Novel power system feature extraction method, device, medium and product
Through jieba word segmentation and TF-IDF algorithm, the text features of the power system are extracted, which solves the problems of time-consuming and labor-intensive traditional methods and inaccurate Chinese feature extraction, and realizes efficient and accurate feature extraction, which improves power system management and intelligent development. ability.
Patent Information
- Application Number
- CN202411260221.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-10
- Publication Date
- 2024-12-13
AI Technical Summary
Traditional power system feature extraction methods are time-consuming and labor-intensive, and difficult to adapt to complex and changeable power system environments. Especially the Chinese text feature extraction is inaccurate, which makes it impossible to be directly used for intelligent analysis.
The text features are divided into word processing using Jieba Chinese vocabulary database combined with white list vocabulary database and black list vocabulary database, and the high-frequency feature vocabulary is extracted using TF-IDF algorithm, and the weighted frequency is determined according to the proportion of data sources to achieve the accuracy of the characteristics of the new power system. extract.
It has improved the accuracy and efficiency of the feature extraction of new power systems, helped governments and enterprises to better understand the development of power systems, improve overall planning and management capabilities, and provide technical support for the intelligent and sustainable development of the power industry.
Smart Images
Figure CN119128158A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of novel power system feature extraction, and in particular to a novel power system feature extraction method, equipment, medium and product. Background Art
[0002] In today's era of rapid development of digitalization and intelligence, natural language processing (NLP) technology has been widely used in various fields. Especially in the power system, with the surge in power equipment and power grid operation data, how to efficiently and accurately extract valuable information from massive data information and identify the development characteristics of new power systems has become a key issue. Traditional power system feature extraction methods mainly rely on artificial rules and statistical models. These methods are not only time-consuming and labor-intensive, but also difficult to adapt to the complex and changing power system environment.
[0003] With the continuous advancement of artificial intelligence technology, especially the development of deep learning and semantic recognition technology, feature extraction methods based on NLP have gradually become a research hotspot. Text processing, as an important link in NLP, is related to the accuracy and precision of feature extraction. Compared with English, Chinese is a language that is highly dependent on the language environment and has no clear word segmentation boundaries. Due to the polysemy and flexibility of combination of Chinese characters, Chinese is more prone to semantic ambiguity than English, which makes the extracted text features related to the power system inaccurate, resulting in the inability to be directly used in the learning process of intelligent analysis. Summary of the invention
[0004] The purpose of this application is to provide a new power system feature extraction method, equipment, medium and product to improve the accuracy of new power system feature extraction.
[0005] To achieve the above objectives, this application provides the following solutions:
[0006] In a first aspect, the present application provides a novel power system feature extraction method, comprising:
[0007] Obtain website source code and policy documents for new power systems;
[0008] Extracting text features of the website source code and the policy document, and constructing a text feature database;
[0009] Performing data cleaning on the text feature database to obtain a text feature database after data cleaning;
[0010] The text features in the text feature database after data cleaning are segmented by using the Chinese word segmentation library jieba, combined with the white-name word library and the black-name word library in the field of the new power system, and the word frequency is counted to obtain the text feature database after word segmentation; the white-name word library is a word library with a custom dictionary added on the basis of the dictionary provided by the Chinese word segmentation library jieba; the custom dictionary includes professional vocabulary in the power industry; the black-name word library is a word library with a stop word corpus added on the basis of the dictionary provided by the Chinese word segmentation library jieba; the stop word corpus includes disordered characters and meaningless characters; the disordered characters include spaces, punctuation marks, numbers and special characters; the meaningless characters include "is", "a kind of" and "etc.;
[0011] Classifying the words in the text feature database after word segmentation according to data sources to obtain a classified text feature database; the data sources include energy policies, energy industry information and news reports; the classified text feature database includes energy policy words, energy industry information words and news report words;
[0012] Using the TF-IDF algorithm to extract high-frequency feature words from the energy policy words, the energy industry information words, and the news report words;
[0013] Determining the weighted frequency of the high-frequency feature words based on the data source ratio;
[0014] The weighted frequencies of the high-frequency characteristic words are sorted in descending order, and a preset number of the high-frequency characteristic words are used as the development characteristics of the new power system.
[0015] Optionally, extracting text features of the website source code specifically includes:
[0016] Build a web crawler model;
[0017] Based on the website source code, using the web crawler model, extracting the target web page;
[0018] The Beautiful Soup library is used to extract text features from the target web page.
[0019] Optionally, extracting text features of the policy document specifically includes:
[0020] Extract the text content of the policy document based on the fitz library in the PyMuPDF library;
[0021] The text features of the text content are extracted using a text extraction function and a page turning function.
[0022] Optionally, performing data cleaning on the text feature database to obtain a text feature database after data cleaning specifically includes:
[0023] Removing the disordered characters in the text feature database;
[0024] Removing the meaningless characters in the text feature database;
[0025] The blank parts of the text in the text feature database are removed to obtain a text feature database after data cleaning.
[0026] Optionally, the TF-IDF algorithm is used to extract high-frequency feature words from the energy policy words, the energy industry information words, and the news report words, specifically including:
[0027] Calculating the frequency of each word in the energy policy vocabulary, the energy industry information vocabulary and the news report vocabulary respectively;
[0028] Calculating the inverse document frequency of each word in the energy policy vocabulary, the energy industry information vocabulary, and the news report vocabulary respectively;
[0029] According to the word frequency and the inverse document frequency, the TF-IDF algorithm is used to determine the high-frequency feature words in the energy policy words, the energy industry information words and the news report words.
[0030] Optionally, according to the word frequency and the inverse document frequency, the high-frequency feature words in the energy policy words, the energy industry information words and the news report words are determined by using the TF-IDF algorithm, specifically including:
[0031] Using the formula TF-IDF=TF ij IDF ij Determine the TF-IDF value for each word; where TF ij is word frequency; IDF ij is the inverse text frequency;
[0032] The words whose TF-IDF values among the energy policy words, the energy industry information words and the news report words are greater than a set threshold are used to obtain high-frequency feature words.
[0033] Optionally, based on the data source ratio, determining the weighted frequency of the high-frequency feature vocabulary specifically includes:
[0034] Using the formula fre i =w i ·v i Determine the weighted frequency of the high-frequency feature words; wherein, frei is the weighted frequency of high-frequency feature words; w i is the frequency of high-frequency feature words; v i It is the weight value of the high-frequency feature words under a certain data source, and the weight value is determined according to the proportion of the data source.
[0035] In a second aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the novel power system feature extraction methods described above.
[0036] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the novel power system feature extraction methods described above.
[0037] In a fourth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the novel power system feature extraction methods described above.
[0038] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0039] This application provides a new power system feature extraction method, equipment, medium and product, which performs word segmentation on the text feature database based on the Jieba Chinese word segmentation library, sets weights for the new power system database under multiple data sources, and uses the TF-IDF algorithm to conduct in-depth analysis and processing of the word segmentation results, thereby achieving more efficient and accurate feature extraction. This application will help deepen the government, companies and energy and power-related enterprises' understanding of the development of new power systems and the reform and innovation of state-owned enterprises, enhance overall planning and management capabilities, and provide technical support for promoting the intelligent and sustainable development of the power industry. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0041] Figure 1 A schematic diagram of a flow chart of a novel power system feature extraction method provided in one embodiment of the present application;
[0042] Figure 2 A flowchart of the novel power system feature extraction method of the present application in practical application;
[0043] Figure 3 This is the flow chart of Jieba word segmentation;
[0044] Figure 4 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION
[0045] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0046] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0047] The present invention provides a novel power system feature extraction method 0 to solve the problems that the traditional power system feature extraction method is time-consuming and labor-intensive and difficult to adapt to the complex and changeable power system environment.
[0048] In an exemplary embodiment, Figure 1 and Figure 2 As shown, a novel power system feature extraction method is provided, comprising the following steps:
[0049] S1: Obtain the website source code and policy documents of the new power system.
[0050] S2: Extracting text features of the website source code and the policy document, and constructing a text feature database.
[0051] As an optional implementation method, extracting text features of the website source code specifically includes:
[0052] Build a web crawler model.
[0053] Based on the website source code, the target web page is extracted using the web crawler model.
[0054] The Beautiful Soup library is used to extract text features from the target web page.
[0055] In practical applications, the source code of the website related to the new power system is obtained and text features are extracted, = a text feature database is constructed in TXT format. The specific steps are as follows:
[0056] S1-1, when extracting relevant web page information, select a suitable seed URL as a starting point to build a web crawler model.
[0057] S1-2, uses the Session object of the Requests library to initiate a request, automatically processes Cookies, thereby maintaining the web page access status, and automatically decodes the response content from the server, so that the model can quickly process the server response, check the status code, and parse the content.
[0058] S1-3, select the Beautiful Soup library to extract text features from the target web page HTML or XML file.
[0059] S1-4, extract the text features from the web page and save them in TXT format to facilitate subsequent text recognition.
[0060] As an optional implementation method, extracting text features of the policy document specifically includes:
[0061] The text content of the policy document is extracted based on the fitz library in the PyMuPDF library.
[0062] The text features of the text content are extracted using a text extraction function and a page turning function.
[0063] In practical applications, we obtain policy documents related to the new power system, extract text from existing PDF files, and merge them with the text feature database constructed above. The specific steps are as follows:
[0064] S2-1, extract text content from PDF files based on the fitz library in the PyMuPDF library.
[0065] S2-2, define a text extraction function to extract text from the target file.
[0066] S2-3, define a page turning function to turn the pages of the target file and extract the content of each page.
[0067] S2-4, close the PDF file and save the extracted text features as a TXT format document.
[0068] S3: performing data cleaning on the text feature database to obtain a text feature database after data cleaning.
[0069] As an optional implementation, S3 specifically includes:
[0070] The disordered characters in the text feature database are removed.
[0071] The meaningless characters in the text feature database are removed.
[0072] Remove the blank parts in the text feature database to obtain the text feature database after data cleaning.
[0073] In practical applications, based on the text feature database constructed above, data cleaning is performed, and the specific process is as follows:
[0074] S3-1, Remove punctuation marks and special characters, that is, scrambled characters.
[0075] S3-2, Remove meaningless stop words such as "de, shi, le, yinci, erqie" (i.e., meaningless characters), reduce data noise, so as to perform modeling and analysis on the text in the database subsequently.
[0076] S3-3, Remove the blank parts of the text to reduce the effective text length.
[0077] S4: Use the Chinese word segmentation library jieba, combine the whitelist word library and blacklist word library in the field of new power systems to perform word segmentation on the text features in the text feature database after data cleaning, and count the word frequencies to obtain the text feature database after word segmentation; the whitelist word library is a word library obtained by adding a custom dictionary on the basis of the built-in dictionary of the Chinese word segmentation library jieba; the custom dictionary includes professional vocabulary in the power industry; the blacklist word library is a word library obtained by adding a stop word corpus on the basis of the built-in dictionary of the Chinese word segmentation library jieba; the stop word corpus includes scrambled characters and meaningless characters; the scrambled characters include spaces, punctuation marks, numbers, and special characters; the meaningless characters include "shi, yizhong, deng".
[0078] In practical applications, define the whitelist word library and blacklist word library in the field of new power systems. The whitelist word library includes adding a custom dictionary containing a large number of professional vocabulary in the power industry on the basis of the built-in dictionary "dict" of "jieba". The blacklist word library includes adding scrambled characters such as spaces, punctuation marks (including (), [], etc.), special characters (including #), numbers, and meaningless characters such as "shi", "yizhong", "deng" as blacklist words to the "Stopwords" library on the basis of the built-in dictionary "dict" of "jieba".
[0079] Use "jieba" for word segmentation on the text features after data cleaning, and count the word frequencies. The specific steps are as follows:
[0080] S4-1, Perform a first word segmentation through the accurate mode of the jieba Chinese word segmentation library, sample some original texts from the text feature databases of different data sources for testing, and observe and identify the word segmentation results.
[0081] S4-2, compare the word segmentation results with the original text, correct the wrong word segmentation results, and add typical wrong word segmentations to the Stopwords library (stop word library) as blacklist words; add professional vocabulary related to the power system to the whitelist library, and reasonably set the vocabulary weights.
[0082] S4-3, perform secondary word segmentation, analyze the word segmentation results, and achieve the ideal word segmentation results by repeatedly iteratively calculating the weights of blacklist words and whitelist words.
[0083] The specific flow chart of word segmentation is as follows Figure 3 As shown, specifically including:
[0084] 1. Text input into the text feature database after data cleaning.
[0085] 2. Load the Stopwords library for data cleaning.
[0086] 3. Load the whitelist vocabulary library to establish the Trie tree segmentation model.
[0087] 4. Determine whether the word segmentation vocabulary is registered in the dictionary.
[0088] 5. For the logged-in vocabulary, a word segmentation DAG word graph is constructed, in which each Chinese character is a node in the graph. If there is a word starting with this Chinese character in the dictionary, point from this Chinese character to the last Chinese character of the word; use dynamic programming to calculate the global probability to obtain the maximum word frequency segmentation combination based on the whitelist vocabulary library.
[0089] 6. For unregistered words, separate the Chinese and English parts based on Token recognition and convert them into Token sequences; load the hidden Markov chain HMM probability model graph, and use the Viterbi algorithm dynamic programming to obtain word segmentation and annotation.
[0090] 7. Finally output the word segmentation results.
[0091] S5: Classify the words in the text feature database after word segmentation according to the data source to obtain a classified text feature database; the data source includes energy policy, energy industry information and news reports; the classified text feature database includes energy policy words, energy industry information words and news report words.
[0092] In practical applications, the text feature database text after data cleaning is classified according to different data sources, and the high-frequency feature vocabulary of the new power system is extracted based on the TF-IDF algorithm. The specific data sources can be divided into energy policies, energy industry information, news reports, etc., aiming to ensure the comprehensiveness of the construction of the new power system text database.
[0093] S6: Using the TF-IDF algorithm, high-frequency feature words are extracted from the energy policy words, the energy industry information words, and the news report words.
[0094] As an optional implementation, S6 specifically includes:
[0095] The frequency of each word in the energy policy vocabulary, the energy industry information vocabulary and the news report vocabulary is calculated respectively.
[0096] In practical applications, the formula Calculate word frequency; among them, TF ij is the normalized frequency of the jth characteristic word of the ith new power system; n ij is the number of times a new type of power system characteristic word appears in the text, ∑ k n kj is the number of all feature words in the text.
[0097] The inverse document frequency of each word in the energy policy vocabulary, the energy industry information vocabulary and the news report vocabulary is calculated respectively.
[0098] In practical applications, IDF represents the inverse document frequency of a word. The closer its value is to 0, the more times the word appears in the text feature database. Its calculation formula is as follows:
[0099]
[0100] Among them, N is the number of all articles in the database, 1+N1 represents the words with obvious characteristics in the article. In order to prevent the word from not existing in the database and the denominator being 0, 1+N1 is used as the denominator. IDF ij is the inverse text frequency of the jth word in the i-th new power system.
[0101] According to the word frequency and the inverse document frequency, the TF-IDF algorithm is used to determine the high-frequency feature words in the energy policy words, the energy industry information words and the news report words.
[0102] As an optional implementation, according to the word frequency and the inverse document frequency, the TF-IDF algorithm is used to determine the high-frequency feature words in the energy policy vocabulary, the energy industry information vocabulary and the news report vocabulary, specifically including:
[0103] Using the formula TF-IDF=TF ij IDF ij Determine the TF-IDF value for each word; where TF ij is word frequency; IDFij The larger the TF-IDF value is, the higher the frequency of the word appears, and the more it can be shown that the word can appear as a new type of power system characteristic word.
[0104] The words whose TF-IDF values among the energy policy words, the energy industry information words and the news report words are greater than a set threshold are used to obtain high-frequency feature words.
[0105] S7: Determine the weighted frequency of the high-frequency feature words based on the data source ratio. In practical applications, based on the importance of the source, the three sources of national policies, energy industry, and news reports are weighted in a ratio of 6:3:1, and the weighted frequency of each word is calculated.
[0106] As an optional implementation, S7 specifically includes:
[0107] Using the formula fre i =wi·vi determines the weighted frequency of the high-frequency feature words; wherein, fre i is the weighted frequency of high-frequency feature words; w i is the frequency of high-frequency feature words; v i It is the weight value of the high-frequency feature words under a certain data source, and the weight value is determined according to the proportion of the data source.
[0108] S8: sorting the weighted frequencies of the high-frequency characteristic words in descending order, and using a preset number of the high-frequency characteristic words as development characteristics of the new power system.
[0109] In practical applications, the weighted frequencies of the obtained high-frequency characteristic words of the new power system are sorted in descending order, and the top ten words are selected as the development characteristics of the new power system.
[0110] This application proposes a new power system feature extraction method, which uses the advanced Jieba Chinese word segmentation library to conduct in-depth analysis and processing of relevant new power system databases from various aspects, thereby achieving more efficient and accurate feature extraction. This method will help deepen the understanding of the government, companies and energy and power-related enterprises on the development of new power systems and the reform and innovation of state-owned enterprises, improve overall planning and management capabilities, and provide technical support for promoting the intelligent and sustainable development of the power industry.
[0111] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 4As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a new method for extracting characteristics of a power system is implemented.
[0112] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above-mentioned novel power system feature extraction method when executing the computer program.
[0113] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the novel power system feature extraction method described above is implemented.
[0114] In an exemplary embodiment, a computer program product is provided, including a computer program, which implements the above-mentioned novel power system feature extraction method when executed by a processor.
[0115] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0116] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0117] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.
[0118] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0119] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A novel power system feature extraction method, characterized in that: include: Obtain website source code and policy documents for new power systems; Extracting text features of the website source code and the policy document, and constructing a text feature database; Performing data cleaning on the text feature database to obtain a text feature database after data cleaning; The text features in the text feature database after data cleaning are segmented by using the Chinese word segmentation library jieba, combined with the white-name word library and the black-name word library in the field of the new power system, and the word frequency is counted to obtain the text feature database after word segmentation; the white-name word library is a word library with a custom dictionary added on the basis of the dictionary provided by the Chinese word segmentation library jieba; the custom dictionary includes professional vocabulary in the power industry; the black-name word library is a word library with a stop word corpus added on the basis of the dictionary provided by the Chinese word segmentation library jieba; the stop word corpus includes disordered characters and meaningless characters; the disordered characters include spaces, punctuation marks, numbers and special characters; the meaningless characters include "is", "a kind of" and "etc.; Classifying the words in the text feature database after word segmentation according to data sources to obtain a classified text feature database; the data sources include energy policies, energy industry information and news reports; the classified text feature database includes energy policy words, energy industry information words and news report words; Using the TF-IDF algorithm to extract high-frequency feature words from the energy policy words, the energy industry information words, and the news report words; Determining the weighted frequency of the high-frequency feature words based on the data source ratio; The weighted frequencies of the high-frequency characteristic words are sorted in descending order, and a preset number of the high-frequency characteristic words are used as the development characteristics of the new power system.
2. The novel power system feature extraction method according to claim 1 is characterized in that: Extracting text features of the website source code, specifically including: Build a web crawler model; Based on the website source code, using the web crawler model, extracting the target web page; The Beautiful Soup library is used to extract text features from the target web page.
3. The novel power system feature extraction method according to claim 1 is characterized in that: Extract the text features of the policy document, including: Extract the text content of the policy document based on the fitz library in the PyMuPDF library; The text features of the text content are extracted using a text extraction function and a page turning function.
4. The novel power system feature extraction method according to claim 1 is characterized in that: Performing data cleaning on the text feature database to obtain a text feature database after data cleaning specifically includes: Removing the disordered characters in the text feature database; Removing the meaningless characters in the text feature database; The blank parts of the text in the text feature database are removed to obtain a text feature database after data cleaning.
5. The novel power system feature extraction method according to claim 1 is characterized in that: The TF-IDF algorithm is used to extract high-frequency feature words from the energy policy words, the energy industry information words, and the news report words, including: Calculating the frequency of each word in the energy policy vocabulary, the energy industry information vocabulary and the news report vocabulary respectively; Calculating the inverse document frequency of each word in the energy policy vocabulary, the energy industry information vocabulary, and the news report vocabulary respectively; According to the word frequency and the inverse document frequency, the high-frequency feature words in the energy policy words, the energy industry information words and the news report words are determined by using the TF-IDF algorithm.
6. The novel power system feature extraction method according to claim 5 is characterized in that: According to the word frequency and the inverse document frequency, the high-frequency feature words in the energy policy words, the energy industry information words and the news report words are determined by using the TF-IDF algorithm, specifically including: Using the formula TF-IDF=TF ij IDF ij Determine the TF-IDF value for each word; where TF ij is word frequency; IDF ij is the inverse text frequency; The words whose TF-IDF values among the energy policy words, the energy industry information words and the news report words are greater than a set threshold are used to obtain high-frequency feature words.
7. The novel power system feature extraction method according to claim 1 is characterized in that: Based on the data source ratio, the weighted frequency of the high-frequency feature words is determined, specifically including: Using the formula fre i =w i ·v i Determine the weighted frequency of the high-frequency feature words; wherein, fre i is the weighted frequency of high-frequency feature words; w i is the frequency of high-frequency feature words; v i It is the weight value of the high-frequency feature words under a certain data source, and the weight value is determined according to the proportion of the data source.
8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the novel power system feature extraction method described in any one of claims 1-7.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the novel power system feature extraction method described in any one of claims 1 to 7 is implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the novel power system feature extraction method described in any one of claims 1 to 7 is implemented.