Scientific and technological policy analysis method

Abstract

The invention discloses a science and technology policy analysis method. The method comprises the following steps: 1, collecting scientific policy documents; 2, classifying the collected scientific policy documents, and completing feature extraction by utilizing a technology of grouping and classifying fragments, paragraphs or documents; 3, establishing a 3 + 5 policy topic statement library, a text abstract library, an entity library and a relation library; and 4, constructing a model, and dividing the database obtained in the step 3 into a training set and a test set for training and testing to obtain a science and technology policy file analysis module capable of completing analysis on the science and technology policy file. According to the method, contents in the aspects of three blueprints and five construction drawings are selected as national science and technology policy analysis elements, so that effects of various policies can be comprehensively evaluated, policy making is optimized, and an analysis basis is provided for boosting and promoting science and technology innovation and social and economic development in China; and the scientific policy understanding and analysis efficiency is improved, and correct understanding of the scientific policy can be ensured.

Description

Technical Field

[0001] This invention belongs to the field of policy analysis, specifically a method for analyzing science and technology policies. Background Technology

[0002] As my country's science and technology scale grows larger, its functions become more powerful, and the process of scientific research organization and management becomes increasingly complex, the sheer number of current science and technology policies means that finding and analyzing relevant documents can be time-consuming. Sometimes, errors in personal understanding can lead to problems in the analysis of science policies, thus affecting subsequent operations. Summary of the Invention

[0003] To reflect the achievements of policies that are centered on innovation, guided by national strategic needs and the forefront of global science and technology, deepen reforms, optimize the environment, and stimulate the innovation vitality of scientific and technological talents, this invention provides a science and technology policy analysis method to address the shortcomings of existing technologies.

[0004] This invention is achieved through the following technical solution: A method for analyzing science and technology policies includes the following steps: Step 1: Collect science policy documents; Step 2: Classify the collected science policy documents, and use techniques for grouping and classifying fragments, paragraphs or documents to complete feature extraction; Step 3: Establish a "3+5" policy theme statement database, text summary database, entity database, and relation database; Step 4: Build a model and use the database obtained in Step 3 to train and test it, resulting in a science and technology policy document analysis module capable of analyzing science and technology policy documents.

[0005] In the science and technology policy analysis method described above, step two involves text preprocessing before classifying science policy documents to remove stop words, punctuation marks, special characters, and case conversion.

[0006] In the above-described method for analyzing science and technology policies, the feature extraction operation in step two is as follows: the text is converted into a word frequency vector and then weighted using TF-IDF (word frequency-inverse document frequency); each word is converted into a high-dimensional vector using tools such as Word2Vec and GloVe, and the results are compared; and the features are filtered according to their importance or relevance to reduce noise and overfitting.

[0007] As described above, in the science and technology policy analysis method, the "3+5" policy theme statement library in step three includes the following: The text contains statements related to the theme of "science and technology policy and strategy". "Science and technology policy and strategy" is a general guideline for solving the overall problems between the development of science and technology and the development of the national economy and society, as well as the relationship between various goals. It is the most concentrated and macro-level manifestation of the science and technology policy system. The text contains sentences related to the theme of "science and technology planning and layout". "Science and technology planning and layout" refers to a long-term plan that is implemented in stages. It is formulated under the guidance of science and technology development policies and strategies. Science and technology planning and layout makes specific provisions and designs for the approaches, measures and steps of science and technology activities. The text containing statements related to the theme of "institutional arrangements for science and technology" states that "science and technology planning and layout" refers to the organizational system and institutional design of science and technology activities; The text contains statements related to the theme of "resource allocation policy." "Resource allocation policy" refers to the allocation of resources to various entities and uses according to certain rules and mechanisms, so as to create as much benefit as possible from limited resources. The text contains statements related to the theme of "science and technology talent policy". "Science and technology talent policy" refers to the code of conduct formulated to regulate the behavior of science and technology talents, which includes a series of laws, measures, methods, and regulations concerning the cultivation, attraction, mobility, incentives and management of science and technology talents. The text of statements related to the theme of "innovation subject policy" refers to a series of laws, measures, methods, and regulations concerning the national innovation system, including the government (leading role), research universities (basic and emerging role), national research institutions (basic and strategic research), and enterprise R&D organizations (the main body of technological innovation). The text of statements related to the theme of "technology transfer policy" refers to the sum of various direct and indirect means, regulations and methods for promoting the transfer of scientific and technological achievements, regulating the transfer process and the behavior of transfer entities; The text contains statements related to the theme of "environmental creation policy." "Environmental creation policy" refers to the environmental factors that the government indirectly influences through policies such as regulations, review and approval, and intellectual property protection, thereby affecting scientific and technological development.

[0008] In the science and technology policy analysis method described above, the model in step four is a Naive Bayes model.

[0009] The above-described method for analyzing science and technology policies, using the Naive Bayes model, employs the Scikit-Learn Naive Bayes algorithm for text classification. It estimates the class probability of the text using the joint probability of feature terms and the class. The specific steps include: Step 1: Import the word vector Bunch object after word segmentation; Step 2: Construct the TF-IDF vector space for the test set; Step 3: Import the bag-of-words from the training set; Step 4: Initialize the vector space model using TFIDFVectorizer; Step 5: Create persistence for the bag of words.

[0010] As described above, in the science and technology policy analysis method, step four, which involves training the model using a training set, includes the following steps: Step 1: Import the Multinomial Bayes algorithm package; Step 2: Import the training set vector space; Step 3: Import the test set vector space; Step 4: Apply the Naive Bayes algorithm; Step 5: Predict the classification result; Step 6: Display the prediction accuracy.

[0011] In the science and technology policy analysis method described above, the distance between the test set samples and the training set samples is recorded using Euclidean distance. The k samples with the shortest distances are selected as the neighbors of the predicted samples. The category with the most frequent distance among the k neighbors is selected as the predicted value of the test set sample through a voting mechanism.

[0012] In the science and technology policy analysis method described above, step four of the Naive Bayes model uses a test set to evaluate the model's performance. Commonly used evaluation metrics include accuracy, recall, and F1 score. At the same time, cross-validation is used to further evaluate the model's stability and generalization ability.

[0013] In the science and technology policy analysis method described above, the model evaluation process involves adjusting feature engineering or model parameters based on the evaluation results to optimize model performance.

[0014] The advantages of this invention are: This invention selects three "blueprints" (strategic layout) and five "construction plans" (policy guidelines) as elements for analyzing science and technology policies of major countries. The purpose is to comprehensively evaluate the effects of various policies, optimize policy formulation, and provide analytical basis for vigorously promoting my country's scientific and technological innovation and socio-economic development. While improving the efficiency of understanding and analyzing science policies, it can also ensure the correct understanding of science policies. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a screenshot of the homepage interface of the present invention in actual use; Figure 2 This is one of the search interface diagrams used in actual application of the present invention; Figure 3 This is the second image of the search interface used in actual application of the present invention; Figure 4 This is a schematic diagram of the search and editing interface in actual use of the present invention; Figure 5 This is a schematic diagram of the search text deletion interface in actual use of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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 scope of protection of the present invention.

[0018] A method for analyzing science and technology policies includes the following steps: Step 1: Collect science policy documents; Step 2: Classify the collected science policy documents, and use techniques for grouping and classifying fragments, paragraphs or documents to complete feature extraction; Step 3: Establish a "3+5" policy theme statement database, text summary database, entity database, and relation database; Step 4: Build a model and use the database obtained in Step 3 to train and test it, resulting in a science and technology policy document analysis module capable of analyzing science and technology policy documents.

[0019] Specifically, in step two of this embodiment, text preprocessing is performed before classifying scientific policy documents to remove stop words, punctuation marks, special characters, and case conversion.

[0020] Specifically, in step two of this embodiment, the feature extraction operation is as follows: after converting the text into a word frequency vector, TF-IDF (word frequency-inverse document frequency) is used for weighting; each word is converted into a high-dimensional vector using Word2Vec or GloVe respectively, and the results are compared; and the features are filtered according to their importance or relevance to reduce noise and overfitting.

[0021] More specifically, the "3+5" policy theme statement library in step three of this embodiment includes the following: The text contains statements related to the theme of "science and technology policy and strategy". "Science and technology policy and strategy" is a general guideline for solving the overall problems between the development of science and technology and the development of the national economy and society, as well as the relationship between various goals. It is the most concentrated and macro-level manifestation of the science and technology policy system. The text contains sentences related to the theme of "science and technology planning and layout". "Science and technology planning and layout" refers to a long-term plan that is implemented in stages. It is formulated under the guidance of science and technology development policies and strategies. Science and technology planning and layout makes specific provisions and designs for the approaches, measures and steps of science and technology activities. The text containing statements related to the theme of "institutional arrangements for science and technology" states that "science and technology planning and layout" refers to the organizational system and institutional design of science and technology activities; The text contains statements related to the theme of "resource allocation policy." "Resource allocation policy" refers to the allocation of resources to various entities and uses according to certain rules and mechanisms, so as to create as much benefit as possible from limited resources. The text contains statements related to the theme of "science and technology talent policy". "Science and technology talent policy" refers to the code of conduct formulated to regulate the behavior of science and technology talents, which includes a series of laws, measures, methods, and regulations concerning the cultivation, attraction, mobility, incentives and management of science and technology talents. The text of statements related to the theme of "innovation subject policy" refers to a series of laws, measures, methods, and regulations concerning the national innovation system, including the government (leading role), research universities (basic and emerging role), national research institutions (basic and strategic research), and enterprise R&D organizations (the main body of technological innovation). The text of statements related to the theme of "technology transfer policy" refers to the sum of various direct and indirect means, regulations and methods for promoting the transfer of scientific and technological achievements, regulating the transfer process and the behavior of transfer entities; The text contains statements related to the theme of "environmental creation policy." "Environmental creation policy" refers to the environmental factors that the government indirectly influences through policies such as regulations, review and approval, and intellectual property protection, thereby affecting scientific and technological development.

[0022] More specifically, the model in step four of this embodiment is a Naive Bayes model.

[0023] More specifically, the Naive Bayes model described in this embodiment: The Naive Bayes algorithm from Scikit-Learn is selected for text classification. The joint probability of feature terms and categories is used to estimate the category probability of the text. The specific operation includes the following steps: Step 1: Import the word vector Bunch object after word segmentation; Step 2: Construct the TF-IDF vector space for the test set; Step 3: Import the bag-of-words from the training set; Step 4: Initialize the vector space model using TFIDFVectorizer; Step 5: Create persistence for the bag of words.

[0024] Furthermore, step four of this embodiment, which involves training the model using a training set, includes the following steps: Step 1: Import the Multinomial Bayes algorithm package; Step 2: Import the training set vector space; Step 3: Import the test set vector space; Step 4: Apply the Naive Bayes algorithm; Step 5: Predict the classification result; Step 6: Display the prediction accuracy.

[0025] Furthermore, in this embodiment, the distance between the test set samples and the training set samples is recorded using Euclidean distance. The k samples with the shortest distances are used as the neighbors of the predicted samples. The class with the most frequent distance among the k neighbors is selected as the predicted value of the test set sample through a voting mechanism.

[0026] Furthermore, in step four of this embodiment, the Naive Bayes model uses a test set to evaluate the model's performance. Commonly used evaluation metrics include accuracy, recall, and F1 score. At the same time, cross-validation is used to further evaluate the model's stability and generalization ability.

[0027] Furthermore, in the model evaluation process described in this embodiment, feature engineering or model parameters are adjusted based on the evaluation results to optimize the model's performance.

[0028] The actual user interface of this invention is described below: (I) Tools Homepage like Figure 1 As shown, the homepage has two main functional areas, corresponding to the two system requirements of document annotation and topic extraction from top to bottom: Search bar: Used for document annotation. Supports single and multiple keyword searches, secondary searches within search results, and specifying (or entering) the topic and keywords for the final search results. Document Upload: Used for topic extraction. Documents are uploaded and topics are extracted by dragging and dropping them. In addition to the two main functional areas mentioned above, there are also two function links in the upper right corner of the homepage: "Topic Statement" and "Text Summary".

[0029] (ii) Search function like Figure 2As shown, this includes displaying search results, searching within search results, specifying or entering a topic, specifying or entering keywords, etc.

[0030] like Figure 3 As shown, after selecting "Search in results," the original keyword turns gray to distinguish it from subsequent keywords. Clicking the magnifying glass icon to the right of the search box displays the search results for "Search in results," as follows: The "Search in results" option can currently only be used once.

[0031] After selecting "Search in results," all keywords in the search box are cleared, and the search returns to its original state. However, the previous search results are still displayed on the page.

[0032] The arrow to the left of "Search in results" is a back button, allowing you to view the original search results. Alternatively, you can use your browser's back button.

[0033] (iii) Search result editing function like Figure 4 As shown, once the search results are determined, the process proceeds to match and rank the topics with the keywords (i.e., tags). This process identifies which tags correspond to the topic or each sub-dimension. Once the tags are determined, the backend can perform the corresponding classification calculations based on a pre-defined algorithm.

[0034] 1. Click "Edit Results" in the left window; the browser page will expand into a two-window window. The browser will automatically maximize. 2. Search results in the left window can be dragged to the right window, and can be freely dragged within the right window to change their top and bottom order; 3. The top of the right window displays a tree-like drop-down menu with eight topics and their corresponding sub-dimensions. The topic selected during annotation applies to all search results within this window. 4. Click the "Edit" button at the top of the right window to make the "Content / Keyword Display" column editable. Content can be freely deleted or added. Except for highlighted keywords, all other keywords are marked with an "X" by default; selecting them will add them. 5. Click the "Confirm" button at the bottom of the right window to add the results to the database. Close the right window, and the browser will return to the state after "Search in results".

[0035] (iv) Topic extraction function The process of extracting policy themes begins on the homepage.

[0036] 1. Drag and drop the document you want to work on regarding policy topics from your PC desktop to the document processing toolbar. Release the mouse button after the toolbar color changes, and the document will be uploaded to the server. 2. Alternatively, you can click the "Extract" button at the bottom of the function area. The system will then select the target document from the desktop or folder and upload it to the server.

[0037] (v) Theme extraction and editing function In principle, topic extraction only requires the display of results. However, for ease of use, the topic extraction results can be edited, including deleting topic statements and adding topics.

[0038] 1. Click the "Edit" button to bring the content and annotations on the results page into editable mode. Highlighted sections can be reselected. Themes can be switched; 2. Click the "Confirm" button to save the edited results to the database and end the editing process.

[0039] (vi) Theme sentence display function Click the "Topic Statements" button on the homepage or other pages to enter the topic statement display page.

[0040] The top of the window features a drop-down menu with eight themes.

[0041] The drop-down menu is empty by default, and the main area of ​​the page displays all sentences or paragraphs corresponding to the 8 topics. 2. Open the drop-down menu, select any policy topic, such as "Science and Technology Policy and Strategy", and the main area of ​​the page will display all statements or paragraphs corresponding to the "Science and Technology Policy and Strategy" topic; 3. The content displayed in the topic sentence is editable. Click the icon on the right side of the record to "Edit" (left) or "Delete" (right).

[0042] 4. Click the "Edit" icon to enter editable mode for the "Theme" section. You can reselect and modify the highlighted text, or change the policy theme; 5. Click the "Delete" icon. The record will be deleted after confirmation (e.g., ...). Figure 5 (As shown).

[0043] (vii) Text summary display function Click the "Text Summary" button on the homepage or other pages to enter the text summary display page.

[0044] The search box is located at the top of the window. Entering keywords will quickly find document names or summary content.

[0045] The search box is empty by default, and the main area of ​​the page displays all completed text summaries; 2. Enter keywords in the search box. After searching, the corresponding filenames and their text summaries will be displayed. Documents without text summaries can be displayed only as document names. 3. The text summary is editable. Click the icon on the right side of the record to "Edit" (left) or "Delete" (right).

[0046] 4. Click the "Edit" icon to make the "Summary" section editable. In principle, only deletion of existing summary statements is allowed; adding or modifying text is not permitted. 5. Click the "Delete" icon, and the record will be deleted after confirmation.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for analyzing science and technology policies, characterized in that: Includes the following steps: Step 1: Collect science policy documents; Step 2: Classify the collected science policy documents, and use techniques for grouping and classifying fragments, paragraphs or documents to complete feature extraction; Step 3: Establish a "3+5" policy theme statement database, text summary database, entity database, and relation database; Step 4: Build a model and use the database obtained in Step 3 to train and test it, resulting in a science and technology policy document analysis module capable of analyzing science and technology policy documents.

2. The method for analyzing science and technology policies according to claim 1, characterized in that: In step two, text preprocessing is performed before classifying scientific policy documents to remove stop words, punctuation marks, special characters, and case conversion.

3. The method for analyzing science and technology policies according to claim 1, characterized in that: The feature extraction operation in step two is as follows: the text is converted into a word frequency vector and then weighted using TF-IDF; each word is converted into a high-dimensional vector using Word2Vec or GloVe respectively, and the results are compared; and the features are filtered according to their importance or relevance to reduce noise and overfitting.

4. The method for analyzing science and technology policies according to claim 1, characterized in that: The "3+5" policy theme statement library in step three includes the following: The text contains statements related to the theme of "science and technology policy and strategy". "Science and technology policy and strategy" is a general framework for solving the overall problems between the development of science and technology and the development of the national economy and society, as well as the relationship between various goals. It is the most concentrated and macro-level manifestation of the science and technology policy system. The text contains sentences related to the theme of "science and technology planning and layout". "Science and technology planning and layout" refers to a long-term plan that is implemented in stages. It is formulated under the guidance of science and technology development policies and strategies. Science and technology planning and layout makes specific provisions and designs for the approaches, measures and steps of science and technology activities. The text contains sentences related to the theme of "institutional arrangements for science and technology," where "science and technology planning and layout" refers to the organizational system and institutional design of science and technology activities. The text contains sentences related to the topic of "resource allocation policy". "Resource allocation policy" refers to the allocation of resources to various entities and uses according to certain rules and mechanisms, so as to make limited resources create as much benefit as possible. The text contains statements related to the theme of "science and technology talent policy". "Science and technology talent policy" refers to the code of conduct formulated to regulate the behavior of science and technology talents, which includes a series of laws, measures, methods, and regulations concerning the cultivation, attraction, mobility, incentives and management of science and technology talents. The text contains statements related to the theme of "innovation entity policies," which refers to a series of laws, measures, methods, and regulations concerning the national innovation system, including the government, research universities, national research institutions, and corporate R&D organizations. The text contains statements related to the theme of "technology transfer policy". "Technology transfer policy" refers to the sum of various direct and indirect means, regulations and methods for promoting the transfer of scientific and technological achievements, regulating the transfer process and the behavior of transfer entities. The text contains sentences related to the theme of "environmental creation policy," which refers to the environmental factors that the government indirectly influences through policies such as regulations, review and approval, and intellectual property protection, thereby affecting scientific and technological development.

5. The method for analyzing science and technology policies according to claim 1, characterized in that: The model in step four is a Naive Bayes model.

6. The method for analyzing science and technology policies according to claim 5, characterized in that: The Naive Bayes model described uses Scikit-Learn's Naive Bayes algorithm for text classification, estimating the class probability of the text using the joint probability of feature terms and categories. The specific steps include: Step 1: Import the word vector Bunch object after word segmentation; Step 2: Construct the TF-IDF vector space for the test set; Step 3: Import the bag-of-words from the training set; Step 4: Initialize the vector space model using TFIDFVectorizer; Step 5: Create persistence for the bag of words.

7. The method for analyzing science and technology policies according to claim 6, characterized in that: Step four, which involves training the model using a training set, includes the following steps: Step 1: Import the Multinomial Bayes algorithm package; Step 2: Import the training set vector space; Step 3: Import the test set vector space; Step 4: Apply the Naive Bayes algorithm; Step 5: Predict the classification result; Step 6: Display the prediction accuracy.

8. The method for analyzing science and technology policies according to claim 7, characterized in that: The distance between the test set samples and the training set samples is recorded using Euclidean distance. The k samples with the shortest distances are selected as the neighbors of the predicted samples. The class with the most frequent distance among the k neighbors is selected as the predicted value of the test set sample through a voting mechanism.

9. The method for analyzing science and technology policies according to claim 5, characterized in that: In step four, the Naive Bayes model uses a test set to evaluate the model's performance. Common evaluation metrics include accuracy, recall, and F1 score. Cross-validation is also used to further evaluate the model's stability and generalization ability.

10. A method for analyzing science and technology policies according to claim 9, characterized in that: During the model evaluation process, feature engineering or model parameters are adjusted based on the evaluation results to optimize model performance.

Images