weighted intermediate feature vector to the weighted multi-scale food safety information associated feature matrix and then dividing it by the difference feature vector to obtain a second intermediate feature vector; passing the second intermediate feature vector through a ReLU function to obtain an optimized multi-scale food safety information associated feature matrix.

[0051] In particular, in the technical solution of the present application, it is considered that the extraction of food safety information associated feature vectors involves features of different scales and types, such as text understanding features and associated features. The poor monotonicity of the overall feature distribution manifold may cause problems in feature representation, making it difficult for the classifier to effectively learn the true distribution of the data, thereby affecting the accuracy of the classification results. When generating multi-scale food safety information associated feature vectors, it is necessary to fuse feature vectors from different scales, which may increase the complexity of feature fusion. The poor monotonicity of the overall feature distribution manifold may make feature fusion more difficult, thereby affecting the classifier's ability to understand and classify data. Due to the poor monotonicity of the overall feature distribution manifold of the multi-scale food safety information associated feature vector, the classifier may not be able to fully utilize the feature information of the data when classifying, resulting in insufficient classification regression constraints of the classifier. This will affect the accuracy of the classification results, making it difficult for the classifier to accurately classify food safety-related news and public opinion data as positive or negative. In order to improve the accuracy of the classification results, in the technical solution of the present application, the multi-scale food safety information associated feature matrix is ​​optimized based on the data distribution dependency of the decoding parameter space to obtain an optimized multi-scale food safety information associated feature matrix.

[0052] The method of optimizing the data distribution dependence of the multi-scale food safety information association feature matrix based on the decoding parameter space to obtain the optimized multi-scale food safety information association feature matrix includes: optimizing the data distribution dependence of the multi-scale food safety information association feature matrix based on the decoding parameter space using the following optimization formula, wherein the optimization formula is:

[0053]

[0054] Among them, v c represents the multi-scale food safety information association feature matrix, M is the classification weight matrix of the classifier, concat represents the concatenation function, W1 represents the first weight matrix, b1 represents the first bias vector, represents matrix multiplication, sigmoid represents an S-shaped activation function, τ represents an activation value, ReLU represents a linear rectification function, and V′ represents the optimized multi-scale food safety information association feature matrix.

[0055] Specifically, in the technical solution of the present application, the manifold monotonicity of the overall feature distribution of the multi-scale food safety information association feature matrix is ​​poor, resulting in insufficient classification regression constraints relative to the overall feature distribution of the classifier when classification regression is performed through the classifier, thereby affecting the accuracy of the classification results.

[0056] Therefore, in the technical solution of the present application, the multi-scale food safety information association feature matrix is ​​optimized based on the data distribution dependence of the decoding parameter space, which is performed by auxiliary analysis and description of the geometric characteristics of the feature manifold of the multi-scale food safety information association feature matrix using the classification weight matrix of the classifier, and the difference between the feature manifold of the auxiliary description analysis result and the original multi-scale food safety information association feature matrix is ​​used as a differential factor to support the descriptiveness of the improved multi-scale food safety information association feature matrix for the predetermined classification result of the classifier, and finally an activation mechanism is added for activation to maintain the reinforcement of the distribution dependence with a positive description. In this way, the manifold monotonicity of the overall feature layout of the multi-scale food safety information association feature matrix is ​​significantly improved, thereby enhancing its classification regression constraint relative to the overall feature layout of the classifier and improving the accuracy of the classification result.

[0057] In summary, the embodiment of the present application first obtains news and public opinion data related to food safety, then uses deep learning technology to perform feature extraction and association analysis, and finally uses a classifier to perform positive and negative sentiment analysis on the news to determine whether the user's emotions are positive or negative, so as to help decision makers better understand the public's attitude and emotions towards food safety incidents, thereby monitoring the dynamics of the food safety field in real time, promptly understanding industry development and event changes, and then taking corresponding measures to avoid harm.

[0058] As described above, the computer network information intelligent extraction system 100 according to the embodiment of the present application can be implemented in various terminal devices. In one example, the computer network information intelligent extraction system 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the computer network information intelligent extraction system 100 can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the computer network information intelligent extraction system 100 can also be one of the many hardware modules of the terminal device.

[0059] Alternatively, in another example, the computer network's intelligent information extraction system 100 and the terminal device may also be separate devices, and the computer network's intelligent information extraction system 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0060] Figure 5 Flowchart of the method for intelligently extracting information from a computer network according to an embodiment of the present application. Figure 5 As shown, according to the intelligent information extraction method of a computer network of an embodiment of the present application, it includes: S110, obtaining news and public opinion data related to food safety; S120, extracting food safety information text understanding feature vectors and food safety information association feature vectors from the food safety related news and public opinion data; S130, judging whether the user emotion is positive or negative based on the food safety information text understanding feature vectors and the food safety information association feature vectors.

[0061] Here, those skilled in the art will appreciate that the specific operations of each step in the above-mentioned computer network information intelligent extraction method have been described in the above reference. Figures 1 to 4 The information intelligent extraction system for computer networks has been introduced in detail, and therefore, its repeated description will be omitted.

[0062] Below, reference Figure 6 To describe an electronic device according to an embodiment of the present application.

[0063] Figure 6 A block diagram of an electronic device according to an embodiment of the present application is illustrated.

[0064] like Figure 6As shown, the electronic device 10 includes an input device 11, an input interface 12, a central processing unit 13, a memory 14, an output interface 15, an output device 16, and a bus 17. The input interface 12, the central processing unit 13, the memory 14, and the output interface 15 are connected to each other through the bus 17, and the input device 11 and the output device 16 are connected to the bus 17 through the input interface 14 and the output interface 15, respectively, and then connected to other components of the electronic device 10.

[0065] Specifically, the input device 11 receives input information from the outside and transmits the input information to the central processing unit 13 through the input interface 12; the central processing unit 13 processes the input information based on the computer executable instructions stored in the memory 14 to generate output information, temporarily or permanently stores the output information in the memory 14, and then transmits the output information to the output device 16 through the output interface 15; the output device 16 outputs the output information to the outside of the electronic device 10 for user use.

[0066] In one embodiment, Figure 6 The electronic device 10 shown can be implemented as a network device, which may include: a memory configured to store programs; a processor configured to run the programs stored in the memory to execute any one of the computer network information intelligent extraction methods described in the above embodiments.

[0067] According to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program tangibly contained on a machine-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network, and / or installed from a removable storage medium.

[0068] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In hardware implementations, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0069] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present application, but the present application is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and substance of the present application, and these modifications and improvements are also considered to be within the scope of protection of the present application.

Claims

1. An intelligent information extraction system for a computer network, characterized in that: include: Food safety review data acquisition module, used to obtain news and public opinion data related to food safety; A food safety comment data extraction module is used to extract food safety information text understanding feature vectors and food safety information association feature vectors from the food safety related news and public opinion data; The user emotion judgment module is used to judge whether the user emotion is positive or negative based on the food safety information text understanding feature vector and the food safety information association feature vector.

2. The computer network information intelligent extraction system according to claim 1, characterized in that: The food safety review data extraction module includes: A food safety comment semantic encoding unit, used for semantically encoding the food safety related news and public opinion data to obtain the food safety information text understanding feature vector; The food safety comment text convolution unit is used to perform text convolution encoding on the food safety-related news and public opinion data to obtain the food safety information-related feature vector.

3. The computer network information intelligent extraction system according to claim 2, characterized in that: The food safety review semantic encoding unit includes: Segmenting the food safety related news and public opinion data to obtain a sequence of food safety information vectors; The sequence of the food safety information vectors is passed through a bidirectional LSTM model to obtain the food safety information text understanding feature vector.

4. The computer network information intelligent extraction system according to claim 3, characterized in that: The food safety review text convolution unit includes: A text semantic feature encoding subunit, used for performing text semantic feature encoding on the food safety related news and public opinion data to obtain a food safety information correlation feature graph; The associated feature pooling subunit is used to pool the food safety information associated feature map to obtain the food safety information associated feature vector.

5. The computer network information intelligent extraction system according to claim 4, characterized in that: The text semantic feature encoding subunit includes: Segmenting and arranging the food safety related news and public opinion data to obtain a food safety word feature matrix; The food safety word feature matrix is ​​passed through a food safety text convolutional neural network as a feature extractor to obtain the food safety information associated feature graph.

6. The computer network information intelligent extraction system according to claim 5, characterized in that: The user emotion judgment module includes: A multi-scale food safety information feature encoding unit, used for performing feature encoding on the food safety information text understanding feature vector and the food safety information association feature vector to obtain a multi-scale food safety information association feature matrix; A food safety information feature optimization unit, used to optimize the multi-scale food safety information association feature matrix based on the data distribution dependency of the decoding parameter space to obtain an optimized multi-scale food safety information association feature matrix; The user emotion judgment generating unit is used to pass the optimized multi-scale food safety information association feature matrix through a classifier to obtain a classification result, and the classification result is used to indicate whether the user emotion is positive or negative.

7. The computer network information intelligent extraction system according to claim 6, characterized in that: The multi-scale food safety information feature encoding unit comprises: Fusion of the food safety information text understanding feature vector and the food safety information rationale feature vector to obtain a multi-scale food safety information text understanding feature matrix; The multi-scale food safety information text understanding feature matrix is ​​passed through a convolutional neural network model with multiple mixed convolutional layers to obtain the multi-scale food safety information association feature matrix.

8. The computer network information intelligent extraction system according to claim 7, characterized in that: The food safety information feature optimization unit comprises: Multiplying the multi-scale food safety information association feature matrix with the classification weight matrix of the classifier to obtain an intermediate feature vector; Concatenating the multi-scale food safety information association feature matrix with the intermediate feature vector to obtain a concatenated feature vector; The first weight matrix is ​​multiplied by the concatenated feature vector and then the first bias vector is added to obtain a coded concatenated feature vector; Activating the encoded concatenated feature vector through a sigmoid function to obtain an activation value; Subtract the activation value from one and multiply it by position with the intermediate feature vector to obtain the weighted intermediate feature vector; Multiplying the activation value by the multi-scale food safety information association feature matrix by position to obtain a weighted multi-scale food safety information association feature matrix; Subtracting the intermediate feature vector from the multi-scale food safety information associated feature matrix by position to obtain a difference feature vector; Adding the weighted intermediate feature vector and the weighted multi-scale food safety information association feature matrix and then dividing by the difference feature vector to obtain a second intermediate feature vector; The second intermediate eigenvector is passed through a ReLU function to obtain an optimized multi-scale food safety information association feature matrix.

9. A method for intelligently extracting information from a computer network, characterized in that: include: Obtain news and public opinion data related to food safety; Extracting food safety information text understanding feature vectors and food safety information association feature vectors from the food safety related news and public opinion data; Based on the food safety information text understanding feature vector and the food safety information association feature vector, it is determined whether the user emotion is positive or negative.

10. The method for intelligently extracting information from a computer network according to claim 9, characterized in that: Extracting food safety information text understanding feature vectors and food safety information association feature vectors from the food safety related news and public opinion data, including: Semantically encoding the food safety related news and public opinion data to obtain a food safety information text understanding feature vector; Text convolution encoding is performed on the food safety-related news and public opinion data to obtain the food safety information-related feature vector.

Citation Information

Patent Citations

  • Food safety public opinion event extraction method based on deep learning

    CN112527956A

  • Food safety public opinion analysis technology extraction method

    CN114186047A

  • Food safety public opinion analysis technology extraction method and system

    CN114996565A

  • Public opinion text sentiment analysis method based on semantic dependency relationship fusion features

    CN115098634A

  • Company public opinion monitoring system and method based on big data

    CN115409018A

Cited By

  • Report data processing system and method of unmanned aerial vehicle management platform based on AI

    CN119763213A