Text data mining and analysis system and method based on artificial intelligence

Abstract

The present application relates to the field of intelligent mining of text data, and specifically discloses a text data mining and analysis system and method based on artificial intelligence, which first collects relevant introduction data of scenic spots and multiple online reviews of scenic spots in the platform as input data, and then uses data mining and deep learning technology to extract features and analyze these input data to obtain classification results for indicating whether there are malicious comments in the platform. In this way, according to the classification results, it helps scenic spots to achieve more effective data processing and information utilization, thereby promoting the development of the smart tourism industry.

Description

Technical Field

[0001] The present application relates to the field of intelligent mining of text data, and more specifically, to an artificial intelligence-based text data mining and analysis system and method. Background Art

[0002] With the rapid development of science and technology, big data has been deeply integrated into the smart tourism industry. Using big data to provide rich information resources for the smart tourism industry is the key. Only by effectively applying these resources can the development of smart tourism be truly realized.

[0003] In a highly competitive market environment, it is crucial to improve tourist satisfaction. Only by ensuring the improvement of tourist satisfaction can we stabilize the existing passenger flow and enhance the competitiveness of tourist attractions. However, it is inevitable that some tourists will post malicious comments, which may have a negative impact on the reputation and business of the scenic spot. Therefore, it is necessary to deeply analyze the online review data of tourists in order to protect the reputation of the scenic spot, improve service quality, ensure the rights and interests of tourists, increase tourist satisfaction, and maintain the order of social platforms.

[0004] With the popularization of the Internet, society has entered the era of big data driven by data. With the rapid growth of data volume, the number of comments is huge and the content is complicated. Traditional data processing methods rely on manual experience, with low classification efficiency and high error rate. This is a realistic challenge that needs to be solved urgently.

[0005] Therefore, we look forward to an artificial intelligence-based text data mining and analysis system and method, which can extract and analyze online review information of scenic spots based on data mining and deep learning technology, so as to achieve more effective data processing and information utilization, thereby promoting the development of the smart tourism industry. Summary of the invention

[0006] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides a text data mining and analysis system and method based on artificial intelligence, which first collects relevant introduction data of the scenic spot and multiple online reviews of the scenic spot in the platform as input data, and then uses data mining and deep learning technology to extract features and analyze these input data to obtain classification results for indicating whether there are malicious comments in the platform. In this way, according to the classification results, it helps the scenic spot to achieve more effective data processing and information utilization, thereby promoting the development of the smart tourism industry.

[0007] According to a first aspect of the present application, there is provided a text data mining and analysis system based on artificial intelligence, comprising:

[0008] The scenic spot related data collection module is used to collect scenic spot related introduction data and multiple scenic spot online review data on the platform;

[0009] A scenic spot related data feature acquisition module is used to extract and analyze the features of the scenic spot related introduction data and the multiple scenic spot online review data in the platform to obtain the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix;

[0010] A scenic spot related data feature fusion module, used for multiplying the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix to obtain a scenic spot evaluation feature vector;

[0011] The scenic spot related data feature classification result generation module is used to obtain a classification result based on the scenic spot evaluation feature vector, and the classification result is used to indicate whether there are malicious comments in the platform.

[0012] In combination with the first aspect of the present application, in an artificial intelligence-based text data mining and analysis system of the first aspect of the present application, the scenic spot related data feature acquisition module includes: a scenic spot introduction feature extraction unit, used to perform feature encoding operations on the scenic spot related introduction data to obtain the scenic spot comprehensive introduction feature vector; a scenic spot online review feature extraction unit, used to perform feature encoding operations on multiple scenic spot online review data in the platform to obtain the scenic spot online review global feature matrix.

[0013] According to a second aspect of the present application, there is provided a text data mining and analysis method based on artificial intelligence, which comprises:

[0014] Collect relevant introduction data of scenic spots and multiple online reviews of scenic spots on the platform;

[0015] Extracting and analyzing the features of the scenic spot introduction data and the multiple scenic spot online review data in the platform respectively to obtain a scenic spot comprehensive introduction feature vector and a scenic spot online review global feature matrix;

[0016] Multiplying the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix to obtain a scenic spot evaluation feature vector;

[0017] A classification result is obtained based on the scenic spot evaluation feature vector, and the classification result is used to indicate whether there are malicious comments in the platform.

[0018] In combination with the second aspect of the present application, in an artificial intelligence-based text data mining and analysis method of the second aspect of the present application, a classification result is obtained based on the scenic spot evaluation feature vector, and the classification result is used to indicate whether there are malicious comments in the platform, including: a scenic spot evaluation feature extraction unit, used to pass the scenic spot evaluation feature vector through a scenic spot evaluation feature extraction module based on a convolutional neural network model to obtain a scenic spot evaluation classification feature vector; a scenic spot evaluation feature optimization unit, used to perform sparsity constraints on the scenic spot evaluation classification feature vector based on the model parameter space to obtain a sparse optimized scenic spot evaluation classification feature vector; a scenic spot evaluation feature classification unit, used to pass the sparse optimized scenic spot evaluation classification feature vector through a classifier to obtain the classification result.

[0019] Compared with the prior art, the text data mining and analysis system and method based on artificial intelligence provided by the present application first collects the relevant introduction data of the scenic spot and multiple online reviews of the scenic spot in the platform as input data, and then uses data mining and deep learning technology to extract features and analyze these input data to obtain classification results for indicating whether there are malicious comments in the platform. In this way, according to the classification results, it helps the scenic spot to achieve more effective data processing and information utilization, thereby promoting the development of the smart tourism industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] 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.

[0021] Figure 1 The figure illustrates a schematic block diagram of an artificial intelligence-based text data mining and analysis system according to an embodiment of the present application.

[0022] Figure 2 The figure illustrates a schematic block diagram of a scenic spot related data feature acquisition module in an artificial intelligence based text data mining and analysis system according to an embodiment of the present application.

[0023] Figure 3 The diagram illustrates a schematic block diagram of a scenic spot introduction feature extraction unit in a scenic spot related data feature acquisition module in an artificial intelligence based text data mining and analysis system according to an embodiment of the present application.

[0024] Figure 4 The figure illustrates a schematic block diagram of a scenic spot online review feature extraction unit in a scenic spot related data feature acquisition module in an artificial intelligence-based text data mining and analysis system according to an embodiment of the present application.

[0025] Figure 5 The figure illustrates a schematic block diagram of a scenic spot related data feature classification result generation module in an artificial intelligence based text data mining and analysis system according to an embodiment of the present application.

[0026] Figure 6 The figure illustrates a flow chart of a text data mining and analysis method based on artificial intelligence according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] 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.

[0028] Exemplary Systems

[0029] Figure 1 FIG. 1 is a schematic block diagram of a text data mining and analysis system based on artificial intelligence according to an embodiment of the present application. Figure 1 As shown, according to the embodiment of the present application, the text data mining and analysis system 100 based on artificial intelligence includes: a scenic spot related data collection module 110, which is used to collect scenic spot related introduction data and multiple scenic spot online review data in the platform; a scenic spot related data feature acquisition module 120, which is used to extract and analyze the features of the scenic spot related introduction data and the multiple scenic spot online review data in the platform respectively to obtain a scenic spot comprehensive introduction feature vector and a scenic spot online review global feature matrix; a scenic spot related data feature fusion module 130, which is used to multiply the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix to obtain a scenic spot evaluation feature vector; a scenic spot related data feature classification result generation module 140, which is used to obtain a classification result based on the scenic spot evaluation feature vector, and the classification result is used to indicate whether there are malicious comments in the platform.

[0030] With the rapid advancement of science and technology, big data has been deeply integrated into the smart tourism industry. The key is to use big data to provide rich information resources for the smart tourism industry. Only by effectively utilizing these resources can the development of smart tourism be truly realized.

[0031] In the fierce market competition, it is crucial to improve tourist satisfaction. Only by ensuring the improvement of tourist satisfaction can we consolidate the existing passenger flow and enhance the competitiveness of tourist attractions. However, it is inevitable that some tourists will post malicious comments, which may have a negative impact on the reputation and business of the scenic spot. Therefore, it is necessary to deeply analyze the online evaluation data of tourists in order to protect the reputation of the scenic spot, improve service quality, ensure the rights and interests of tourists, improve tourist satisfaction, and maintain the order of social platforms.

[0032] With the popularization of the Internet, society has entered the era of big data driven by data. With the rapid growth of data volume, the number of reviews is huge and the content is complicated. Traditional data processing methods rely on manual experience, with low classification efficiency and high error rate. This is a realistic challenge that needs to be solved urgently. Therefore, a text data mining and analysis system and method based on artificial intelligence is expected to extract and analyze the online review information of scenic spots based on data mining and deep learning technology to achieve more effective data processing and information utilization, thereby promoting the development of the smart tourism industry.

[0033] In particular, data mining technology is a process of analyzing large amounts of data to discover hidden patterns, associations, and trends. It combines machine learning, statistics, and database technology to extract useful information from massive amounts of data. Deep learning is a machine learning technology based on artificial neural networks. Its core idea is to learn abstract feature representations of data through multi-level neural networks. Deep learning models can automatically learn representations from data without manually specifying features.

[0034] In the text data mining and analysis system based on artificial intelligence in the embodiment of the present application, data mining and deep learning technology play an important role. For example, by combining data mining and deep learning technology, the smart tourism industry can better utilize big data resources, improve service quality, protect reputation, enhance competitiveness, and achieve a smarter and more efficient tourism experience. Therefore, in the specific embodiment of the present application, feature extraction and analysis of input data are performed through data mining and deep learning technology.

[0035] In the embodiment of the present application, the scenic spot related data collection module 110 is used to collect scenic spot related introduction data and multiple scenic spot online review data in the platform. It should be understood that collecting multiple scenic spot online review data can improve the system's ability to understand and analyze the content of the review, and help to more accurately identify whether there are malicious comments. That is, the comprehensive analysis of multiple data can reduce the misjudgment rate and improve the system's detection accuracy of malicious comments, thereby more effectively protecting the reputation of the scenic spot and improving the service quality. Therefore, in order to more accurately detect malicious comments, first collect multiple scenic spot online review data in the platform. In addition, the scenic spot related introduction data provides detailed information on the basic information of the scenic spot, scenic spot introduction and service facilities, etc., to help the system fully understand the characteristics and features of the scenic spot. Combining the scenic spot related introduction data can help the system better understand the semantics and context of the online review data, thereby improving the recognition accuracy of malicious comments. Therefore, in order to better identify potential malicious comments, it is also necessary to collect scenic spot related introduction data (such as scenic spot basic information, scenic spot introduction, service facilities, etc.).

[0036] Specifically, when collecting scenic spot introduction data, crawler technology can be used to capture scenic spot introduction data from scenic spot official websites, travel platforms, social media and other channels. When collecting multiple scenic spot online review data on the platform, the API of social media platforms can be used to obtain scenic spot review data posted by users, such as Twitter, Facebook, Instagram, etc.

[0037] In the embodiment of the present application, the scenic spot related data feature acquisition module 120 is used to extract and analyze the features of the scenic spot related introduction data and the multiple scenic spot online review data in the platform to obtain the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix. It should be understood that after collecting these input data, the input data is further subjected to feature extraction and analysis tasks.

[0038] Specifically, Figure 2 The figure shows a schematic block diagram of a scenic spot related data feature acquisition module in an artificial intelligence based text data mining and analysis system according to an embodiment of the present application. Figure 2 As shown, the scenic spot related data feature acquisition module 120 includes: a scenic spot introduction feature extraction unit 121, which is used to perform feature coding operations on the scenic spot related introduction data to obtain the scenic spot comprehensive introduction feature vector; a scenic spot online review feature extraction unit 122, which is used to perform feature coding operations on multiple scenic spot online review data in the platform to obtain the scenic spot online review global feature matrix.

[0039] First, feature extraction and analysis are performed on the scenic spot related introduction data. Specifically, Figure 3 The figure shows a schematic block diagram of a scenic spot introduction feature extraction unit in a scenic spot related data feature acquisition module in a text data mining and analysis system based on artificial intelligence according to an embodiment of the present application. Figure 3 As shown, the scenic spot introduction feature extraction unit 121 includes: a scenic spot introduction text embedding coding subunit 121-1, which is used to arrange the scenic spot related introduction data into multiple scenic spot related introduction text sequences according to the type dimension and then embed them respectively to obtain multiple scenic spot introduction embedding vectors; a scenic spot introduction context feature acquisition subunit 121-2, which is used to pass the multiple scenic spot introduction embedding vectors through a scenic spot comprehensive introduction feature extraction module based on a context encoder to obtain the scenic spot comprehensive introduction feature vector.

[0040] It should be understood that the scenic spot introduction data includes basic information of the scenic spot, introduction of scenic spots, service facilities, etc. Directly extracting features from all the data may not be able to fully capture the semantic information and rich contextual relationships in the text data, thereby affecting the model's accurate understanding of the text data. Therefore, the scenic spot introduction data is first arranged into multiple scenic spot introduction text sequences according to the type dimension. Furthermore, considering that the order and contextual information of the text sequence are crucial to the understanding of the model, directly extracting features from these text sequences may result in the loss of some information. Embedded coding can help the model capture the semantic similarity between words, thereby better understanding the text content and improving the efficiency and accuracy of text data processing and analysis. Therefore, in order to better understand the text context information, the multiple scenic spot introduction text sequences are further embedded coded to obtain the multiple scenic spot introduction embedding vectors.

[0041] Next, it should be understood that the basic information of the scenic spot, the introduction of scenic spots, service facilities and other information constitute a complete introduction to the scenic spot. Integrating this information into a complete introduction to the scenic spot can provide a more comprehensive description of the scenic spot. By integrating this information, the system can have a more comprehensive understanding of the characteristics and highlights of each scenic spot. Considering that the context encoder containing the embedding layer can capture the contextual associations and semantic information in the text data. This helps to better understand the semantics and associations between the contexts in the scenic spot introduction text and improve the accuracy of feature extraction. Therefore, a scenic spot comprehensive introduction feature extraction module based on the context encoder is used to contextually encode the multiple scenic spot introduction embedding vectors to obtain the scenic spot comprehensive introduction feature vector. In this way, the feature vector extracted by the context encoder can more comprehensively reflect the characteristics of the scenic spot, including scenic spots, service facilities and other aspects, which helps to improve the understanding and utilization of scenic spot information and promote the development of the smart tourism industry.

[0042] In a specific embodiment of the present application, the scenic spot introduction context feature acquisition subunit 121-2 is used to: input the multiple scenic spot introduction embedding vectors into the converter-based context encoder to obtain multiple context scenic spot introduction feature vectors; and cascade the multiple context scenic spot introduction feature vectors to obtain the scenic spot comprehensive introduction feature vector.

[0043] Then, the features of multiple scenic spot online review data in the platform are extracted and analyzed. Specifically, Figure 4 The figure shows a schematic block diagram of a scenic spot online review feature extraction unit in a scenic spot related data feature acquisition module in a text data mining and analysis system based on artificial intelligence according to an embodiment of the present application. Figure 4As shown, the scenic spot online review feature extraction unit 122 includes: a scenic spot online review encoding subunit 122-1, which is used to embed and encode the multiple scenic spot online review data in the platform to obtain multiple scenic spot online review embedding vectors; a scenic spot online review feature extraction subunit 122-2, which is used to pass the multiple scenic spot online review embedding vectors through a scenic spot online review feature extraction module based on a feature encoder to obtain multiple scenic spot online review feature vectors; a scenic spot online review global feature extraction subunit 122-3, which is used to arrange the multiple scenic spot online review feature vectors into a scenic spot online review comprehensive input matrix and then pass them through a scenic spot online review global feature extraction module based on a two-dimensional convolutional neural network model to obtain the scenic spot online review global feature matrix.

[0044] It should be understood that text data is discrete and variable in length, making it difficult for machines to recognize. Therefore, in order to enable machines to better recognize text data, the multiple scenic spot online review data in the platform are first embedded and encoded to obtain multiple scenic spot online review embedding vectors. Among them, embedded coding can map text data into a high-dimensional vector space, so that each word or text fragment can be represented in the vector space with rich semantics. This helps to capture the semantic relationship between words and improve the representation ability of text data.

[0045] Next, considering that embedded coding converts text data into vector form, it does not guarantee that all information is useful. Feature extraction can help the system identify and retain the most relevant information and improve the effectiveness and accuracy of data representation. Therefore, in order to capture more advanced abstract features and improve the accuracy of feature extraction, a scenic spot online review feature extraction module based on a feature encoder is used to perform feature encoding on the multiple scenic spot online review embedding vectors to obtain the multiple scenic spot online review feature vectors.

[0046] In a specific embodiment of the present application, the scenic spot online review feature extraction subunit 122-2 is used to: use each layer of the feature encoder to perform the following on the input data in the forward pass of the layer: perform convolution processing based on a one-dimensional convolution kernel on the input data to obtain a convolution feature map; perform pooling processing based on a feature matrix on the convolution feature map to obtain a pooled feature map; perform nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the input of the first layer of the feature encoder is the multiple scenic spot online review embedding vectors, and the output of the last layer of the feature encoder is the multiple scenic spot online review feature vectors.

[0047] Next, considering that there is a large amount of online review data about scenic spots in social media, only analyzing a single online review data may not be able to obtain more comprehensive and objective information. Therefore, in order to obtain more reliable online review feature information, these online review information are first integrated. Specifically, in an embodiment of the present application, the multiple scenic spot online review feature vectors are arranged into a scenic spot online review comprehensive input matrix. Then, a scenic spot online review global feature extraction module based on a two-dimensional convolutional neural network model is used to extract and analyze the scenic spot online review comprehensive input matrix to obtain the scenic spot online review global feature matrix. Among them, the two-dimensional convolutional neural network model can effectively capture local and global features in text data, improve the efficiency of data processing and information utilization, and thus promote the development of the smart tourism industry.

[0048] In a specific embodiment of the present application, the scenic spot online review global feature extraction subunit 122-3 is used to: use each layer of the two-dimensional convolutional neural network model to perform the following operations on the input data in the forward pass of the layer: use the convolution units of each layer of the two-dimensional convolutional neural network model to perform convolution processing based on a two-dimensional convolution kernel on the input data to obtain a convolution feature map; use the pooling units of each layer of the two-dimensional convolutional neural network model to perform pooling processing on the convolution feature map along the channel dimension to obtain a pooling feature map; and use the activation units of each layer of the two-dimensional convolutional neural network model to perform nonlinear activation on the eigenvalues ​​of each position in the pooling feature map to obtain an activation feature map; wherein the output of the last layer of the two-dimensional convolutional neural network model is the global feature matrix of the scenic spot online review.

[0049] In an embodiment of the present application, the scenic spot related data feature fusion module 130 is used to multiply the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix to obtain a scenic spot evaluation feature vector. It should be understood that the comprehensive introduction feature vector of the scenic spot usually contains the basic information, characteristics, facilities, etc. of the scenic spot, while the global feature matrix of the scenic spot online review reflects the tourists' overall evaluation and experience of the scenic spot. Multiplying the two can comprehensively utilize the static description information and dynamic evaluation information of the scenic spot, making the evaluation more comprehensive and accurate, thereby better capturing malicious comments from a large number of comments. Therefore, the comprehensive introduction feature vector of the scenic spot and the global feature matrix of the scenic spot online review are further feature fused to obtain the scenic spot evaluation feature vector.

[0050] In the embodiment of the present application, the scenic spot related data feature classification result generation module 140 is used to obtain a classification result based on the scenic spot evaluation feature vector, and the classification result is used to indicate whether there are malicious comments in the platform. Specifically, Figure 5 The figure shows a schematic block diagram of a scenic spot related data feature classification result generation module in an artificial intelligence based text data mining and analysis system according to an embodiment of the present application. Figure 5 As shown, the scenic spot related data feature classification result generation module 140 includes: a scenic spot evaluation feature extraction unit 141, which is used to pass the scenic spot evaluation feature vector through a scenic spot evaluation feature extraction module based on a convolutional neural network model to obtain a scenic spot evaluation classification feature vector; a scenic spot evaluation feature optimization unit 142, which is used to perform sparsity constraints on the scenic spot evaluation classification feature vector based on the model parameter space to obtain a sparse optimized scenic spot evaluation classification feature vector; a scenic spot evaluation feature classification unit 143, which is used to pass the sparse optimized scenic spot evaluation classification feature vector through a classifier to obtain the classification result.

[0051] It should be understood that for complex data relationships and patterns, simple feature fusion methods may not be able to effectively capture the complex associations between features, affecting the performance of the model. Therefore, in order to further capture the complex association features between scenic spot introductions and scenic spot reviews, a convolutional neural network model is used to extract features from the scenic spot evaluation feature vector to obtain the scenic spot evaluation classification feature vector. Among them, the convolutional neural network model has a strong nonlinear modeling capability, can better handle nonlinear relationships, and is suitable for modeling complex data relationships.

[0052] In particular, in the technical solution of the present application, it is considered that when collecting scenic spot introductions and online review data, erroneous information, incomplete records or inconsistent data formats may be included, which can be regarded as noise. During the preprocessing process, if the outliers are not effectively cleaned and processed, these outliers will be brought into the subsequent feature extraction process. When the embedding layer converts the text into a vector, some irrelevant features may be captured, such as spelling errors, non-standard words, etc., which may be regarded as noise. Since the embedding layer is pre-trained based on a large amount of text data, for data in some specific fields, it may not be able to capture the specific semantics of scenic spot reviews well, resulting in noise in the feature vector. When extracting the features of scenic spot introductions and online reviews, some irrelevant or redundant features may be introduced, which will increase the complexity of the model and lead to overfitting. When extracting the global features of scenic spot online reviews, the two-dimensional convolutional neural network may over-focus on some unimportant local features, which may be regarded as noise. When extracting classification features, if the number of network layers is too large or the parameters are improperly set, the convolutional neural network may also overfit the noise and outliers in the training data. In order to eliminate the noise and outliers of the scenic spot evaluation classification feature vector, in the technical solution of the present application, the scenic spot evaluation classification feature vector is subjected to sparsity constraints based on the model parameter space.

[0053] Specifically, in a specific embodiment of the present application, the scenic spot evaluation feature optimization unit 142 includes: extracting a model parameter space, wherein the model parameter space includes a model weight matrix and a model bias vector; performing sparsity constraints on the model weight matrix and the model bias vector based on a regularization term to obtain a sparse model weight matrix and a sparse model bias vector; based on the sparse model weight matrix and the sparse model bias vector, processing the scenic spot evaluation classification feature vector with the following formula to obtain the sparse optimized scenic spot evaluation classification feature vector, wherein the formula is:

[0054] V sparse =W T V c +B

[0055] Among them, V c represents the scenic spot evaluation classification feature vector, W represents the sparse model weight matrix, B represents the sparse model bias vector, T represents the transpose of the matrix, V sparse Represents the sparse optimized scenic spot evaluation classification feature vector.

[0056] Here, by performing sparsity constraints based on the model parameter space on the scenic spot evaluation classification feature vector, it can be ensured that only a few elements in the scenic spot evaluation classification feature vector are non-zero, thereby achieving the selection and optimization of features. This approach not only helps to simplify the model structure, making it more transparent and easy to explain, but also reduces the overfitting of the model to the training data and enhances its generalization ability. In addition, the sparse scenic spot evaluation classification feature vector means that the model will rely on fewer features for decision-making, which not only reduces the computational cost, but also improves the robustness of the model in the face of noise and outliers. In the process of implementing the sparsity constraints based on the model parameter space, the degree of sparsity can be clearly controlled, so as to find an appropriate balance between model complexity and prediction performance. This method can also assist in feature selection, highlighting those features that are most critical to the classification task, while protecting sensitive information in the data and ensuring privacy security.

[0057] Among them, based on the sparse model weight matrix and the sparse model bias vector, the scenic spot evaluation classification feature vector is processed with the following formula to obtain the sparse optimized scenic spot evaluation classification feature vector, including: creating a controller class, the controller class is used to process the sparsity constraint request; in response to receiving the sparsity constraint request, extracting the sparse model weight matrix, the sparse model bias vector and the scenic spot evaluation classification feature vector from the sparsity constraint request; using the controller class, and using the sparse model weight matrix and the sparse model bias vector, sparsely constraining the scenic spot evaluation classification feature vector to obtain the sparse optimized scenic spot evaluation classification feature vector; returning the sparse optimized scenic spot evaluation classification feature vector.

[0058] In particular, using the Spring Web mechanism to implement "application sparsity constraints" has many advantages, such as scalability: Spring Web provides an extensible framework that can easily handle high concurrent requests; flexibility: Spring Web supports multiple deployment options, including application servers and cloud platforms, providing flexibility; security: Spring Web provides built-in security features such as authentication and authorization to protect applications from unauthorized access; maintainability: Spring Web follows the principles of loose coupling and dependency injection, making applications easy to maintain and extend. In addition, using the Spring Web mechanism can simplify the deployment and management of sparsity constraints because Spring Web provides out-of-the-box features such as request processing, model loading, and response generation.

[0059] The specific code implementation process is:

[0060]

[0061] Then, considering that the classifier can help the system identify malicious comments, that is, those containing false information, malicious attacks or inappropriate content, so as to help the platform to timely discover and deal with these harmful information and protect the interests of users and platforms. Therefore, the classifier is used to extract and analyze the sparse optimized scenic spot evaluation classification feature vector to obtain a classification result, which is used to indicate whether there are malicious comments in the platform.

[0062] In a specific embodiment of the present application, the scenic spot evaluation feature classification unit 143 is used to: use the fully connected layer of the classifier to fully connect the sparse optimized scenic spot evaluation classification feature vector to obtain a fully connected encoded feature vector; pass the fully connected encoded feature vector through the Softmax classification function of the classifier to obtain a first probability attributable to the presence of malicious comments in the platform and a second probability attributable to the absence of malicious comments in the platform; and determine the classification result based on a comparison between the first probability and the second probability.

[0063] It is worth mentioning that in addition to using a classifier to perform feature classification on the sparse optimized scenic spot evaluation classification feature vector, a clustering algorithm can also be used to analyze and process the scenic spot online review information. The following is the implementation process of this method: 1. Data preprocessing: First, the scenic spot online review information is cleaned and preprocessed, including removing special characters, stop words and punctuation marks, and performing operations such as stem extraction or word form restoration to prepare the data for subsequent processing. 2. Feature extraction: After data preprocessing, the text data can be converted into numerical feature vectors using techniques such as bag of words or word embeddings for clustering analysis. 3. Cluster analysis: Cluster analysis is performed on the feature vectors of scenic spot online review information using clustering algorithms (such as K-means clustering, hierarchical clustering, DBSCAN, etc.), and the review texts are divided into different groups according to their similarities. 4. Malicious comment identification: After clustering is completed, potential malicious comment groups can be identified by manual inspection or using some rules. Malicious comments are usually clustered in the same category because they may contain similar language patterns or content. 5. Result display: Finally, based on the clustering results, the system can display which comments are classified as potential malicious comment groups to help platform managers better understand and handle these comments.

[0064] In summary, the text data mining and analysis system based on artificial intelligence according to the embodiment of the present application is explained, which first collects the relevant introduction data of the scenic spot and multiple online reviews of the scenic spot in the platform as input data, and then uses data mining and deep learning technology to extract features and analyze these input data to obtain classification results for indicating whether there are malicious comments in the platform. In this way, according to the classification results, it helps the scenic spot to achieve more effective data processing and information utilization, thereby promoting the development of the smart tourism industry.

[0065] As described above, the text data mining and analysis system 100 based on artificial intelligence according to the embodiment of the present application can be implemented in various wireless terminals, such as a server for text data mining and analysis based on artificial intelligence. In one example, the text data mining and analysis system 100 based on artificial intelligence according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the text data mining and analysis system 100 based on artificial intelligence can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the text data mining and analysis system 100 based on artificial intelligence can also be one of the many hardware modules of the wireless terminal.

[0066] Alternatively, in another example, the artificial intelligence-based text data mining and analysis system 100 and the wireless terminal may also be separate devices, and the artificial intelligence-based text data mining and analysis system 100 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0067] Exemplary Methods

[0068] Figure 6 The figure shows a flow chart of a text data mining and analysis method based on artificial intelligence according to an embodiment of the present application. Figure 6 As shown, according to the artificial intelligence-based text data mining and analysis method of the embodiment of the present application, the method includes: S1, collecting scenic spot related introduction data and multiple scenic spot online review data in the platform; S2, respectively extracting and analyzing the scenic spot related introduction data and the multiple scenic spot online review data in the platform to obtain a scenic spot comprehensive introduction feature vector and a scenic spot online review global feature matrix; S3, multiplying the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix to obtain a scenic spot evaluation feature vector; S4, obtaining a classification result based on the scenic spot evaluation feature vector, and the classification result is used to indicate whether there are malicious comments in the platform.

[0069] Here, those skilled in the art can understand that the specific functions and operations of each step in the above-mentioned text data mining and analysis method based on artificial intelligence have been referred to above. Figure 1 The invention has been introduced in detail in the description of the artificial intelligence-based text data mining and analysis system, and therefore, its repeated description will be omitted.

[0070] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, apparatuses and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0071] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0072] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0073] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0074] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or systems stated in a system claim can also be implemented by one unit or system through software or hardware. Secondary words are used to indicate names, but not to indicate any particular order.

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution 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 solution of the present invention can be modified or replaced by equivalents without departing from the spirit of the technical solution of the present invention.

Claims

1. A text data mining and analysis system based on artificial intelligence, characterized in that: include: The scenic spot related data collection module is used to collect scenic spot related introduction data and multiple scenic spot online review data on the platform; A scenic spot related data feature acquisition module is used to extract and analyze the features of the scenic spot related introduction data and the multiple scenic spot online review data in the platform to obtain the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix; A scenic spot related data feature fusion module, used for multiplying the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix to obtain a scenic spot evaluation feature vector; The scenic spot related data feature classification result generation module is used to obtain a classification result based on the scenic spot evaluation feature vector, and the classification result is used to indicate whether there are malicious comments in the platform.

2. The text data mining and analysis system based on artificial intelligence according to claim 1, characterized in that: The scenic spot related data feature acquisition module includes: A scenic spot introduction feature extraction unit, used for performing feature encoding operation on the scenic spot related introduction data to obtain a comprehensive scenic spot introduction feature vector; The scenic spot online review feature extraction unit is used to perform feature encoding operations on multiple scenic spot online review data in the platform to obtain the scenic spot online review global feature matrix.

3. The text data mining and analysis system based on artificial intelligence according to claim 2 is characterized in that: The scenic spot introduction feature extraction unit comprises: A scenic spot introduction text embedding coding subunit is used to arrange the scenic spot related introduction data into multiple scenic spot related introduction text sequences according to the type dimension and then perform embedding coding respectively to obtain multiple scenic spot introduction embedding vectors; The scenic spot introduction context feature acquisition subunit is used to pass the multiple scenic spot introduction embedding vectors through the scenic spot comprehensive introduction feature extraction module based on the context encoder to obtain the scenic spot comprehensive introduction feature vector.

4. The text data mining and analysis system based on artificial intelligence according to claim 3 is characterized in that: The scenic spot introduction context feature acquisition subunit is used to: Inputting the plurality of scenic spot introduction embedding vectors into the converter-based context encoder to obtain a plurality of contextual scenic spot introduction feature vectors; as well as The multiple contextual scenic spot introduction feature vectors are cascaded to obtain the scenic spot comprehensive introduction feature vector.

5. The text data mining and analysis system based on artificial intelligence according to claim 4 is characterized in that: The scenic spot online review feature extraction unit includes: A scenic spot online review encoding subunit, used to embed and encode multiple scenic spot online review data in the platform to obtain multiple scenic spot online review embedding vectors; A scenic spot online review feature extraction subunit, used to pass the plurality of scenic spot online review embedding vectors through a scenic spot online review feature extraction module based on a feature encoder to obtain a plurality of scenic spot online review feature vectors; The scenic spot online review global feature extraction subunit is used to arrange the multiple scenic spot online review feature vectors into a scenic spot online review comprehensive input matrix and then pass it through a scenic spot online review global feature extraction module based on a two-dimensional convolutional neural network model to obtain the scenic spot online review global feature matrix.

6. The text data mining and analysis system based on artificial intelligence according to claim 5 is characterized in that: The scenic spot related data feature classification result generation module includes: A scenic spot evaluation feature extraction unit, used for passing the scenic spot evaluation feature vector through a scenic spot evaluation feature extraction module based on a convolutional neural network model to obtain a scenic spot evaluation classification feature vector; A scenic spot evaluation feature optimization unit, used for performing sparsity constraints on the scenic spot evaluation classification feature vector based on the model parameter space to obtain a sparse optimized scenic spot evaluation classification feature vector; The scenic spot evaluation feature classification unit is used to pass the sparse optimized scenic spot evaluation classification feature vector through a classifier to obtain the classification result.

7. The text data mining and analysis system based on artificial intelligence according to claim 6 is characterized in that: The scenic spot evaluation feature optimization unit includes: Extracting a model parameter space, wherein the model parameter space includes a model weight matrix and a model bias vector; Performing sparsity constraints based on regularization terms on the model weight matrix and the model bias vector to obtain a sparse model weight matrix and a sparse model bias vector; Based on the sparse model weight matrix and the sparse model bias vector, the scenic spot evaluation classification feature vector is processed by the following formula to obtain the sparse optimized scenic spot evaluation classification feature vector, wherein the formula is: V sparse =W T V c +B Among them, V c represents the scenic spot evaluation classification feature vector, W represents the sparse model weight matrix, B represents the sparse model bias vector, T represents the transpose of the matrix, V sparse Represents the sparse optimized scenic spot evaluation classification feature vector.

8. The text data mining and analysis system based on artificial intelligence according to claim 7, characterized in that: Based on the sparse model weight matrix and the sparse model bias vector, the scenic spot evaluation classification feature vector is processed to obtain the sparse optimized scenic spot evaluation classification feature vector, including: Creating a controller class, wherein the controller class is used to process a sparsity constraint request; In response to receiving the sparsity constraint request, extracting the sparse model weight matrix, the sparse model bias vector and the scenic spot evaluation classification feature vector from the sparsity constraint request; Using the controller class, the sparse model weight matrix and the sparse model bias vector, sparsely constraining the scenic spot evaluation classification feature vector to obtain the sparse optimized scenic spot evaluation classification feature vector; Returns the sparse optimized scenic spot evaluation classification feature vector.

9. A text data mining and analysis method based on artificial intelligence, characterized in that ,include: Collect relevant introduction data of scenic spots and multiple online reviews of scenic spots on the platform; Extracting and analyzing the features of the scenic spot introduction data and the multiple scenic spot online review data in the platform respectively to obtain a scenic spot comprehensive introduction feature vector and a scenic spot online review global feature matrix; Multiplying the scenic spot comprehensive introduction feature vector and the scenic spot online review global feature matrix to obtain a scenic spot evaluation feature vector; A classification result is obtained based on the scenic spot evaluation feature vector, and the classification result is used to indicate whether there are malicious comments in the platform.

10. The text data mining and analysis method based on artificial intelligence according to claim 9 is characterized in that: Feature extraction and analysis are performed on the scenic spot related introduction data and multiple scenic spot online review data in the platform to obtain a scenic spot comprehensive introduction feature vector and a scenic spot online review global feature matrix, including: A scenic spot introduction feature extraction unit, used for performing feature encoding operation on the scenic spot related introduction data to obtain a comprehensive scenic spot introduction feature vector; The scenic spot online review feature extraction unit is used to perform feature encoding operations on multiple scenic spot online review data in the platform to obtain the scenic spot online review global feature matrix.

Images