An evaluation method and system for carbon emission policies in transportation

Abstract

This invention discloses a method and system for evaluating carbon emission policies in the transportation sector. The method includes the following steps: acquiring policy documents related to carbon emissions in the transportation sector, performing word segmentation and word frequency analysis on them, constructing an industry lexicon based on the word segmentation results, extracting keywords from the industry lexicon, and updating the word segmenter using high-frequency words and keywords from the word segmentation; constructing policy evaluation indicators based on the GDP and carbon emissions of the policy documents for the corresponding year, classifying policy documents into policies with significant or insignificant effects based on the comparison between their values ​​and set thresholds, and labeling the policy documents according to the classification results; constructing a prediction model, embedding words into the prediction model based on the industry lexicon, and training the word-embedded prediction model using the labeled policy documents; segmenting the policy documents to be evaluated using the updated word segmenter, inputting the word segmentation results into the trained prediction model, and determining whether the effects of the input policy documents are significant.

Description

Technical Field

[0001] This invention relates to the field of transportation technology, and specifically to an assessment method and system for transportation carbon emission policies. Background Technology

[0002] Globally, climate change and environmental problems are becoming increasingly severe. Reducing greenhouse gas emissions, primarily carbon dioxide, has become a major issue that humanity urgently needs to address, and this has reached a global consensus. Against this backdrop, comprehensive decision-making data is urgently needed to support energy conservation and emission reduction efforts and drive industries towards a green and low-carbon transformation.

[0003] Currently, in policy analysis, traditional research mainly relies on manual, word-by-word interpretation of policy texts, using the experience and knowledge of professionals to determine the key points and potential impacts of policies. In the data processing stage, the common practice is to use general-purpose text analysis software, which uses basic word segmentation algorithms and simple word frequency statistics to perform preliminary text processing. As for model building, most directly adopt general-purpose machine learning models, such as some traditional classification models, inputting the simply processed data for training and prediction.

[0004] However, these existing technological methods have serious shortcomings when dealing with policy analysis and effectiveness evaluation in the transportation sector. While manual interpretation possesses a certain degree of professionalism, it is extremely inefficient and lacks unified standards, making it difficult to process large volumes of transportation environment policy texts on a large scale. General-purpose text analysis software is not optimized for transportation-specific terminology and language expressions, failing to accurately identify industry keywords. This results in the loss of important information during word segmentation and frequency statistics, failing to provide high-quality data support for subsequent analysis. Similarly, general-purpose machine learning models do not consider the unique characteristics of the transportation industry. During feature extraction and model training, they fail to effectively capture industry patterns and semantic logic in transportation environment policies, leading to significant discrepancies between predictions and actual results. This makes it difficult to provide decision-makers with targeted and reliable policy recommendations, and ultimately fails to meet the actual needs of the transportation sector for policy effectiveness evaluation. Summary of the Invention

[0005] This invention proposes an evaluation method and system for carbon emission policies in transportation, which solves the problem that existing policy analysis methods are not specific enough in the field of carbon emission in transportation.

[0006] To address the aforementioned technical problems, this invention provides a method for assessing carbon emission policies in transportation, comprising the following steps:

[0007] Step S1: Obtain several policy documents on carbon emissions in transportation, perform word segmentation and word frequency analysis on the text content of all policy documents, construct an industry lexicon based on the word segmentation results, extract keywords from the industry lexicon, and update the word segmenter using high-frequency words and keywords from the word segmentation.

[0008] Step S2: Construct policy evaluation indicators based on the relationship between GDP and carbon emissions in the corresponding year of the policy document. Based on the comparison between the values ​​of the policy evaluation indicators and the set thresholds, classify the policy documents into policies with significant effects or policies with insignificant effects, and label the policy documents according to the classification results.

[0009] Step S3: Construct a prediction model, embed words into the prediction model based on the industry vocabulary, and train the word-embedded prediction model using the annotated policy documents;

[0010] Step S4: Use the updated word segmenter to segment the policy document to be evaluated, input the segmentation results into the trained prediction model, and judge whether the effect of the input policy document is obvious.

[0011] Preferably, the keyword extraction from the industry glossary in step S1 includes the following steps:

[0012] Step S11: Calculate the term frequency (TF) and inverse document frequency (IDF) of each word in the industry lexicon. Calculate the first importance score of the word segment based on the TF and IDF. The expression for calculating the first importance score is as follows:

[0013] TF-IDF(t,d)=TF(t,d)×IDF(t);

[0014]

[0015] In the above formula, TF-IDF(t,d) is the first importance score of the word segment; TF(t,d) is the term frequency of the word segment; IDF(t) is the inverse document frequency of the word segment; f t,d t represents the number of times the word segment t appears in the environmental policy text d; D represents the total number of word segments; N represents the total number of environmental policy texts.

[0016] Step S12: Based on the word segmentation network constructed according to the relationships of all word segments, calculate the weight of each word segment in the word segmentation network as the second importance score of the word segmentation. The expression for calculating the second importance score is as follows:

[0017]

[0018] In the formula, S(t) is the second importance score of word segmentation; ξ is the damping coefficient; ln(t) i ) refers to the segmentation t iThe vocabulary set; out(t) j ) is the word segmentation t in the word segmentation network j The degree of departure;

[0019] Step S13: Weight and fuse the first importance score and the second importance score to obtain the importance score for each word segment. The expression for calculating the importance score is as follows:

[0020] Score(t)=α×TF-IDF(t)+(1-α)×S(t);

[0021] In the formula, α is the weight adjustment parameter;

[0022] Step S14: Sort all word segments in descending order of importance score, and select the top N word segments as keywords.

[0023] Preferably, the expression for the policy evaluation indicators in step S2 is:

[0024]

[0025] In the above formula, DPR is the carbon emission efficiency ratio; PEIR China To improve the effectiveness of domestic policies; PEIR Global To improve the global policy effectiveness rate; ratio China,i The ratio is the ratio of China's GDP to carbon emissions in year i. Global,i Let be the ratio of global GDP to carbon emissions in year i; GDP China,i CE China,i These represent the country's gross domestic product and carbon emissions, respectively. Global,i CE Global,i These represent global GDP and carbon emissions, respectively.

[0026] Preferably, the word embedding of the prediction model based on the industry vocabulary in step S3 includes the following steps:

[0027] Step S31: Define the window size. For each word in the industry glossary, count the co-occurrence frequency of the word with other words in the window to obtain the co-occurrence probability of all word pairs.

[0028] Step S32: Randomly sample the central word from the industry vocabulary and obtain the context word pairs of the central word. Use the prediction model to obtain the probability distribution of the predicted context words of the sampled central word.

[0029] Step S33: Based on the predicted context word probability distribution and the actual context words, calculate the loss function of the prediction model, update the parameters of the prediction model through the optimizer, obtain the word vector of each word, and form an industry embedding matrix;

[0030] Step S34: Introduce an industry term embedding matrix into the embedding layer of the prediction model using a splicing method.

[0031] Preferably, the expression for calculating the co-occurrence probability of the word pair in step S31 is:

[0032]

[0033] In the formula, p(w) j |w i ) is the word w i And the word w j The co-occurrence probability of M; ij For the word w i And the word w j The number of co-occurrences; V industry This is a glossary of industry terms.

[0034] Preferably, the expression for calculating the loss function in step S33 is:

[0035]

[0036] In the formula, p(w) j |w i ) is the word w i And the word w j Co-occurrence probability; Context(w i ) for the word w i Analysis of the context; V industry This is a glossary of industry terms.

[0037] Preferably, step S1 involves obtaining historical environmental policy texts in the transportation sector through a proxy pool and web crawler, including the following steps:

[0038] Step S101: Collect several proxy IPs, verify the validity of the collected proxy IPs, store the verified proxy IPs in the database, and build a proxy pool;

[0039] Step S102: Randomly assign proxy IPs from the proxy pool, and use web crawler software to crawl a large amount of environmental policy texts in the transportation field from public websites using the proxy IPs;

[0040] Step S103: Periodically verify and update the proxy IPs in the proxy pool.

[0041] This invention also provides an evaluation system for carbon emission policies in transportation, based on the aforementioned evaluation method for carbon emission policies in transportation, comprising: a data acquisition module, a text preprocessing module, a policy effectiveness evaluation module, a model training module, and a text prediction module;

[0042] The data acquisition module collects textual data such as regulations, policies, and academic papers related to carbon emissions in transportation.

[0043] The text preprocessing module: constructs a stop word list, cleans the collected text data, segments the cleaned text data into words and counts word frequencies, extracts keywords from all word segments, and updates the user dictionary of the word segmenter using the top 100 high-frequency words and keywords.

[0044] The policy effectiveness evaluation module: constructs evaluation indicators for assessing the effectiveness of transportation carbon emission policies, divides policies into two categories: those with significant effects and those with insignificant effects, and annotates the policy text according to the policy effects;

[0045] The model training module: constructs a prediction model, adds word embeddings based on an industry lexicon to the embedding layer of the prediction model, and trains the prediction model using two types of labeled policy text data;

[0046] The text prediction module preprocesses the transportation carbon emission policy text to be evaluated, inputs the processed policy text into the trained prediction model, and outputs the prediction result of whether the policy effect is obvious.

[0047] Preferably, the system further includes a user interaction module, which includes two pages. The first page displays the latest transportation and carbon emission related policies and regulations, and the second page accepts policy text input by the user. After the background processes the article, the second page displays a word cloud and semantic network analysis diagram of the policy text input by the user, so as to help the user understand the content and semantic relationships of the policy text in an intuitive way.

[0048] Preferably, the data acquisition module constructs a proxy pool, crawls text data related to carbon emissions in transportation from public websites based on the constructed proxy pool, and regularly updates the proxy IPs in the proxy pool.

[0049] The advantages of this invention include at least the following:

[0050] 1. By specifically acquiring historical environmental policy texts in the transportation field and constructing a stop word list for this field, the text can be accurately preprocessed. In the process of word segmentation, word frequency statistics and keyword extraction, the focus is on the core content of transportation environmental policies, effectively filtering out key information, providing a high-quality data foundation for subsequent analysis and model training, ensuring that the data is closely integrated with actual application scenarios, and avoiding interference from irrelevant information.

[0051] 2. Constructing an industry-specific glossary and embedding words enhances the predictive model's understanding and processing capabilities of specialized terms in the transportation sector. Training with two types of environmental policy texts allows the model to fully learn the textual features and semantic patterns of policies with different effects, continuously optimize its parameters, improve the accuracy and reliability of predictions, and better adapt to the task requirements of predicting the effects of transportation environmental policies. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of the proxy pool construction process according to an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram of the proxy pool framework according to an embodiment of the present invention;

[0055] Figure 4 This is a schematic diagram of the word frequency analysis process according to an embodiment of the present invention;

[0056] Figure 5 This is a schematic diagram of the word meaning network analysis process according to an embodiment of the present invention;

[0057] Figure 6 This is a schematic diagram of the text prediction process according to an embodiment of the present invention;

[0058] Figure 7 This is a schematic diagram of the system framework of an embodiment of the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0060] like Figure 1 As shown, this embodiment of the invention provides a method for evaluating carbon emission policies in transportation, including the following steps:

[0061] Step S1: Obtain several policy documents on carbon emissions in transportation through a proxy pool and web crawler. Perform word segmentation and word frequency analysis on the text content of all policy documents. Construct an industry lexicon based on the word segmentation results. Extract keywords from the industry lexicon. Update the word segmenter using high-frequency words and keywords from the word segmentation.

[0062] Specifically, embodiments of the present invention use a Redis proxy pool and Python software to crawl a large number of policy documents in the field of transportation from public websites.

[0063] like Figure 2 As shown, building a Redis proxy pool includes the following steps:

[0064] Step S101: Collect several proxy IPs and verify their validity, specifically verifying their connectivity, anonymity, speed, and stability.

[0065] Step S102: Store the verified proxy IPs in the database to build a Redis proxy pool.

[0066] Step S103: Randomly or strategically assign proxy IPs from the Redis proxy pool, and use web crawling software to crawl a large amount of environmental policy texts in the transportation field from public websites using the proxy IPs.

[0067] Step S104: Periodically verify and update the proxy IPs in the Redis proxy pool.

[0068] like Figure 3 The diagram shows the structure of the proxy pool constructed according to an embodiment of the present invention. After constructing the Redis proxy pool, Python is used to crawl a large number of legal, policy, and standard documents related to carbon emissions in transportation from public websites using proxy IPs.

[0069] The collected text data is parsed and saved as a .txt file for later processing. The collected text data is cleaned to remove special characters, punctuation marks, numbers, and HTML tags. Simultaneously, the text format is standardized, and blank lines and extra spaces are filtered out to optimize text quality.

[0070] Construct a dedicated stop word list for the field of transportation policy, removing words that lack distinctiveness. This list includes common words such as "is," "of," and "and," as well as words from unrelated fields such as "agriculture," "construction," "manufacturing," and "electricity."

[0071] like Figure 4 The diagram illustrates the word frequency analysis process of this invention. The cleaned text is segmented using the Jieba library in Python, generating a segmentation result list. All segmentations are checked to determine if they are in the stop word character list and to check if their length is greater than 1. Segments in the stop word list or with a length less than 1 are filtered out. The filtered segmentation list is converted into a long text string, and a word cloud image is generated using functions from the WordCloud library to display the word segments in historical environmental policy texts to the user.

[0072] The word frequency in the segmentation results is statistically analyzed, and the `collections.Counter` function is used to identify high-frequency words. After analyzing the high-frequency words in the policy text, it is also necessary to identify the keywords. Therefore, this embodiment of the invention innovatively combines the TF-IDF and TextRank algorithms to extract keywords from the segmentation results. The TF-IDF algorithm is a statistical method for evaluating the importance of words in text. It measures the relative importance of words in a specific document by calculating the term frequency (TF) and inverse document frequency (IDF), thereby effectively filtering out common words and highlighting characteristic words. The TextRank algorithm is an unsupervised keyword extraction algorithm based on a graph model. It treats words in the text as nodes in a graph, establishes edges based on word co-occurrence relationships, and uses the PageRank algorithm to calculate the weight of each word, thereby extracting the most representative keywords. This embodiment of the invention combines the two algorithms to extract keywords from the segmentation results, including the following steps:

[0073] Step S111: Calculate the term frequency (TF) and inverse document frequency (IDF) of each word segment. Calculate the first importance score of the word segment based on the TF and IDF. The expression for calculating the first importance score is:

[0074] TF-IDF(t,d)=TF(t,d)×IDF(t);

[0075]

[0076] In the above formula, TF-IDF(t,d) is the first importance score of the word segment; TF(t,d) is the term frequency of the word segment; IDF(t) is the inverse document frequency of the word segment; f t,d t represents the number of times the word segment t appears in the environmental policy text d; D represents the total number of word segments; and N represents the total number of environmental policy texts.

[0077] Step S12: Based on the word segmentation network constructed according to the relationships of all word segments, calculate the weight of each word segment in the word segmentation network as the second importance score of the word segmentation. The expression for calculating the second importance score is as follows:

[0078]

[0079] In the formula, S(t) is the second importance score of word segmentation; ξ is the damping coefficient, which is taken as 0.85 in this embodiment of the invention; ln(t) i ) refers to the segmentation t i The vocabulary set; out(t) j ) is the word segmentation t in the word segmentation network j The degree of departure.

[0080] Step S13: Weight and fuse the first importance score and the second importance score to obtain the importance score for each word segment. The expression for calculating the importance score is as follows:

[0081] Score(t)=α×TF-IDF(t)+(1-α)×S(t);

[0082] In the formula, α is a weight adjustment parameter, which is 0.5 in this embodiment of the invention.

[0083] Step S14: Sort all word segments in descending order of importance score, and select the top N word segments as keywords.

[0084] The top 100 high-frequency words and extracted keywords are automatically added to a custom word library file, and the user dictionary of the Jieba word segmenter is dynamically updated to achieve adaptive optimization of the word segmentation library. The generated word segmentation model and word frequency data are persistently stored using the pickle library for later use.

[0085] like Figure 5 The diagram illustrates the flow chart of word semantic network analysis in this embodiment of the invention. By calling the word segmentation function in the Jieba library, all environmental policy texts are segmented. The process checks if all segmented words are in the stop word character list and if the length of each segmented word is greater than 1, filtering out words that do not meet these two conditions. The word frequency of all filtered segmented words is calculated, and the top N words with the highest frequency are selected as keywords. Keywords are used as nodes, and the values ​​in the word segmentation association matrix are used as edge weights. If two keywords appear in the same sentence, the value at the corresponding position in the association matrix is ​​increased by a set value. A graph structure is constructed using the Networkx library, generating a PNG file. The latest transportation carbon emission-related policies and regulations are displayed through a PyWebIO-based web application; the Python libraries used include requests, json, scrapy, time, jieba, redis, BeautifulSoup, and threading.

[0086] Step S2: Construct policy evaluation indicators based on the relationship between GDP and carbon emissions in the corresponding year of the policy document. Based on the comparison between the values ​​of the policy evaluation indicators and the set thresholds, classify the policy documents into policies with significant effects or policies with insignificant effects, and label the policy documents according to the classification results.

[0087] Specifically, constructing an evaluation indicator system to determine whether the effects of environmental policies are significant includes the following steps:

[0088] Step S21: Use a web crawler to retrieve all articles from the policy and regulation section of the carbon trading website, saving the publication date of each article. The metadata for each article includes policy content, issuing organization, policy category, and publication time. Calculate the domestic and global policy effectiveness improvement rate (PEIR) based on the carbon emissions and GDP corresponding to the policy text for that year.

[0089]

[0090] In the formula, PEIR China To improve the effectiveness of domestic policies; PEIR Global To improve the global policy effectiveness rate; ratio China,i The ratio is the ratio of China's GDP to carbon emissions in year i. Global,i Let be the ratio of global GDP to carbon emissions in year i; GDP China,i CE China,i These represent the country's gross domestic product and carbon emissions, respectively. Global,i CE Global,i These represent global GDP and carbon emissions, respectively.

[0091] Step S22: Calculate the domestic and global carbon emission efficiency ratio (DPR) based on PEIR:

[0092]

[0093] Step S23: When the DPR is greater than 20%, the carbon emission policy of the previous year is considered to have a significant effect; otherwise, it is considered insignificant. Organize the two types of policy texts into two separate JSON files. The structure of each file is: {"Policy Content": "Specific Text", "Effect Classification": "Significant / Insignificant", "Publication Date": "YYYY-MM-DD"}, to facilitate subsequent calling and classification by the BERT model.

[0094] Step S3: Build a prediction model, embed words into the prediction model based on the industry vocabulary, and train the word-embedded prediction model using annotated policy documents.

[0095] Specifically, the prediction model is constructed by calling the Torch, Transformers, and Sklearn libraries and defining the SingleInputBERT class. In this embodiment of the invention, the BERT model is used. Considering the diversity of policy texts, in order to further improve the classification accuracy of the BERT model for long texts, this embodiment of the invention adds pre-trained word embeddings based on industry vocabularies to the BERT embedding layer to enhance the model's ability to recognize professional terms such as transportation and carbon emissions.

[0096] The Word2Vec method is used to generate embedding vectors, and the BERT model is pre-trained using an industry vocabulary for word embedding, including the following steps:

[0097] Step S31: Collect a large amount of domain text data containing industry-specific vocabulary from industry documents such as transportation carbon emission regulations, policies, and related academic papers. Preprocess the corpus by removing stop words, punctuation marks, and non-linguistic symbols, and use jieba for text segmentation to obtain the industry vocabulary list V. industry .

[0098] Step S32: Define the window size. For each word in the industry glossary, count the co-occurrence frequency of the word with other words in the window to obtain the co-occurrence probability of all word pairs.

[0099]

[0100] In the formula, p(w) j |w i ) is the word w i And the word w j The co-occurrence probability of M; ij For the word w i And the word w j The number of times they co-occur.

[0101] Step S33: Using the Skip-gram model, randomly sample the center words from the industry vocabulary and obtain the context word pairs of the center words. Then, use the prediction model to obtain the probability distribution of the predicted context words of the sampled center words.

[0102] Step S34: Calculate the loss function of the prediction model based on the predicted probability distribution of context words and the actual context words:

[0103]

[0104] In the formula, L is the loss function; Context(w i ) for the word w i Analysis of the context.

[0105] Step S35: Update the parameters of the prediction model through the optimizer to obtain the word vector of each word, form an industry embedding matrix, and introduce the industry word embedding matrix into the embedding layer of the prediction model by concatenation.

[0106] Step S4: Use the updated word segmenter to segment the policy document to be evaluated, input the segmentation results into the trained prediction model, and judge whether the effect of the input policy document is obvious.

[0107] like Figure 6The diagram illustrates the text prediction process of this invention. The trained word segmenter and BERT model are loaded, and the Jieba library is used to segment the user-input policy text, removing specific characters and markers such as full-width spaces, line breaks, and years to form a new text. The preprocessed text is then input into the BERT model to predict the effectiveness of the policy text, determining whether the policy has a relatively significant or relatively insignificant effect.

[0108] Simultaneously, singular value decomposition can be performed on the word frequency matrix of the policy text to obtain the singular value of each word and the singular vector corresponding to each text, thereby constructing a keyword-keyword semantic distance table and a text-keyword semantic distance table to display the key content of the policy text to users.

[0109] like Figure 7 As shown, this embodiment of the invention also provides an evaluation system for transportation carbon emission policies, which is based on the above-mentioned evaluation method for transportation carbon emission policies. The system includes: a data acquisition module, a text preprocessing module, a policy effect evaluation module, a model training module, a text prediction module, and a user interaction module.

[0110] The data acquisition module is used to build a proxy pool. Based on the built proxy pool, it crawls text data related to carbon emissions in transportation from public websites and updates the proxy IPs in the proxy pool regularly.

[0111] The text preprocessing module is used to build a stop word list, clean the collected text data, segment the cleaned text data and count word frequencies, extract keywords from all segmented words, and update the user dictionary of the word segmenter using the top 100 high-frequency words and keywords.

[0112] The policy effectiveness evaluation module constructs evaluation indicators to assess the effectiveness of carbon emission policies in transportation, classifies policies into two categories: those with significant effects and those with insignificant effects, and annotates policy texts according to their effects.

[0113] The model training module is used to build a prediction model. Word embeddings based on an industry vocabulary are added to the embedding layer of the prediction model, and the prediction model is trained using two types of labeled policy text data.

[0114] The text prediction module is used to preprocess the transportation carbon emission policy text to be evaluated. The processed policy text is then input into the trained prediction model, which outputs a prediction of whether the policy's effect is significant.

[0115] The user interaction module includes two pages. The first page displays the latest transportation and carbon emission related policies and regulations, while the second page accepts policy text input by the user. After the backend processes the text, it displays a word cloud and semantic network analysis diagram of the policy text input by the user on the second page, which helps users understand the content and semantic relationships of the policy text in an intuitive way.

[0116] This invention's system integrates Python web scraping technology and a Redis proxy pool. Utilizing a Redis-designed proxy IP pool, it achieves efficient and stable management of proxy IPs. Through the collection, verification, storage, allocation, and updating of proxy IPs, it ensures the smooth execution of Python web scraping tasks. In practical applications in transportation, the Redis data structure and operation methods can be adjusted according to needs to adapt to different carbon emission monitoring scenarios. This enables real-time monitoring and data collection of transportation carbon emission-related policies and regulations. Compared to traditional periodic data collection methods, this system provides more timely and accurate data support, offering decision-makers the latest policy dynamics and carbon emission information. A user-friendly web application interface based on the PyWebIO library allows even non-professional users to easily perform policy text analysis and prediction. This design lowers the barrier to entry and expands the potential user base. Furthermore, the system allows for easy integration of new data sources and analysis tools, exhibiting excellent scalability. As the field of transportation carbon emissions continues to develop, this system can flexibly adapt to new needs and challenges, maintaining long-term technological competitiveness.

[0117] The method provided by this invention not only performs basic word frequency analysis on policy texts, but also utilizes advanced text analysis techniques such as jieba word segmentation and the BERT model to deeply mine the semantic information of policy texts, generating word clouds and semantic network diagrams. This enables policymakers to understand policy content and potential impacts more intuitively and deeply, improving the depth and breadth of policy analysis. By introducing a text prediction function based on a pre-trained BERT model, it is possible to predict policy effects and determine whether a policy has relatively significant or insignificant effects. This prediction mechanism provides policymakers with early warning and response strategies for crisis transformation, improving the system's practicality and the foresight of decision-making.

[0118] The method of this invention applies data acquisition, data analysis, and data prediction methods, is user-friendly, comprehensively utilizes relevant artificial intelligence methods, and has a smooth text processing process. It provides new ideas and reference suggestions for the evaluation of transportation carbon emission policies, reduces the difficulty of decision-making, and has strong feasibility.

[0119] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described; only preferred embodiments of the present invention are illustrated. The descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. As long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0120] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this invention should be determined by the appended claims.

Claims

1. A method for assessing carbon emission policies in transportation, characterized in that, Includes the following steps: Step S1: Obtain several policy documents on carbon emissions in transportation, perform word segmentation and word frequency analysis on the text content of all policy documents, construct an industry lexicon based on the word segmentation results, extract keywords from the industry lexicon, and update the word segmenter using high-frequency words and keywords from the word segmentation. Step S2: Construct policy evaluation indicators based on the relationship between GDP and carbon emissions for the corresponding year of the policy document. Based on the comparison between the values ​​of these indicators and set thresholds, classify the policy documents into policies with significant effects or those with insignificant effects. Label the policy documents according to the classification results. The expression for the policy evaluation indicators is: In the above formula, DPR is the carbon emission efficiency ratio; PEIR China To improve the effectiveness of domestic policies; PEIR Global To improve the global policy effectiveness rate; ratio China,i The ratio is the ratio of China's GDP to carbon emissions in year i. Global,i Let be the ratio of global GDP to carbon emissions in year i; GDP China,i CE China,i These represent the country's gross domestic product and carbon emissions, respectively. Global,i CE Global,i These are global GDP and carbon emissions, respectively. Step S3: Construct a prediction model, embed words into the prediction model based on the industry vocabulary, and train the word-embedded prediction model using the annotated policy documents; Step S4: Use the updated word segmenter to segment the policy document to be evaluated, input the segmentation results into the trained prediction model, and judge whether the effect of the input policy document is obvious.

2. The method for evaluating a transportation carbon emission policy according to claim 1, characterized in that: Step S1, which involves extracting keywords from the industry glossary, includes the following steps: Step S11: Calculate the term frequency (TF) and inverse document frequency (IDF) of each word in the industry lexicon. Calculate the first importance score of the word segment based on the TF and IDF. The expression for calculating the first importance score is as follows: TF-IDF(t,d)=TF(t,d)×IDF(t); In the above formula, TF-IDF(t,d) is the first importance score of the word segment; TF(t,d) is the term frequency of the word segment; IDF(t) is the inverse document frequency of the word segment; f t,d t represents the number of times the word segment t appears in the environmental policy text d; D represents the total number of word segments; N represents the total number of environmental policy texts. Step S12: Based on the word segmentation network constructed according to the relationships of all word segments, calculate the weight of each word segment in the word segmentation network as the second importance score of the word segmentation. The expression for calculating the second importance score is as follows: In the formula, S(t) is the second importance score of word segmentation; ξ is the damping coefficient; ln(t) i ) refers to the segmentation t i The vocabulary set; out(t) j ) is the word segmentation t in the word segmentation network j The degree of departure; Step S13: Weight and fuse the first importance score and the second importance score to obtain the importance score of each word segment. The expression for calculating the importance score is as follows: Score(t)=α×TF-IDF(t)+(1-α)×S(t); In the formula, α is the weight adjustment parameter; Step S14: Sort all word segments in descending order of importance score, and select the top N word segments as keywords.

3. The method for evaluating a transportation carbon emission policy according to claim 1, characterized in that: Step S3, which involves embedding words into the prediction model based on the industry vocabulary, includes the following steps: Step S31: Define the window size. For each word in the industry glossary, count the co-occurrence frequency of the word with other words in the window to obtain the co-occurrence probability of all word pairs. Step S32: Randomly sample the central word from the industry vocabulary list and obtain the context word pairs of the central word. Use the prediction model to obtain the probability distribution of the predicted context words of the sampled central word. Step S33: Based on the predicted context word probability distribution and the actual context words, calculate the loss function of the prediction model, update the parameters of the prediction model through the optimizer, obtain the word vector of each word, and form an industry embedding matrix; Step S34: Introduce an industry term embedding matrix into the embedding layer of the prediction model using a splicing method.

4. The method for evaluating a transportation carbon emission policy according to claim 3, characterized in that: The expression for calculating the co-occurrence probability of the word pair in step S31 is as follows: In the formula, p(w) j |w i ) for the word w i And the word w j The co-occurrence probability of M; ij For the word w i And the word w j The number of co-occurrences; V industry This is a glossary of industry terms.

5. The method for evaluating a transportation carbon emission policy according to claim 3, characterized in that: The expression for calculating the loss function in step S33 is as follows: In the formula, L is the loss function; p(w j |w i ) for the word w i And the word w j Co-occurrence probability; Context(w i ) for the word w i Analysis of the context; V industry This is a glossary of industry terms.

6. The method for evaluating a carbon emission policy for transportation according to claim 1, characterized in that: Step S1 involves obtaining several policy documents on carbon emissions from transportation through a proxy pool and web crawlers, including the following steps: Step S101: Collect several proxy IPs, verify the validity of the collected proxy IPs, store the verified proxy IPs in the database, and build a proxy pool; Step S102: Randomly assign proxy IPs from the proxy pool, and use web crawler software to crawl a large amount of environmental policy texts in the transportation field from public websites using the proxy IPs; Step S103: Periodically verify and update the proxy IPs in the proxy pool.

7. An assessment system for carbon emission policies in transportation, implemented based on an assessment method for carbon emission policies in transportation as described in any one of claims 1 to 6, characterized in that, include: The system includes a data acquisition module, a text preprocessing module, a policy effectiveness evaluation module, a model training module, and a text prediction module. The data acquisition module collects textual content such as regulations, policies, and academic papers related to carbon emissions in transportation. The text preprocessing module: constructs a stop word list, cleans the collected text content, segments the cleaned text data into words and counts word frequencies, extracts keywords from all word segments, and updates the user dictionary of the word segmenter using the top 100 high-frequency words and keywords. The policy effectiveness evaluation module: constructs evaluation indicators for assessing the effectiveness of transportation carbon emission policies, divides policies into two categories: those with significant effects and those with insignificant effects, and annotates the policy text according to the policy effects; The model training module: constructs a prediction model, adds word embeddings based on an industry lexicon to the embedding layer of the prediction model, and trains the prediction model using two types of labeled policy text data; The text prediction module preprocesses the transportation carbon emission policy text to be evaluated, inputs the processed policy text into the trained prediction model, and outputs the prediction result of whether the policy effect is obvious.

8. The assessment system for carbon emission policies in transportation according to claim 7, characterized in that: The system also includes a user interaction module, which consists of two pages. The first page displays the latest transportation and carbon emission related policies and regulations, while the second page accepts policy text input by the user. After the background processes the text, the second page displays a word cloud and semantic network analysis diagram of the policy text input by the user, in an intuitive way to help users understand the content and semantic relationships of the policy text.

9. The assessment system for carbon emission policies in transportation according to claim 7, characterized in that: The data acquisition module constructs a proxy pool, and based on the constructed proxy pool, crawls text data related to carbon emissions in transportation from public websites, and regularly updates the proxy IPs in the proxy pool.

Images