A new energy policy information extraction method and system

Abstract

The present invention proposes a new energy policy information extraction method and system, which processes the new energy policy into a new energy policy text in a set format; based on a pre-established new energy policy feature dictionary library, the new energy policy text is divided into sentence levels and annotated using a method of automatically identifying sentences using punctuation marks, and the division and annotation results are output to generate a text sentence library; based on a unified service interface, key information of a new energy policy feature dictionary library and a text sentence library is generated and output according to the text sentences.

Description

Technical Field

[0001] The present invention belongs to the field of new energy policy evaluation and text mining, and specifically relates to a new energy policy information extraction method and system. Background Art

[0002] Under the multiple challenges of environmental pollution, climate change and the depletion of fossil energy, new energy, mainly renewable energy, has played an increasingly important role in the energy strategies of various countries. In order to promote the development of the renewable energy industry, my country promulgated the Renewable Energy Law in 2005 and established a renewable energy target and planning system, a fixed electricity price and mandatory grid connection system, a cost sharing and subsidy system, etc. Subsequently, as of May 2018, the central government and governments at all levels have issued a series of policies and related supporting documents around the implementation of the Renewable Energy Law, gradually forming a new energy and renewable energy law and policy system dominated by the Renewable Energy Law.

[0003] In the traditional model, government departments often face technical difficulties such as limited data sources, insufficient data processing capabilities, and low timeliness in the process of policy formulation. There are problems such as difficulty in defining quantitative indicators, difficulty in assessing policy implementation, and lack of evaluation methods for policy rationality. It is urgent to adopt new methods and means to analyze and evaluate the effectiveness and role of policy implementation. To this end, on July 1, 2015, the General Office of the State Council issued the "Opinions of the General Office of the State Council on Using Big Data to Strengthen Services and Supervision of Market Subjects", proposing "using big data to scientifically formulate and adjust regulatory systems and policies, track and monitor the implementation effects of relevant systems and policies" and "making full use of big data technology to improve economic operation monitoring and forecasting and risk warning."

[0004] To analyze and evaluate policies, we first need to interpret the core content of the policies and extract the core information. The traditional processing method is for analysts and evaluators to manually interpret the policies, obtain the key information required for analysis and evaluation, and then use this information as a basis for further analysis and processing. However, in the face of a large number of new energy policies, manual retrieval and manual classification are not only inefficient, but also create the risk of inaccurate extraction of effective information. Summary of the invention

[0005] In order to meet the above needs, the present invention proposes a new energy policy information extraction method and system, which can use a computer to quickly, accurately and concisely extract the main information expressed by the policy, obtain the policy's classification, the policy's impact range and other contents, and avoid the trouble of manual retrieval and manual classification, thereby saving time and improving efficiency.

[0006] The purpose of the present invention is achieved by adopting the following technical solutions:

[0007] A method for extracting new energy policy information, the method comprising:

[0008] Process the new energy policy into a new energy policy text in a set format;

[0009] Based on a pre-established new energy policy feature dictionary library, the new energy policy text is divided into sentence levels and annotated using a method of automatic sentence recognition using punctuation marks, and the division and annotation results are output to generate a text sentence library;

[0010] Based on the unified service interface, key information of a new energy policy feature dictionary library and a text sentence library is generated according to the text sentence and outputted.

[0011] Preferably, the establishment of the new energy policy feature dictionary library includes:

[0012] Pre-process the collected new energy policies;

[0013] By using the expert method, a sampling analysis is conducted on the preprocessed new energy policy text database to generate a new energy policy feature dictionary for storing basic feature words.

[0014] Preferably, based on the pre-established new energy policy feature dictionary library, the method of automatically identifying sentences by punctuation marks is used to divide and annotate the new energy policy text into sentence levels, and the division and annotation results are output to generate a text sentence library, which includes:

[0015] Based on the pre-established new energy policy feature dictionary, the new energy policy text is segmented to generate a text segmentation database;

[0016] According to the word frequency of the text word segmentation library, the new energy policy feature dictionary library is updated;

[0017] Establish a text sentence library based on the updated new energy policy feature dictionary.

[0018] Furthermore, the pre-processing of the collected new energy policies includes:

[0019] Collect new energy policies, and construct a text database based on new energy policies according to a predefined coding format, as well as a policy data table for managing information related to the new energy policies;

[0020] The policy data table includes the uploader, upload time, source, number, policy name, issuing department and issuing date.

[0021] Furthermore, the basic characteristic words include: electricity price, subsidy, finance and taxation, value-added tax refund, income tax exemption, financial subsidy, value-added tax refund, law, bill, action plan, planning, construction plan, full guaranteed acquisition, new energy installed capacity, newly added production capacity, cumulative grid-connected installed capacity, annual power generation, installed capacity share, power generation share, CO2 emission reduction, SO2 emission reduction and NOX emission reduction.

[0022] Furthermore, the word segmentation processing of the new energy policy text is performed based on the pre-established new energy policy feature dictionary library to generate the text word segmentation library, including:

[0023] Calling an open source word segmentation tool, loading the basic feature words into a custom word library of the open source word segmentation tool, and using the default word library of the open source word segmentation tool to perform word segmentation on the new energy policy text to obtain the part of speech, word frequency and position of the word segmentation;

[0024] Using the TF-IDF algorithm to calculate the weight of the basic feature words;

[0025] A text segmentation library is constructed based on the part of speech, word frequency and position of the segmented words and the weights of the basic feature words.

[0026] Furthermore, the weight of the basic feature word is determined by the following formula:

[0027] TF = t;

[0028] IDF = log(N / n); c

[0029] TF-IDF=TF*IDF=t*log(N / n)

[0030] In the formula, t is the number of times the basic feature word K appears in the text, N is the total number of new energy policy text documents, and n is the number of documents containing the basic feature word K.

[0031] Further, the updating of the new energy policy feature dictionary according to the word frequency of the text word segmentation library includes:

[0032] According to the weight of the basic characteristic words, determining the frequency of occurrence of the basic characteristic words in the new energy policy text;

[0033] Based on the frequency of occurrence of the basic characteristic words in the new energy policy text, a secondary expert evaluation is conducted on the basic characteristic words not included in the new energy policy characteristic dictionary library, and the basic characteristic words that pass the evaluation are updated to the new energy policy characteristic dictionary library.

[0034] Furthermore, the method of automatically identifying sentences using punctuation marks to divide and annotate the new energy policy text into sentences includes:

[0035] Use periods, question marks, and exclamation marks as sentence separators;

[0036] Treat emphasis, hyphens, spaces, book title marks and proper name marks as sentence break marks that are automatically ignored in programming;

[0037] Modify the initial program using commas, quotation marks, and brackets to complete the sentence.

[0038] Furthermore, the division and labeling results include: sentence identification, sentence content, sentence position and sentence weight, and the positional association between participles and sentences; wherein,

[0039] The sentence weight is the sum of the frequency and weight of non-stop words in the sentence.

[0040] Preferably, based on the unified service interface, the text sentence generates key information of a text word library and a text sentence library, and outputs the following information:

[0041] According to the sentence weights, the sentences to be decomposed are selected, and a key information template is generated based on the Chinese grammatical structure; the participle of the sentence in the key information template is used to generate key information; and the key information is output through a unified service interface.

[0042] Preferably, after obtaining the key information, the step further includes:

[0043] Construct a new energy policy text classification strategy library based on the types of key information; including macro laws, development plans, consumption policies, administrative management policies, fiscal and taxation policies, and electricity price policies;

[0044] According to the sentence weights and the importance of key information, the importance values ​​of the sentences are calculated, and the sentences whose importance values ​​exceed a preset threshold are defined as text summaries.

[0045] A new energy policy information extraction system, the system comprising:

[0046] A processing module, used for processing the new energy policy into a new energy policy text in a set format;

[0047] A division and annotation module is used to divide and annotate the new energy policy text into sentences based on a pre-established new energy policy feature dictionary library and a method of automatically identifying sentences using punctuation marks, output the division and annotation results, and generate a text sentence library;

[0048] The acquisition module is used to generate key information of a new energy policy feature dictionary library and a text sentence library based on the unified service interface and output the key information according to the text sentence.

[0049] Compared with the closest prior art, the beneficial effects of the present invention are:

[0050] The present invention proposes a new energy policy information extraction method and system, which first processes the new energy policy into a new energy policy text in a set format; through a unified data model, only a small amount of manual assistance is required, which is very suitable for massive data.

[0051] Secondly, based on the pre-established new energy policy feature dictionary library, the new energy policy text is divided and annotated by using punctuation marks to automatically identify sentences, and the division and annotation results are output to generate a text sentence library; there is no need to redesign the document isomorphism architecture, and the document isomorphism format definition and its operation interface can be appropriately modified to meet the needs of this field; by dividing and annotating the new energy policy text into sentences, the whole sentence in the new energy policy text can be split, and the distinction between words can be achieved, which provides conditions for the subsequent extraction of key text information. The new energy policy word library, sentence library, feature dictionary library, and classification strategy library are created separately and coupled, so that the focus of new energy policy text analysts is more concentrated and strategy optimization and improvement are more convenient.

[0052] Finally, based on the unified service interface, the key information of the new energy policy feature dictionary and the text sentence library is generated and output according to the text sentence. In the word segmentation method based on the new energy policy feature dictionary, a variety of methods are adopted to achieve the matching between the new energy policy text and the feature dictionary. Selecting the appropriate method according to the actual situation can improve the accuracy of matching with the feature dictionary, and then improve the accuracy of information extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A general flow chart of information extraction for new energy policy content provided for a specific implementation method of the present invention;

[0054] Figure 2 A flowchart of new energy policy text mining provided for a specific implementation of the present invention. DETAILED DESCRIPTION

[0055] The specific implementation modes of the present invention are further described in detail below in conjunction with the accompanying drawings.

[0056] The extraction of new energy policy content mainly refers to summarizing the new energy policy text content for the special field of new energy. The traditional text content extraction method mainly extracts the feature words that can represent the main content of the text according to the word weight, and then calculates the sentence weight according to the physical information of the sentence and the feature words contained in the sentence, and extracts the candidate summary sentence according to the sentence weight, etc. However, this method is not completely suitable for the special field of new energy.

[0057] Based on the above analysis, the purpose of the present invention is to provide a method for extracting information on new energy policy content, extracting representative keyword information that can distinguish the policy text category from the main content of the new energy policy, including policy classification information, whether the policy is quantifiable, and the scope of policy influence, etc., thereby realizing automatic parsing of new energy policy texts and extraction of key indicators, laying the foundation for the construction of a policy text library and new energy policy analysis and evaluation work.

[0058] Example 1: A method for extracting information on new energy policy content provided by the present invention, the specific business process is as follows Figure 1 and Figure 2 The detailed description is as follows:

[0059] S1 processes the new energy policy into a new energy policy text with a set format;

[0060] S2 divides and annotates the new energy policy text into sentences based on a pre-established new energy policy feature dictionary library and uses a punctuation mark automatic sentence recognition method, outputs the division and annotation results, and generates a text sentence library;

[0061] S3 generates key information of a new energy policy feature dictionary library and a text sentence library based on a unified service interface and outputs the key information according to the text sentence.

[0062] In step S1, the establishment of a new energy policy feature dictionary includes:

[0063] a. Pre-process the collected new energy policies;

[0064] b. Using the expert method, a sampling analysis is performed on the preprocessed new energy policy text database to generate a new energy policy feature dictionary for storing basic feature words.

[0065] In step a, the new energy policy text is preprocessed, that is, the new energy policy is collected and organized, a txt policy text database is constructed, the encoding format is UTF-8 for word segmentation, and a policy data table is established to manage other dimensional data information of the policy, including uploader, upload time, source, number, policy name, issuing department, issuing date and other information.

[0066] The basic characteristic words in step b include: electricity price, subsidy, finance and taxation, value-added tax refund, income tax exemption, financial subsidy, value-added tax refund, law, bill, action plan, planning, construction plan, full guarantee acquisition, new energy installed capacity, newly added production capacity, cumulative grid-connected installed capacity, annual power generation, installed capacity share, power generation share, CO2 emission reduction, SO2 emission reduction and NOX emission reduction.

[0067] The expert method is used to sample and analyze the new energy policy text database information, and a preliminary new energy policy feature dictionary is constructed based on the machine learning text automatic classification method, as shown in Table 1.

[0068] Table 1 New energy policy characteristics dictionary

[0069]

[0070] Step S2, based on the pre-established new energy policy feature dictionary library, the new energy policy text is divided into sentence levels and annotated using a method of automatic sentence recognition using punctuation marks, and the division and annotation results are output to generate a text sentence library, including:

[0071] 1) Based on the pre-established new energy policy feature dictionary, the new energy policy text is segmented to generate a text segmentation database;

[0072] 2) updating the new energy policy feature dictionary according to the word frequency of the text word segmentation library;

[0073] 3) Establish a text sentence library based on the updated new energy policy feature dictionary.

[0074] In step 1), based on the pre-established new energy policy feature dictionary library, the new energy policy text is segmented, and the generated text segmentation library includes:

[0075] Calling an open source word segmentation tool, loading the basic feature words into a custom word library of the open source word segmentation tool, and using the default word library of the open source word segmentation tool to perform word segmentation on the new energy policy text to obtain the part of speech, word frequency and position of the word segmentation;

[0076] The TF-IDF algorithm is used to calculate the weights of the basic feature words.

[0077] The weight of the basic feature word is determined by the following formula:

[0078] TF = t;

[0079] IDF = log(N / n); c

[0080] TF-IDF=TF*IDF=t*log(N / n)

[0081] In the formula, t is the number of times the basic feature word K appears in the text, N is the total number of new energy policy text documents, and n is the number of documents containing the basic feature word K.

[0082] The construction of the text segmentation library specifically includes: for a certain new energy policy txt text preprocessed in step a, calling the open source Chinese segmentation tool - Jieba Segmentation, loading the basic feature words in step b into the custom word library of Jieba Segmentation, combining the default word library of Jieba Segmentation to segment the text, obtain the part of speech, word frequency and position of the segmented words, and calculate the weight information of the basic feature words. The weight calculation method is as follows:

[0083] The weights of feature words are calculated based on the TF-IDF (term frequency–inverse document frequency) algorithm. Term frequency TF is the number of times a feature word appears in the text, and inverse document frequency IDF is the logarithm of the ratio of the total number of documents in the policy text library to the number of documents containing the feature word, which indicates the ability of the feature word to effectively distinguish the target text, that is, the weight coefficient representing the importance of the feature word in the text, which is inversely proportional to the prevalence.

[0084] Let t be the number of times the basic feature word K appears in the text (since the length of the policy texts does not vary much, t is no longer standardized), N is the total number of documents in the policy text library, and n is the number of documents containing the keyword, then:

[0085] TF = t;

[0086] IDF = log(N / n); c

[0087] TF-IDF=TF*IDF=t*log(N / n);

[0088] Step 2) According to the word frequency of the text word segmentation library, the new energy policy feature dictionary library is updated, including:

[0089] According to the weight of the basic characteristic words, determining the frequency of occurrence of the basic characteristic words in the new energy policy text;

[0090] Based on the frequency of occurrence of basic characteristic words in new energy policy texts, a secondary expert evaluation is conducted on the basic characteristic words that are not included in the new energy policy characteristic dictionary, and the basic characteristic words that pass the evaluation are updated to the new energy policy characteristic dictionary.

[0091] Step S2, using a method of automatically identifying sentences using punctuation marks to divide and annotate the new energy policy text into sentences, includes:

[0092] Use periods, question marks, and exclamation marks as sentence separators;

[0093] Treat emphasis, hyphens, spaces, book title marks and proper name marks as sentence break marks that are automatically ignored in programming;

[0094] Modify the initial program using commas, quotation marks, and brackets to complete the sentence.

[0095] The division and annotation results include: sentence identification, sentence content, sentence position and sentence weight, as well as the positional relationship between participles and sentences;

[0096] The sentence weight is the sum of the frequency and weight of non-stop words in the sentence.

[0097] In step S2, the generation of the text sentence library includes:

[0098] According to the automatic word segmentation and part-of-speech tagging work, the method of automatic sentence recognition by punctuation is adopted to automatically divide and annotate the sentence levels and relationships of the new energy policy txt text selected in step a. The output result includes sentence identification, sentence content, sentence position and sentence weight, and completes the positional association relationship between word segmentation and sentence.

[0099] ■ Punctuation marks include period, question mark, exclamation mark, comma, comma, semicolon, colon, quotation mark, bracket, dash, ellipsis, emphasis mark, hyphen, spacer, book title mark and proper name mark. First, when dividing sentences, focus on judging period, question mark and exclamation mark as sentence dividing marks (including simple and complex sentences). Secondly, exclude emphasis mark, hyphen, spacer, book title mark and proper name mark when programming, so that the computer will not use these punctuation marks as sentence breaking marks when encountering them. Finally, considering the special functions of comma, quotation mark and bracket, slightly modify the initial program and finally complete the sentence division.

[0100] ■The sentence weight is the sum of the frequency and weight of non-stop words in the sentence.

[0101] Step S3, based on the unified service interface, generates key information of the new energy policy feature dictionary and text sentence library according to the text sentence and outputs the following information:

[0102] According to the sentence weights, select the sentences that need to be decomposed, and generate key information templates based on the Chinese grammatical structure; for example, noun + verb + numeral + quantifier + time word + place word;

[0103] The key information is generated by using the part of speech of the sentence in the key information template; and the key information is output through a unified service interface.

[0104] Key information generated includes:

[0105] According to the sentence weight, the sentence to be decomposed is selected, and the key information template is generated based on the Chinese grammatical structure. According to the completeness of the generated information, the information importance of the sentence is evaluated. The rule is the sum of the importance of the part of speech in the template. The specific values ​​of the importance of the part of speech are as follows:

[0106] Part of Speech noun verb numeral quantifier Time Words Place words value 5 4 3 3 2 2

[0107] After step S3 generates key information, it also includes:

[0108] Construct a new energy policy text classification strategy library based on the type of key information; including macro laws, development plans, consumption policies, administrative management policies, fiscal and taxation policies, and electricity price policies.

[0109] According to the sentence weights and the importance of key information, the importance values ​​of the sentences are calculated, and the sentences whose importance values ​​exceed a preset threshold are defined as text summaries.

[0110] Among them, the expert method is used to build a new energy policy text classification strategy library to judge and return the policy classification. The basic strategy is shown in Table 2.

[0111] Table 2 New energy policy text classification strategy library

[0112]

[0113] 1) Text summary generation

[0114] According to the sentence weight in 4) and the information importance of the clauses in 5), the importance value of the sentence is recalculated, a certain threshold is set, and the sentences exceeding the threshold are output to generate a summary.

[0115] 2) Text information output

[0116] Design a universal interface to output key information and ensure the normative and standard nature of information transmission.

[0117] The general xml.webservice is used to develop the interface and standardize the output of new energy policy text information. The specific output examples are as follows:

[0118]

[0119]

[0120] In summary, through the separate creation and coupling processing of the above-mentioned new energy policy word segmentation library, sentence library, feature dictionary library, and classification strategy library, the focus of new energy policy text analysts is more focused and strategy optimization and improvement is more convenient.

[0121] Subsequently, the key information obtained through this embodiment can be imported into the new energy policy evaluation and risk analysis system, and the implementation effect of the new energy policy can be evaluated in combination with the implementation of subsequent policies, thereby helping to provide a basis for adjusting the policy and providing a basis for the next policy formulation.

[0122] The above method adopts a relatively wide range of evaluation functions for the selection of evaluation functions at each stage, and extracts relevant information according to the selection of appropriate evaluation functions according to different model establishment, thereby improving the accuracy of information extraction. Compared with a single algorithm, the diversity and accuracy of text mining are improved. Compared with the prior art, the present invention has the advantages of accurate data extraction, strong targeting and strong method flexibility.

[0123] The specific implementation is as follows:

[0124] 1) Policy text loading

[0125] After the policy is released, the main platform converts the policy content in the URL (or other address) into txt text (with the encoding format of UTF-8).

[0126] The text mining module loads the policy into the system.

[0127] 2) Basic keyword loading

[0128] Receive the basic keywords transmitted from the main platform (see Table 3), match these keywords with the policy text, and output the keyword status in standard XML format (see Table 4).

[0129] Table 3 Basic keyword list

[0130]

[0131] Table 4 Basic keyword information table

[0132]

[0133] 3) Semantic Web Acquisition

[0134] Get the most keyword-relevant content and filter it with built-in strategies to get the valid content in the "Notice of the National Development and Reform Commission on the Price Policy of Photovoltaic Power Generation Projects in 2018".

[0135]

[0136]

[0137] 4) Text topic generation

[0138] The text mining function uses the topic analysis algorithm and the LDA topic model to analyze the effective content and obtain the matching results of the subject words. The content is as follows:

[0139]

[0140] 5) Text summary generation

[0141] According to the semantic network and subject content, relevant valid content is selected to generate a summary of the text, as follows:

[0142] The Development and Reform Commission, Price Bureau, Energy Bureau, Poverty Alleviation Office of each province, autonomous region, and municipality directly under the central government, the State Grid Corporation of China, the China Southern Power Grid Company, and the Inner Mongolia Electric Power Company will reduce the benchmark on-grid electricity price for photovoltaic power stations put into operation after January 1, 2018. The benchmark on-grid electricity price for Class I, Class II, and Class III resource areas will be adjusted to RMB 0.55, RMB 0.65, and RMB 0.75 per kilowatt-hour (tax included), respectively.

[0143] Starting from 2019, all photovoltaic power generation projects included in the annual scale management of fiscal subsidies will implement the corresponding benchmark electricity price according to the time of commissioning.

[0144] For distributed photovoltaic power generation projects put into operation after January 1, 2018, which adopt the "self-generation for own use and surplus power connected to the grid" model, the full electricity metering subsidy standard will be reduced by 0.05 yuan, that is, the subsidy standard will be adjusted to 0.37 yuan per kilowatt-hour (tax included).

[0145] The benchmark electricity price for village-level photovoltaic poverty alleviation power stations (0.5 MW and below) and the subsidy standard per kilowatt-hour for household distributed photovoltaic poverty alleviation projects remain unchanged.

[0146] 6) Result interface output

[0147] The text mining results are output in standard XML format and provided to the main platform for calling.

[0148] Embodiment 2: Based on the same inventive concept, the present application also proposes a new energy policy information extraction system, the system comprising:

[0149] A processing module, used for processing the new energy policy into a new energy policy text in a set format;

[0150] A division and annotation module is used to divide and annotate the new energy policy text into sentences based on a pre-established new energy policy feature dictionary library and a method of automatically identifying sentences using punctuation marks, output the division and annotation results, and generate a text sentence library;

[0151] The acquisition module is used to generate key information of a new energy policy feature dictionary library and a text sentence library based on the unified service interface and output the key information according to the text sentence.

[0152] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0153] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0154] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0155] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0156] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application rather than to limit its protection scope. Although the present application has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that after reading the present application, those skilled in the art can still make various changes, modifications or equivalent substitutions to the specific implementation methods of the application, and these changes, modifications or equivalent substitutions are all within the scope of the claims to be approved by the application.

Claims

1. A new energy policy information extraction method, characterized in that: The method comprises: Process the new energy policy into a new energy policy text in a set format; Based on a pre-established new energy policy feature dictionary library, the new energy policy text is divided into sentence levels and annotated using a method of automatic sentence recognition using punctuation marks, and the division and annotation results are output to generate a text sentence library; Based on the unified service interface, key information of a new energy policy feature dictionary library and a text sentence library is generated according to the text sentence, and outputted; The establishment of the new energy policy characteristics dictionary library includes: Pre-process the collected new energy policies; Using the expert method, a sample analysis is conducted on the preprocessed new energy policy text database to generate a new energy policy feature dictionary for storing basic feature words; The pre-processing of the collected new energy policies includes: Collect new energy policies, and construct a text database based on new energy policies according to a predefined coding format, as well as a policy data table for managing information related to the new energy policies; The policy data table includes the uploader, upload time, source, number, policy name, issuing department and issuing date; The basic characteristic words include: electricity price, subsidy, finance and taxation, value-added tax refund, income tax exemption, financial subsidy, value-added tax refund, law, bill, action plan, planning, construction plan, full guarantee acquisition, new energy installed capacity, newly added production capacity, cumulative grid-connected installed capacity, annual power generation, installed capacity share, power generation share, CO2 emission reduction, SO2 emission reduction and NOX emission reduction; After obtaining key information, it also includes: Construct a new energy policy text classification strategy library based on the types of key information; including macro laws, development plans, consumption policies, administrative management policies, fiscal and taxation policies, and electricity price policies; According to the sentence weights and the importance of key information, the importance values ​​of the sentences are calculated, and the sentences whose importance values ​​exceed a preset threshold are defined as text summaries.

2. The method according to claim 1, characterized in that The method of automatically identifying sentences based on punctuation marks based on the pre-established new energy policy feature dictionary library is used to divide and annotate the new energy policy text into sentences, output the division and annotation results, and generate a text sentence library, including: Based on the pre-established new energy policy feature dictionary, the new energy policy text is segmented to generate a text segmentation database; According to the word frequency of the text word segmentation library, the new energy policy feature dictionary library is updated; Establish a text sentence library based on the updated new energy policy feature dictionary.

3. The method according to claim 2, characterized in that The method of performing word segmentation processing on the new energy policy text based on the pre-established new energy policy feature dictionary library to generate the text word segmentation library includes: Calling an open source word segmentation tool, loading the basic feature words into a custom word library of the open source word segmentation tool, and using the default word library of the open source word segmentation tool to perform word segmentation on the new energy policy text to obtain the part of speech, word frequency and position of the word segmentation; Using the TF-IDF algorithm to calculate the weight of the basic feature words; A text segmentation library is constructed based on the part of speech, word frequency and position of the segmented words and the weights of the basic feature words.

4. The method according to claim 3, characterized in that The weight of the basic feature word is determined by the following formula: TF = t; IDF = log(N / n); c TF-IDF=TF*IDF=t*log(N / n) In the formula, t is the number of times the basic feature word K appears in the text, N is the total number of new energy policy text documents, and n is the number of documents containing the basic feature word K.

5. The method according to claim 3, characterized in that: According to the frequency of word segmentation in the text word segmentation library, the new energy policy feature dictionary library is updated, including: According to the weight of the basic characteristic words, determining the frequency of occurrence of the basic characteristic words in the new energy policy text; Based on the frequency of occurrence of the basic characteristic words in the new energy policy text, a secondary expert evaluation is conducted on the basic characteristic words not included in the new energy policy characteristic dictionary library, and the basic characteristic words that pass the evaluation are updated to the new energy policy characteristic dictionary library.

6. The method according to claim 2, characterized in that The method of automatically identifying sentences using punctuation marks to divide and annotate the new energy policy text into sentences includes: Use periods, question marks, and exclamation marks as sentence separators; Treat emphasis, hyphens, spaces, book title marks and proper name marks as sentence break marks that are automatically ignored in programming; Modify the initial program using commas, quotation marks, and brackets to complete the sentence.

7. The method according to claim 2, characterized in that The division and labeling results include: sentence identification, sentence content, sentence position and sentence weight, and the positional relationship between participles and sentences; wherein, The sentence weight is the sum of the frequency and weight of non-stop words in the sentence.

8. The method according to claim 1, characterized in that Based on the unified service interface, the text sentence generates key information of the text word library and the text sentence library, and outputs the following information: According to the sentence weights, the sentences to be decomposed are selected, and a key information template is generated based on the Chinese grammatical structure; the participle of the sentence in the key information template is used to generate key information; and the key information is output through a unified service interface.

9. A new energy policy information extraction system, characterized in that: The system comprises: A processing module, used for processing the new energy policy into a new energy policy text in a set format; A division and annotation module is used to divide and annotate the new energy policy text into sentences based on a pre-established new energy policy feature dictionary library and a method of automatically identifying sentences using punctuation marks, output the division and annotation results, and generate a text sentence library; An acquisition module, used to generate key information of a new energy policy feature dictionary library and a text sentence library according to the text sentence based on a unified service interface, and output the information; The establishment of the new energy policy characteristics dictionary library includes: Pre-process the collected new energy policies; Using the expert method, a sample analysis is conducted on the preprocessed new energy policy text database to generate a new energy policy feature dictionary for storing basic feature words; The pre-processing of the collected new energy policies includes: Collect new energy policies, and construct a text database based on new energy policies according to a predefined coding format, as well as a policy data table for managing information related to the new energy policies; The policy data table includes the uploader, upload time, source, number, policy name, issuing department and issuing date; The basic characteristic words include: electricity price, subsidy, finance and taxation, value-added tax refund, income tax exemption, financial subsidy, value-added tax refund, law, bill, action plan, planning, construction plan, full guarantee acquisition, new energy installed capacity, newly added production capacity, cumulative grid-connected installed capacity, annual power generation, installed capacity share, power generation share, CO2 emission reduction, SO2 emission reduction and NOX emission reduction; After obtaining key information, it also includes: Construct a new energy policy text classification strategy library based on the types of key information; including macro laws, development plans, consumption policies, administrative management policies, fiscal and taxation policies, and electricity price policies; According to the sentence weights and the importance of key information, the importance values ​​of the sentences are calculated, and the sentences whose importance values ​​exceed a preset threshold are defined as text summaries.

Images