A method and system for constructing and displaying popular science content based on a graph

Abstract

The application discloses a kind of based on atlas construction display popular science content method and system, adopt three-dimensional force guide as the display carrier of popular science content, and popular science knowledge structure model is constructed by atlas technology, wherein the popular science knowledge structure model includes multiple nodes, each node is a popular science knowledge point, and the associated relationship between knowledge points is represented by connecting line between nodes;Utilize three-dimensional technology to stereoscopic rendering is carried out to the popular science knowledge structure model, and the popular science content interface with intuitive, image display effect is generated;User is clicked after any node on the popular science content interface, and the popular science article or article list corresponding to the clicked node is called out for user to consult.The application is based on atlas construction display popular science content method, and the deficiencies of existing popular science means in knowledge presentation, user interaction and content correlation are effectively solved by knowledge structure visualization, stereoscopic rendering and interactive experience and accurate content push.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of popular science popularization, and particularly relates to a method and system for constructing and displaying popular science content based on a graph. BACKGROUND

[0002] With the development of information technology and the increasing demand for scientific literacy among the public, popular science education is gradually moving towards digitization, networking and personalization. Traditional linear text or single media forms have been unable to meet the needs of modern users for diversified knowledge acquisition, interactivity and in-depth exploration. Therefore, how to innovate the construction and display of popular science content to make it more intuitive, vivid and easy to understand has become an important issue to be solved in the current popular science field.

[0003] Existing popular science methods often rely on static text, pictures or simple multimedia combinations, lacking systematic revelation and dynamic presentation of the internal structure of knowledge. Although some online platforms provide rich popular science resources, users often have difficulty quickly locating the required content when faced with a large amount of information, and it is even more difficult to intuitively perceive the correlation and context between knowledge points. In addition, the lagging update of popular science content, the lack of in-depth analysis and interdisciplinary perspectives also restrict the effectiveness of popular science education. SUMMARY

[0004] The purpose of the present application is to provide a method and system for constructing and displaying popular science content based on a graph, using a three-dimensional force-directed graph as a display carrier for popular science content, and skillfully using graph technology and three-dimensional rendering technology, aiming to build a highly visual, interactive and clear knowledge structure popular science platform, to overcome the limitations of existing popular science methods, improve the quality of popular science education and user experience, and solve the problems in the prior art mentioned in the background.

[0005] To achieve the above purpose, the present application adopts the following technical solution: a method for constructing and displaying popular science content based on a graph, comprising the following steps: using a three-dimensional force-directed graph as a display carrier for popular science content, constructing a popular knowledge structure model through graph technology, wherein the popular knowledge structure model comprises a plurality of nodes, each node being a popular knowledge point, and the nodes are connected by lines to represent the correlation between knowledge points; using three-dimensional technology to render the popular knowledge structure model in three dimensions to generate a popular content interface with intuitive and visual display effects; after a user clicks on any node on the popular content interface, the corresponding popular article or article list is displayed for the user to review.

[0006] Preferably, the construction of the popular knowledge structure model comprises: determining the position coordinates of each node and the curvature of the connecting line according to the importance of the popular knowledge points and the logical relationship between them through formula 1:

[0007] Formula 1: (x_i, y_i, z_i) = f(importance_i, relations_i).

[0008] Preferably, the construction of the popular science knowledge structure model further comprises: using Formula 2 to statistically analyze the keywords of the popular science knowledge points and their frequencies:

[0009] Formula 2: TF-IDF(k_i, d_j) = tf(k_i, d_j) * log(N / df(k_i)).

[0010] Preferably, the popular science content interface further comprises a search box, which is used to calculate the similarity between the user query keywords and the node labels using Formula 3 after the user inputs the query keywords, and display the relevant nodes in order of similarity:

[0011] Formula 3: similarity(q, t) = cosine(vec(q), vec(t)).

[0012] Preferably, the method further comprises the following steps to achieve personalized recommendation of nodes:

[0013] Collecting user browsing behavior data during the use of the popular science content interface, processing the collected browsing behavior data, and calculating the interest degree of the user to each popular science knowledge point based on Formula 4; Formula 4;

[0014] interest_u(k) = a*click_u(k) + β*read_u(k) + γ*dwell_time_u(k) + δ*search_freq_u(k);

[0015] According to the user interest degree matrix and the browsing behavior data of all users, the collaborative filtering algorithm is used to calculate the predicted interest degree of the user to the popular science knowledge points that have not been contacted.

[0016] Marking the popular science knowledge points with a predicted interest degree higher than a preset threshold as the potential interest points of the user, and displaying them with an identifier in the popular science content interface, or recommending them to the user through a push notification.

[0017] Preferably, the method further comprises the following steps to achieve knowledge Q&A of popular science content:

[0018] Designing and constructing a knowledge Q&A system, which receives the question input of the user related to the popular science content, and the question input is analyzed by a natural language processing module, key information is extracted and converted into a standardized query statement;

[0019] The semantic similarity of the standardized query statement and each node label in the popular science knowledge structure model is calculated by using formula 5, formula 5; sim(Q, T) =∑w_i*cos(vec(q_i), vec(t_i)), i = 1, 2, …, n;

[0020] Nodes with a semantic similarity higher than a preset threshold are marked as related to the question, and are arranged in descending order of similarity;

[0021] The system extracts answer information from the popular science articles or article list corresponding to the related nodes according to the sorting result, generates a reply text, returns the reply text to the user, and highlights the nodes related to the question on the popular science content interface.

[0022] Preferably, the method further comprises the following steps to achieve personalized recommendation of popular science content:

[0023] User interest behavior data is collected, and machine learning algorithms are used to analyze the user interest behavior data to extract the user's interest theme, attention degree and learning progress characteristics in the popular science field, and a user portrait model is established;

[0024] The nodes in the popular science knowledge graph are multi-dimensionally labeled, covering knowledge point difficulty level, subject field, applicable age group, and associated hot event factors;

[0025] A recommendation algorithm is designed, and formula 6 is used to calculate the matching degree of the user and each node in the graph according to the user portrait model and node labeling information, formula 6; M(U, N) = α*P(N|U) + β*D(N) + γ*R(N);

[0026] According to the sorting result of the matching degree, the popular science articles or article list corresponding to the top-K nodes are selected as the recommended content, and the personalized recommendation result is displayed in the popular science content interface, and the recommendation list is dynamically updated to respond to user behavior changes and knowledge graph updates.

[0027] Preferably, the method further comprises the following steps to achieve community interaction and knowledge Q&A of popular science content:

[0028] A user interaction module is established, so that the user can perform social interaction operations under the popular science content interface, and relevant interaction data is recorded;

[0029] An online Q&A system is integrated, and the user can ask questions about the popular science articles, the system automatically identifies the relevance of the questions and the knowledge graph nodes, assigns appropriate labels to the questions, and pushes them to community members or experts with relevant professional knowledge backgrounds;

[0030] Receiving and displaying the answers of community members or experts to the user's questions, the answer content can be directly linked to the related graph node, analyzing the user's submitted questions, answers and comment texts, extracting keywords and sentiment tendency information to enrich the user portrait, and at the same time, used for optimizing the retrieval and recommendation of community content;

[0031] Periodically sorting the high-value question and answer content in the community, and highlighting the display of the popular science content interface through system recommendation or manual editing.

[0032] Preferably, the method further comprises the following steps to realize the personalized push and learning path planning of the popular science content:

[0033] Establishing a user portrait system, collecting and analyzing the multi-dimensional data of user's browsing history, interactive behavior, interest preference and knowledge level, and forming a user personalized feature model;

[0034] Combining the user portrait and the knowledge graph structure, designing a push algorithm based on content similarity, knowledge dependency relationship and user interest matching degree, and recommending the popular science articles most relevant to the current learning state and interest of the user in real time;

[0035] Providing a user-defined subscription function, so that the user selects a specific theme, field or expert focus list, and the system pushes the customized popular science content regularly according to the list;

[0036] Building a learning path planning engine, intelligently generating a personalized learning path graph based on the knowledge nodes mastered by the user, learning goals and logical associations of the knowledge graph, including recommended learning order, key knowledge points and supplementary learning resources;

[0037] Implementing a learning progress tracking and feedback mechanism, recording the user's reading and understanding of the pushed content, obtaining the user's satisfaction and learning effect feedback of the pushed content through interactive testing and questionnaire survey, and dynamically adjusting the push strategy and learning path planning.

[0038] On the other hand, the present application proposes a system for constructing and displaying popular science content based on a graph, comprising:

[0039] A popular science knowledge structure model construction module is used to adopt a three-dimensional force guide map as a display carrier of popular science content, and a popular science knowledge structure model is constructed through graph technology, wherein the popular science knowledge structure model contains a plurality of nodes, each node is a popular science knowledge point, and the association relationship between knowledge points is represented through a connecting line between nodes;

[0040] A popular science content interface establishment module is used to perform three-dimensional rendering on the popular science knowledge structure model by using three-dimensional technology, and a popular science content interface with intuitive and visual display effect is generated.

[0041] The popular science content display module is configured to display a popular science article or an article list corresponding to the clicked node for the user to browse after the user clicks any node on the popular science content interface.

[0042] The technical effects and advantages of the present application are as follows.

[0043] The method for constructing and displaying popular science content based on a graph atlas effectively solves the deficiencies of existing popular science means in knowledge presentation, user interaction and content relevance through knowledge structure visualization, three-dimensional rendering and interactive experience and precise content pushing, realizes intuitive and visual display of popular science content, significantly improves the learning efficiency, understanding and memory effect of the user, and effectively promotes the modernization process of popular science education. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The flowchart of the method for constructing and displaying popular science content based on a graph atlas in the embodiment of the present application is shown in the figure.

[0045] Figure 2 The flowchart of the personalized recommendation of the node in the embodiment of the present application is shown in the figure.

[0046] Figure 3 The flowchart of the knowledge Q&A of the popular science content in the embodiment of the present application is shown in the figure.

[0047] Figure 4 The flowchart of the personalized recommendation of the popular science content in the embodiment of the present application is shown in the figure.

[0048] Figure 5 The flowchart of the community interaction and knowledge Q&A of the popular science content in the embodiment of the present application is shown in the figure.

[0049] Figure 6 The flowchart of the personalized pushing and learning path planning of the popular science content in the embodiment of the present application is shown in the figure.

[0050] Figure 7 The module diagram of the system for constructing and displaying popular science content based on a graph atlas in the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0051] 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. The specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention. 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.

[0052] This invention provides a method for constructing and displaying popular science content based on graphs. It uses three-dimensional force diagrams as the display carrier for popular science content and cleverly utilizes graph technology and three-dimensional rendering technology. The aim is to build a highly visualized, highly interactive, and clearly structured popular science platform to overcome the limitations of existing popular science methods and improve the quality of popular science education and user experience.

[0053] like Figure 1 As shown, Figure 1 This is a flowchart of a method for constructing and displaying popular science content based on a graph according to an embodiment of the present invention.

[0054] In this embodiment, a method for constructing and displaying popular science content based on a graph includes the following steps:

[0055] Step 1: Use a three-dimensional force diagram as the display carrier for popular science content, and construct a popular science knowledge structure model through graph technology. The popular science knowledge structure model contains multiple nodes, each node is a popular science knowledge point, and the relationship between knowledge points is represented by lines between nodes.

[0056] Specifically, the construction of the science popularization knowledge structure model includes: determining the position coordinates of each node and the curvature of the connecting lines based on the importance of science popularization knowledge points and their logical relationships, using Formula 1.

[0057] Formula 1: (x_i,y_i,z_i)=f(importance_i,relations_i);

[0058] Where (x_i, y_i, z_i) are the three-dimensional coordinates of the i-th node, importance_i is the importance coefficient of the popular science knowledge point represented by the i-th node, and relations_i represents the logical relationship parameters between the i-th node and other nodes. The importance coefficient of a node can be determined comprehensively based on factors such as its core position in the scientific system, public attention, and teaching syllabus requirements. The logical relationship parameters reflect the logical connections between knowledge points, such as causal relationships, inclusion relationships, and comparative relationships, and can be obtained through expert evaluation, literature analysis, or automatic reasoning from knowledge graphs.

[0059] The specific implementation function f of Formula 1 can select a suitable layout algorithm according to actual needs, such as force-directed layout, hierarchical layout, etc. The force-directed layout simulates the action of physical force, and adjusts the node position through attractive force (to keep the connection line between nodes as short as possible) and repulsive force (to prevent the nodes from being too dense); the hierarchical layout arranges the nodes in the vertical or horizontal direction according to the hierarchical relationship between the nodes. These layout algorithms can dynamically calculate the node coordinates according to the node importance and logical relationship parameters, to ensure the clarity and readability of the knowledge structure model.

[0060] The curvature of the connection line reflects the closeness or complexity of the relationship between the nodes, which can be represented by visual features such as the length, degree of bending, color, and thickness of the connection line. In some cases, the curvature can be directly calculated from the node position, while in other scenarios, additional parameters or algorithms may be needed to determine it.

[0061] In order to further enrich the information amount of the popular science knowledge structure model, Formula 2 is used to statistically analyze the keywords of popular science knowledge points and their frequency of occurrence:

[0062] Formula 2: TF-IDF(k_i, d_j) = tf(k_i, d_j) * log(N / df(k_i));

[0063] Wherein, TF-IDF(k_i, d_j) represents the weight of keyword k_i in document d_j, tf(k_i, d_j) is the number of occurrences of keyword k_i in document d_j, N is the total number of documents, and df(k_i) is the number of documents containing keyword k_i. TF-IDF is a statistical method widely used in information retrieval and text mining fields, which is used to measure the importance of a word to a document set or a document in a specific context.

[0064] Term Frequency (TF): The frequency of a keyword appearing in a document, which intuitively reflects the local importance of the keyword in the document. High-frequency keywords usually have strong representativeness for the theme of the document.

[0065] Inverse Document Frequency (IDF): A measure of the universality of a keyword in the entire document set, calculated as log(N / df(k_i)) in the formula. The smaller df(k_i) is, the fewer documents containing the keyword, indicating that the word has high uniqueness or specificity. The higher the IDF value, the more discriminative the word is in the global context.

[0066] The TF-IDF value comprehensively considers the local importance and global importance of keywords, and can filter out the most representative and discriminative keywords from a large number of documents, which are used to identify the core concepts, key terms or professional vocabulary of popular science knowledge points. These keywords can be displayed as node labels, annotations or auxiliary information on the popular science knowledge structure model, enhancing the information density and query convenience of the model.

[0067] The specific implementation is as follows: based on the above principles, the specific implementation process of constructing a popular science knowledge structure model based on a three-dimensional force-directed graph is as follows:

[0068] Data collection and preprocessing: Collect various popular science resources such as popular science articles, textbooks, reports, video scripts, etc., and perform text cleaning, word segmentation, and standardization processing to form a document set for analysis.

[0069] Node generation and attribute assignment: According to the popular science theme, divide the knowledge points, create a node for each knowledge point, and assign corresponding attribute values (importance_i, relations_i) according to its importance, correlation, etc. These attribute values will be used to calculate the coordinates of the node in the three-dimensional space through formula 1.

[0070] Layout algorithm application: Use the selected layout algorithm (such as force-directed layout) to calculate the precise coordinates (x_i, y_i, z_i) of the node in the three-dimensional space according to the node attributes, forming a preliminary knowledge structure model framework.

[0071] Keyword statistics and TF-IDF calculation: Extract keywords from each document and count the number of occurrences, then calculate the TF-IDF value of each keyword in each document and the entire document set using formula 2.

[0072] Keyword integration and visualization: Associate the high TF-IDF value keywords calculated with the corresponding popular science knowledge point nodes, and display them in the form of labels, annotations, etc. on the three-dimensional force-directed graph to increase the information richness of the model.

[0073] Model optimization and interactive design: Adjust the model layout, color, font, and other visual elements according to user feedback and test results to optimize user experience. At the same time, design user interaction functions such as node click triggering related content display, keyword search, path navigation, etc. to improve the interactivity and practicality of the model.

[0074] Through the above implementation, the popular science knowledge structure model based on a three-dimensional force-directed graph is constructed, which not only visually displays the popular science knowledge points and their correlation, but also enriches the information content of the model through keyword statistics and analysis, providing a three-dimensional, dynamic, information-intensive and easy-to-interact popular science learning environment for users, which helps to improve the effectiveness of popular science education and public scientific literacy.

[0075] Step two: use three-dimensional technology to render the popular science knowledge structure model in three dimensions, generating a popular science content interface with intuitive and visual display effects;

[0076] Further, the popular science content interface also includes a search box, which is used to calculate the similarity between the user's query keywords and each node label using formula 3 after the user inputs the query keywords, and display the relevant nodes in order of similarity:

[0077] Formula 3: similarity(q, t) = cosine(vec(q), vec(t));

[0078] Where similarity(q, t) represents the similarity between the query keyword q and the node label t, vec(q) and vec(t) are the vectorized representations of the query keyword q and the node label t, respectively, and cosine() is a cosine similarity calculation function.

[0079] Specifically, three-dimensional model construction and rendering

[0080] First, based on the constructed popular science knowledge structure model, use three-dimensional modeling software or programming languages (such as Three.js, Unity, etc.) to convert nodes and connections into three-dimensional geometric entities. Nodes can be designed as spheres, cubes, cylinders, etc., and their size, color, texture can be customized according to the importance, category or user preference of the node. The connection can be represented as a curve, a straight line, a tubular structure, etc., and its thickness, color, transparency can be adjusted according to the closeness or complexity of the relationship between nodes.

[0081] Next, apply lighting, shading, material, animation, and other three-dimensional rendering techniques to make the popular science knowledge structure model present a three-dimensional, realistic visual effect. For example, by setting up environmental light, point light source, spotlight, etc. to simulate the lighting conditions of the real world, give the model rich light and shadow changes; use texture mapping, particle system, etc. to enhance the expressiveness of nodes and connections, making them more visually appealing. In addition, you can also add perspective control, zoom, rotation, etc. functions to allow users to observe the model from different angles and distances, enhancing the interactive experience.

[0082] Search function integration

[0083] The popular science content interface should be equipped with a search box to allow users to quickly locate the popular science knowledge points of interest. When the user inputs the query keywords, the system calculates the similarity between the user's query keywords and each node label using formula 3.

[0084] Cosine similarity is a commonly used measure of the angle between two non-zero vectors, with a value range of [-1, 1], and the closer the value is to 1, the smaller the angle between the two vectors, i.e., the higher the similarity in direction. In the field of text analysis, vectorization representation is usually achieved by methods such as bag-of-words model, TF-IDF model, Word2Vec, BERT, etc. to convert text into numerical vectors. For query keywords q and node labels t, the same vectorization method can be used to obtain their vector representations vec(q) and vec(t), and then the semantic similarity between them can be obtained by calculating the cosine similarity between the two vectors.

[0085] The specific implementation process is as follows:

[0086] User input and preprocessing: The user inputs query keywords in the search box, and the system performs preprocessing operations such as word segmentation, stop word removal, and stemming on the input, forming a list of keywords to be queried.

[0087] Node label vectorization: Apply the same vectorization method (such as TF-IDF, Word2Vec, etc.) to all node labels to generate vector representations vec(t) of the node labels.

[0088] Query keyword vectorization: Apply the same method to the preprocessed query keyword list to obtain the vector representation vec(q) of the query keyword q.

[0089] Similarity calculation: Calculate the cosine similarity between the query keyword vector vec(q) and each node label vector vec(t) using formula 3 to obtain a set of similarity values.

[0090] Result sorting and display: Sort the similarity values in descending order, and the corresponding nodes are the most relevant science popularization knowledge points. In the science popularization content interface, highlight the relevant nodes according to the sorting results, or list the node names, brief descriptions, and similarity scores in the search result area for users to quickly browse and select.

[0091] Through three-dimensional technology, the structure model of science popularization knowledge is rendered in three dimensions, not only realizing the three-dimensional and visual display of science popularization content, but also providing a convenient way to query knowledge points by integrating search function. The search function is based on cosine similarity calculation, which can accurately identify the semantic association between the user input query keywords and the node labels, quickly locate the relevant science popularization knowledge points, and present them to the user in the form of sorting. This combination of three-dimensional visualization and intelligent search in the science popularization content interface greatly improves the efficiency and experience of users in obtaining science popularization knowledge, and is conducive to promoting the popularization and in-depth of science popularization education.

[0092] Step three: The user clicks on any node in the science popularization content interface, and the corresponding science popularization article or article list is displayed for the user to review.

[0093] Specifically, in the science popularization content interface, the user can click on any node to display the corresponding science popularization article or article list, realizing deep knowledge exploration and learning. The specific implementation is as follows:

[0094] Node click event response and article display

[0095] 1. Node and article association storage

[0096] A node and science popularization article association database is established in advance. Each node not only contains its label information, but also associates one or more science popularization article identifiers (such as article ID, URL, etc.). This association can be one-to-one (each node corresponds to a specific article), one-to-many (one node is associated with a series of related articles), or many-to-many (there is cross-association between nodes and articles). The association data should be stored in the database for easy querying and updating.

[0097] 2. Node click event processing

[0098] In the front-end code of the science popularization content interface, add a click event listener for each interactive node. When the user clicks on a node, the following operations are triggered:

[0099] a. Get node ID: Capture the node ID (or other unique identifier) clicked by the user, which is the key information for querying associated articles.

[0100] b. Send request: The front-end sends an asynchronous request to the back-end server, and the request data includes the clicked node ID and possibly other parameters (such as user permissions, language settings, etc.).

[0101] c. Server response: After the server receives the request, it queries the association database and retrieves all science popularization article information associated with the node ID. These information may include article title, abstract, author, publication time, link, etc.

[0102] d. Return result: The server returns the query result to the front-end in JSON or other suitable data format.

[0103] 3. Article display and interaction

[0104] After the front-end receives the article data returned by the server, it dynamically generates an article list or directly loads a single article view, depending on the node and article association mode:

[0105] a. One-to-one mode: If a node directly corresponds to a specific article, the frontend directly loads the detailed content of the article, including the title, body, multimedia elements such as pictures and videos, and social functions such as comments and sharing. The article should be presented in an easy-to-read layout, such as paragraphs, lists, quotes, and other formats, and support user operations such as scrolling, zooming, and full-screen.

[0106] b. One-to-many mode: If a node is associated with multiple related articles, the frontend generates an article list view, with list items containing article titles, summaries, authors, publication times, and other basic information, as well as links or buttons to jump to complete articles. The list can be sorted by publication time, relevance, user ratings, and other criteria, and provide pagination, filtering, and search functions to facilitate user browsing and selection of interesting articles.

[0107] When a user clicks on any node in the science popularization content interface, the system can accurately call out the science popularization article or article list associated with the clicked node by triggering the node click event. This process relies on a pre-established node-article association database and a front-end and back-end collaborative event handling mechanism. The front-end is responsible for capturing user operations, sending requests, and dynamically displaying article content, while the back-end is responsible for receiving requests, querying the database, and returning results. Whether loading a single article directly or presenting an article list, the goal is to provide users with rich, detailed, and targeted science popularization knowledge to further deepen their understanding and exploration of the clicked node topic. Such design not only enhances the interactivity and practicality of the science popularization content interface, but also greatly enriches users' personalized learning paths, helping to improve the effectiveness of science popularization education and user satisfaction.

[0108] In this embodiment, as Figure 2 The above-mentioned method for constructing and displaying science popularization content based on a graph further includes the following steps to achieve personalized recommendation of nodes:

[0109] Collect user browsing behavior data during use of the science popularization content interface, process the collected browsing behavior data, and calculate the user's interest in each science knowledge point based on formula 4; formula 4;

[0110] interest_u(k)=a*click_u(k)+b*read_u(k)+g*dwell_time_u(k)+d*search_freq_u(k), wherein, interest_u(k) represents the interest degree of user u to science popularization knowledge point k, click_u(k) is the number of times that user u clicks science popularization knowledge point k, read_u(k) is the number of times that user u inquires the article related to science popularization knowledge point k, dwell_time_u(k) is the dwell time of user u in the interface related to science popularization knowledge point k, search_freq_u(k) is the frequency that user u searches the keyword related to science popularization knowledge point k, a, b, g and d are preset weight coefficients; these coefficients can be set according to actual business requirements, data analysis results or expert experience, and generally, the sum of these coefficients is required to be 1, so as to ensure the dimensional consistency of the interest degree.

[0111] According to the user interest degree matrix and the browsing behavior data of all users, the collaborative filtering algorithm is used to calculate the predicted interest degree of user u to the science popularization knowledge point not contacted.

[0112] The science popularization knowledge point with a predicted interest degree higher than a preset threshold is marked as the potential interest point of user u, and is displayed in the science popularization content interface with an identifier, or is recommended to the user in the form of a push notification.

[0113] The specific implementation is as follows:

[0114] I: User behavior data collection and processing

[0115] 1. Data source and type

[0116] In the process of using the science popularization content interface, the user will generate various browsing behavior data, and these data are the basis for calculating the user interest degree. The main behavior data includes:

[0117] Click behavior: the number of times that the user clicks the science popularization knowledge point node (click_u(k)), which reflects the attention degree of the user to the knowledge point.

[0118] Reading behavior: the number of times that the user inquires the article related to science popularization knowledge point k (read_u(k)), which reflects the willingness of the user to learn the knowledge point in depth.

[0119] Dwell time: the cumulative dwell time of the user in the interface related to science popularization knowledge point k (such as the article detail page, the related topic page and the like) (dwell_time_u(k)), which indirectly measures the concentration degree and absorption condition of the user to the content.

[0120] Search behavior: The frequency of users searching for keywords related to science popularization knowledge point k (search_freq_u(k)) reveals the intensity of users' demand for actively seeking relevant information.

[0121] 2. Data cleaning and standardization

[0122] After collecting the original behavior data, clean it to remove invalid data such as outliers and duplicate records. For different types of indicators, standardization may be needed, such as normalization or logarithmic transformation, to ensure that each indicator is comparable when calculating interest.

[0123] Two: User interest calculation

[0124] Use formula 4 to calculate the interest of user u in science popularization knowledge point k. The principle of formula 4 is as follows:

[0125] This formula uses a weighted sum model to comprehensively measure the user's interest in science popularization knowledge points. The meaning of each behavior indicator is as follows:

[0126] click_u(k): The number of clicks reflects the user's initial attention and curiosity about the knowledge point. High-frequency clicks may indicate that the user has a high immediate interest in the knowledge point.

[0127] read_u(k): The number of readings represents the user's in-depth exploration of the knowledge point. Multiple readings of the same knowledge point-related articles indicate that the user has a continuous learning demand and interest in the knowledge point.

[0128] dwell_time_u(k): Longer dwell time indicates that the user has invested more effort in the knowledge point content, which may be due to recognition of the content quality or because the knowledge point is difficult to understand and requires more time. Overall, longer dwell time is positively correlated with higher interest.

[0129] search_freq_u(k): Search frequency reflects the user's initiative to actively seek information about a specific knowledge point. High-frequency searches often correspond to strong learning desire and potential interest.

[0130] By assigning appropriate weight coefficients to each behavior indicator, the formula integrates the user's behavior performance in different dimensions to generate a value that comprehensively reflects the user's interest in knowledge point k.

[0131] Three: Collaborative filtering prediction and potential interest point labeling

[0132] 1. User interest matrix construction

[0133] Organize all users' interest degree data of all science popularization knowledge points into a user-knowledge point interest degree matrix. Each element (u, k) in the matrix is interest_u(k).

[0134] 2. Collaborative filtering algorithm application

[0135] Using the user interest degree matrix, the predicted interest degree of user u for the science popularization knowledge points not contacted is calculated by a collaborative filtering algorithm (such as user-based collaborative filtering or item-based collaborative filtering). The algorithm is based on the principle of "like attracts like, and people are divided into groups", assuming that users with similar browsing behaviors may have similar interest degrees for the same knowledge points not contacted, or the user's interest in known knowledge points can infer his interest in unknown knowledge points.

[0136] 3. Prediction interest degree threshold setting and potential interest point identification

[0137] Set a preset threshold, and mark the science popularization knowledge points with a predicted interest degree higher than the threshold as potential interest points of user u. These knowledge points are the content that the system speculates the user may be interested in but has not contacted according to the user's historical behavior and group behavior patterns.

[0138] Four: personalized recommendation display

[0139] The science popularization knowledge points marked as potential interest points are highlighted in the science popularization content interface with special marks (such as highlighting, starring, tagging, etc.), guiding the user to pay attention. In addition, the knowledge points with high predicted interest degree can also be sent to the user as personalized recommendation content through push notification, reminding the user to view.

[0140] In summary, the embodiment collects the user's browsing behavior data in the science popularization content interface, calculates the user's interest degree for each science popularization knowledge point using formula 4, predicts the user's interest in the knowledge points not contacted by combining the collaborative filtering algorithm, and finally recommends the knowledge points with high predicted interest degree as potential interest points. This process makes full use of user behavior data, realizes the precision and personalization of science popularization content recommendation, and helps to improve user experience and knowledge acquisition efficiency.

[0141] In the embodiment, as Figure 3 The above-mentioned method for displaying science popularization content based on a graph further includes the following steps to realize knowledge question and answer of science popularization content:

[0142] Design and build a knowledge question and answer system, which receives user input of questions related to science popularization content, and the question input is analyzed by a natural language processing module to extract key information and convert it into a standardized query statement;

[0143] The semantic similarity between the standardized query statement and each node label in the popular science knowledge structure model is calculated using formula 5, formula 5; sim(Q, T) =∑w_i*cos(vec(q_i), vec(t_i)), i = 1, 2, …, n, wherein sim(Q, T) represents the semantic similarity between the query statement Q and the node label T, Q is composed of n words q_1, q_2, …, q_n, T is composed of m words t_1, t_2, …, t_m, w_i is the weight of the word q_i in the query statement Q, vec(q_i) and vec(t_i) are the vectorized representations of the words q_i and t_i respectively, and cos() is a cosine similarity calculation function;

[0144] Nodes with a semantic similarity higher than a preset threshold are marked as related to the question, and are arranged in descending order of similarity;

[0145] The system extracts answer information from the popular science articles or article lists corresponding to the related nodes according to the sorting results, generates a reply text, returns the reply text to the user, and highlights the nodes related to the question on the popular science content interface.

[0146] The specific implementation is as follows:

[0147] One: Construction of a knowledge Q&A system

[0148] A knowledge Q&A system is designed and developed, which integrates natural language processing technology, can receive user input of questions related to popular science content, perform semantic understanding and information extraction, and then match popular science knowledge nodes in the graph, generate answers and present them to the user.

[0149] Step two: question analysis and standardized query generation

[0150] Question input reception: Users submit questions related to popular science content on the popular science content interface or dedicated Q&A portal, such as “What is a black hole?” and “How does photosynthesis work?”

[0151] Natural language processing: The question input is parsed by the natural language processing module (NLP), which includes lexical analysis, syntactic analysis and semantic analysis sub-modules. Lexical analysis identifies words and phrases in the question, syntactic analysis determines the grammatical relationship between words, and semantic analysis understands the overall meaning and potential intent of the question. Through these processes, the system can accurately grasp the core concepts and themes of the user's question.

[0152] Key information extraction and standardized query statement generation: Based on the analysis results, the system extracts key information (such as entities, actions, attributes, etc.) from the question and converts it into a standardized query statement. The standardized query statement expresses the core demands of the question in a structured form, facilitating accurate matching with node labels in the graph.

[0153] Three: Query statement and node label semantic similarity calculation

[0154] The semantic similarity (sim(Q, T)) between the standardized query statement Q and each node label T in the science popularization knowledge structure model is calculated using Formula 5: Formula 5 principle explanation, this formula calculates the semantic similarity based on the term weight and the cosine similarity of the word vector:

[0155] Term weight (w_i): reflects the importance of the word q_i in the query statement Q. The weight can be calculated by TF-IDF, TextRank, etc. The word with high weight has a greater impact on the overall semantic similarity.

[0156] Word vector (vec(q_i) and vec(t_i)): maps words to a high-dimensional vector space, so that semantically similar words have a closer vector distance in this space. Common word vector models include Word2Vec, GloVe, BERT, etc. Vector representation makes the semantic relationship between words that are difficult to compare directly quantifiable.

[0157] Cosine similarity (cos(vec(q_i), vec(t_i))): calculates the cosine value of the included angle between two word vectors, with a value range of [-1, 1]. The closer the value is to 1, the closer the two vectors are, i.e. the more similar the semantics of the words q_i and t_i are; the closer the value is to -1, the more opposite the direction of the two vectors, i.e. the greater the semantic difference between the words q_i and t_i; a value of 0 indicates that the two vectors are perpendicular and have no semantic relationship.

[0158] Formula 5 calculates the overall semantic similarity between the query statement Q and the node label T by accumulating the similarity of each word pair multiplied by its weight. This calculation method takes into account the importance of words and the semantic similarity between words, effectively measuring the semantic matching degree between the query statement and the node label.

[0159] Four: Related node screening and sorting

[0160] Semantic similarity threshold setting: set a preset threshold, mark the nodes with a semantic similarity higher than the threshold as related to the question.

[0161] Related node sorting: sort the related nodes in descending order of similarity value. The nodes with higher semantic matching degree are more likely to contain the answer to the question.

[0162] Five: Answer extraction and presentation

[0163] Answer information retrieval: The system extracts answer information from the relevant nodes' corresponding popular science articles or article lists based on the sorting results. This step may involve deep text mining, summary generation, knowledge extraction, and other technologies to ensure that the extracted answers are accurate, concise, and closely related to the question.

[0164] Answer text generation: The extracted answer information is organized into easy-to-understand answer text, ensuring fluent language, clear logic, and meeting the user's knowledge acquisition needs.

[0165] User feedback and interface display: Return the answer text to the user and highlight the relevant nodes on the popular science content interface. Highlighting helps users intuitively see the relationship between the question and the knowledge graph nodes, deepens their understanding of the answer, and enhances the interactive experience.

[0166] In summary, this embodiment builds a knowledge question and answer system, receives user questions and performs natural language processing, calculates the semantic similarity between the query statement and the node label using formula 5, filters and sorts relevant nodes, extracts answer information and generates answer text, and finally highlights relevant nodes on the popular science content interface, forming a complete knowledge question and answer process based on the graph. This function improves the knowledge acquisition efficiency and user experience of the popular science content platform, allowing users to quickly and accurately acquire the desired popular science knowledge.

[0167] In this embodiment, as Figure 4 The above-mentioned method for constructing and displaying popular science content based on a graph further includes the following steps to achieve personalized recommendation of popular science content:

[0168] Collect user interest behavior data, analyze user interest behavior data using machine learning algorithms, extract user interest themes, attention levels, and learning progress characteristics in the field of popular science, and establish a user portrait model;

[0169] Multi-dimensional labeling of nodes in the popular science knowledge graph, covering knowledge point difficulty level, subject field, applicable age group, and associated hot event factors;

[0170] Design a recommendation algorithm based on the user portrait model and node labeling information, and use formula 6 to calculate the matching degree between the user and each node in the graph:

[0171] Formula 6; M(U, N) = a * P(N|U) + b * D(N) + g * R(N), where M(U, N) represents the matching degree between user U and graph node N, P(N|U) is the probability of user interest in node N based on user behavior data, D(N) is the difficulty of node N's knowledge points and the adaptability score of user learning progress, R(N) is the degree of node N's association with current hot events, and a, b, g are weight coefficients;

[0172] According to the matching degree ranking result, the Top-K nodes corresponding to the popular science articles or article list are selected as the recommended content, and the personalized recommendation result is displayed in the popular science content interface, and the recommendation list is dynamically updated to respond to user behavior changes and knowledge graph updates.

[0173] The specific implementation is as follows:

[0174] One: User interest behavior data collection and analysis

[0175] Data collection: The system continuously monitors and records various interactive behaviors of users on the popular science content platform, including but not limited to browsing history, search keywords, likes, collections, shares, comments, etc. In addition, user preferences for popular science content can also be actively obtained through questionnaires, user feedback, etc.

[0176] Data analysis: Use machine learning algorithms (such as decision trees, clustering analysis, implicit feedback models, etc.) to analyze the collected user interest behavior data in depth, and extract key features reflecting users' interest themes, attention levels, and learning progress in the field of popular science. For example, by analyzing the popular science categories that users frequently browse, dwell time, and repeated access times, we can identify scientific topics that users are particularly interested in; by tracking users' completed learning tasks, test scores, etc., we can assess their learning progress in specific subject areas.

[0177] User portrait model establishment: Based on the above analysis results, a user portrait model is established. This model uses data-driven methods to depict the unique characteristics of each user, such as interest theme distribution, depth and breadth of focus themes, mastered knowledge points, and expected learning knowledge points, forming a personalized description of user knowledge needs.

[0178] Two: Multi-dimensional annotation of popular science knowledge graph nodes

[0179] Knowledge point difficulty level annotation: For each node in the graph, according to its content complexity, theoretical depth, understanding threshold, etc., divide the difficulty level suitable for users of different knowledge levels (such as primary, intermediate, and advanced).

[0180] Discipline field annotation: Clearly define the discipline classification of each node, such as physics, biology, chemistry, astronomy, earth science, and environmental science, so as to recommend content based on users' interest in specific disciplines.

[0181] Applicable age range annotation: According to the appropriate reading age of popular science content, label the corresponding age range (such as children, adolescents, and adults) for the node. This ensures that the recommended content is consistent with the user's age and meets their cognitive development stage.

[0182] Correlation with hot event factor labeling: Real-time monitoring and analysis of social hotspots, scientific discoveries, and major technological events, etc. The relevant science popularization nodes are labeled as hot content, increasing the timeliness and attractiveness of recommendations.

[0183] Three: recommendation algorithm design and matching degree calculation

[0184] Design the recommendation algorithm, based on the user portrait model and node labeling information, use formula 6 to calculate the matching degree (M(U, N)) between user U and graph node N. The principle of the formula is explained, formula 6 integrates three core indicators by weighted fusion, and comprehensively evaluates the matching degree between user and node:

[0185] Interest prediction probability P(N|U): Based on user historical behavior data, use collaborative filtering, deep learning, etc. Algorithm to predict the user's interest in node N. The higher the probability value, the more interested the user is in the science popularization content represented by the node.

[0186] Difficulty adaptability score D(N): According to the user's learning progress, evaluate whether the difficulty of node N matches its current knowledge level. If the user has mastered the relevant basic knowledge, recommending higher-level content will help them deepen their learning; if the user is in the early stage, recommend content with moderate or slightly lower difficulty than their level to stimulate their interest and confidence in learning. The score is usually between 0 and 1, and the higher the value, the higher the adaptability.

[0187] Hot correlation degree R(N): Measure the relevance of node N to the current hot event. If the node is associated with a hot event, its recommendation value is usually high, because users are often more interested in fresh and popular science topics. This index also takes values between 0 and 1, and the higher the value, the closer the node is associated with the hot event.

[0188] Weight coefficients α, β, γ are used to adjust the relative importance of each index in the calculation of the total matching degree. According to the actual application scene and user demand, these coefficients can be dynamically adjusted to optimize the recommendation effect.

[0189] Four: personalized recommendation generation and display

[0190] Matching degree sorting and Top-K selection: Calculate the matching degree between the user and all nodes in the graph, and sort them from high to low according to the matching degree value. Select the Top-K nodes in the front row, and the corresponding science popularization articles or article list will become the recommended content.

[0191] Personalized recommendation display: In the popular science content interface, display the recommended results in the form of list, card, carousel, etc. The recommendation list should include node title, brief introduction, difficulty level, hot spot identification, etc. to help users quickly understand the summary of the recommended content. The recommended results should be updated regularly or in real time to respond to changes in user behavior (such as newly generated interests, completed learning tasks, etc.) and updates to the knowledge graph (such as adding new nodes, updating node attributes, etc.).

[0192] In summary, the present embodiment realizes personalized recommendation of popular science content by collecting user interest behavior data, analyzing and constructing user portrait models, multi-dimensional labeling of popular science knowledge graph nodes, and designing a recommendation algorithm to calculate the matching degree between users and nodes. This function not only improves the efficiency of users finding interesting content in a large amount of popular science resources, but also promotes the orderly construction and expansion of users' knowledge system, and enhances the user stickiness and educational value of the popular science platform.

[0193] In the present embodiment, as Figure 5 The above-mentioned method for displaying popular science content based on a knowledge graph further includes the following steps to realize community interaction and knowledge Q&A of popular science content:

[0194] Establish a user interaction module to enable users to perform social interaction operations under the popular science content interface and record relevant interaction data;

[0195] Integrate an online Q&A system, users can ask questions about popular science articles, the system automatically identifies the relevance between the question and the knowledge graph node, assigns appropriate tags to the question and pushes it to community members or experts with relevant professional knowledge background;

[0196] Receive and display the answers of community members or experts to user questions, the answer content can be directly linked to the relevant graph node, analyze the user's submitted questions, answers and comments, extract keywords and sentiment information to enrich the user portrait, and at the same time, optimize the search and recommendation of community content;

[0197] Periodically organize high-value Q&A content in the community and highlight it in the popular science content interface through system recommendation or manual editing.

[0198] The present embodiment realizes the interaction and knowledge Q&A function of the popular science content community by introducing a user interaction module, integrating an online Q&A system, and optimizing content management and recommendation mechanism, which produces the following significant technical effects:

[0199] Improve user engagement: Social interaction operations and Q&A functions encourage users to actively participate in the discussion and problem solving of popular science content, create a knowledge co-creation atmosphere, and enhance users' sense of belonging and stickiness to the platform.

[0200] Deepen knowledge learning: the Q&A session directly links to the knowledge graph nodes, helping users quickly locate relevant knowledge points from the answers for in-depth learning, improving the effectiveness of science popularization education.

[0201] Precise user profiling: through the analysis of user questions, answers, and comments, the user profile is continuously updated, making personalized recommendations more in line with individual knowledge needs and emotional states, improving the accuracy of the recommendation system.

[0202] Optimize content retrieval and recommendation: Q&A data provides more diverse and real-time indexing information for content retrieval, improving the relevance of search results. At the same time, Q&A data is also used to optimize the recommendation algorithm to ensure that recommended content keeps pace with the times and meets users' changing knowledge needs.

[0203] Knowledge sedimentation and inheritance: regularly sorting and highlighting high-value Q&As promotes the accumulation and dissemination of high-quality knowledge, improves the knowledge content and value of the entire science popularization community, and further stimulates users' learning enthusiasm and exploration desire.

[0204] In this embodiment, as Figure 6 The above-mentioned method for constructing and displaying science popularization content based on a knowledge graph further includes the following steps to achieve personalized content pushing and learning path planning:

[0205] Establish a user profiling system to collect and analyze multi-dimensional data such as user browsing history, interaction behavior, interest preferences, and knowledge level, forming a user individualized feature model;

[0206] Combine user profiling and knowledge graph structure to design a pushing algorithm based on content similarity, knowledge dependency relationship, and user interest matching degree, and recommend science popularization articles most relevant to the user's current learning status and interests in real time;

[0207] Provide user-defined subscription function, allowing users to select specific topics, fields, or expert focus lists, and the system will push customized science popularization content accordingly;

[0208] Build a learning path planning engine to intelligently generate personalized learning path diagrams based on the user's mastered knowledge nodes, learning goals, and logical associations in the knowledge graph, including recommended learning order, key knowledge points, and supplementary learning resources;

[0209] Implement learning progress tracking and feedback mechanism, record user reading and understanding of pushed content, and obtain user satisfaction and learning effect feedback through interactive testing and questionnaire survey, dynamically adjust pushing strategy and learning path planning.

[0210] In summary, the embodiment based on the atlas construction method for displaying popular science content, through knowledge structure visualization, three-dimensional rendering and interactive experience, and accurate content pushing, effectively solves the deficiencies of existing popular science means in knowledge presentation, user interaction and content relevance, realizes intuitive and visual display of popular science content, significantly improves the learning efficiency, understanding and memory effect of users, and effectively promotes the modernization process of popular science education. This method is expected to innovate the popular science communication mode, stimulate the public's enthusiasm for scientific knowledge, and help improve the scientific literacy of the whole society.

[0211] The application also provides a system for displaying popular science content based on an atlas construction, as shown in the accompanying drawings, comprising: a general knowledge structure model construction module, a popular science content interface establishment module and a popular science content display module. Figure 7

[0212] Specifically, the general knowledge structure model construction module is used to adopt a three-dimensional force guide map as a display carrier of popular science content, and to construct a popular science knowledge structure model through atlas technology, wherein the popular science knowledge structure model comprises a plurality of nodes, each node being a popular science knowledge point, and the nodes are connected through lines to represent the association relationship between the knowledge points.

[0213] Specifically, the popular science content interface establishment module is used to perform three-dimensional rendering on the popular science knowledge structure model by using three-dimensional technology, to generate a popular science content interface with intuitive and visual display effect.

[0214] Specifically, the popular science content display module is used to call out a popular science article or an article list corresponding to the clicked node for the user to review after the user clicks any node on the popular science content interface.

[0215] In addition, the general knowledge structure model construction module, the popular science content interface establishment module and the popular science content display module described above are also used to realize other functions of the above-mentioned method for displaying popular science content based on an atlas construction when executed, which will not be described here.

[0216] In addition, the application also provides a terminal device, and the method for displaying popular science content based on an atlas construction involved in the embodiment is mainly applied in the terminal device, which can be a PC, a portable computer, a mobile terminal or other devices with display and processing functions.

[0217] ​Specifically, the terminal device can include a processor (e.g., a CPU), a communication bus, a user interface, a network interface, and a memory. The communication bus is used to realize the connection communication between these components; the user interface can include a display screen (Display) and an input unit such as a keyboard (Keyboard); the network interface can optionally include a standard wired interface, a wireless interface (such as a WI-FI interface); the memory can be a high-speed RAM memory, or a stable memory (non-volatile memory) such as a disk memory, and the memory can also be a storage device independent of the aforementioned processor.

[0218] The memory stores a readable storage medium, and the readable storage medium stores a popular science content display program. The processor can call the popular science content display program stored in the memory and execute the method for displaying popular science content based on a graph according to the embodiments of the present application.

[0219] It can be understood that the readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium can be, for example but not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the above. More specific examples (a non-exhaustive list) of the computer readable storage medium include a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punched card or a concave-convex structure in a slot, and any suitable combination of the above. The computer readable storage medium used herein is not to be interpreted as a transitory signal per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission medium (e.g., optical pulses through an optical fiber cable), or electrical signals transmitted through a wire.

[0220] The computer readable program instructions described herein can be downloaded from the computer readable storage medium to the respective computing / processing device, or to an external computer or external storage device through a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, fiber-optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer readable program instructions from the network and forwards the computer readable program instructions to the computer readable storage medium for storage in the respective computing / processing device.

[0221] Computer readable program instructions for carrying out operations of the present disclosure can be assembly instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) can execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present disclosure.

[0222] Finally, it should be noted that the above-mentioned only is the preferred embodiment of the present application, and is not used to limit the present application, although the present application is described in detail with reference to the foregoing embodiments, for those skilled in the art, the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced, any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application, should be included in the protection scope of the present application.

Claims

1. A method for constructing and displaying popular science content based on graphs, characterized in that, Includes the following steps: A three-dimensional force-guided graph is used as a display carrier for popular science content. A popular science knowledge structure model is constructed through graph technology. The popular science knowledge structure model contains multiple nodes, each node is a popular science knowledge point, and the relationship between knowledge points is represented by lines between nodes. The popular science knowledge structure model is rendered in three dimensions using 3D technology to generate a popular science content interface with an intuitive and vivid display effect; After clicking on any node on the science popularization content interface, users can access the science popularization articles or article lists corresponding to the clicked node for browsing. A three-dimensional force-guided graph is used as the display medium for popular science content. A knowledge structure model is constructed using graphing technology, including: assigning corresponding attribute values ​​(importance_i, relations_i) to each node based on the importance and logical relationships between the popular science knowledge points; calculating the position coordinates of each node in three-dimensional space based on its attribute values ​​using Formula 1; and representing the curvature of the connecting lines by the closeness or complexity of the relationships between the nodes through visual characteristics, including the length, curvature, color, and thickness of the connecting lines. Formula 1: (x_i,y_i,z_i)=f(importance_i,relations_i); Where (x_i, y_i, z_i) are the three-dimensional coordinates of the i-th node, importance_i is the importance coefficient of the popular science knowledge point represented by the i-th node, and relations_i represents the logical relationship parameters between the i-th node and other nodes. Formula 2 was used to statistically analyze the keywords and their frequencies of popular science knowledge points. The keywords with high TF-IDF values ​​were then associated with the corresponding popular science knowledge point nodes and displayed as tags and annotations on the 3D force map. Formula 2; TF-IDF(k_i,d_j)=tf(k_i,d_j)*log(N / df(k_i)); Where TF-IDF(k_i,d_j) represents the weight of keyword k_i in document d_j, tf(k_i,d_j) is the number of times keyword k_i appears in document d_j, N is the total number of documents, and df(k_i) is the number of documents containing keyword k_i; The science popularization content interface also includes a search box, which is used to calculate the similarity between the user's query keywords and the tags of each node using Formula 3 after the user enters query keywords, and displays relevant nodes in sorted order by similarity: Formula 3; similarity(q,t)=cosine(vec(q),vec(t)); Where, similarity(q,t) represents the similarity between query keyword q and node label t, vec(q) and vec(t) are the vectorized representations of query keyword q and node label t, respectively, and cosine() is the cosine similarity calculation function; The method further includes the following steps to achieve personalized recommendations for nodes: Collect user browsing behavior data during the use of the science popularization content interface, process the collected browsing behavior data, and calculate the user's interest in each science popularization knowledge point based on Formula 4; Formula 4; interest_u(k)=α*click_u(k)+β*read_u(k)+γ*dwell_time_u(k)+δ*search_freq_u(k); Where, interest_u(k) represents user u's interest in science knowledge point k, click_u(k) is the number of times user u clicks on science knowledge point k, read_u(k) is the number of times user u reads articles related to science knowledge point k, dwell_time_u(k) is the time user u stays on the interface related to science knowledge point k, search_freq_u(k) is the frequency of user u searching for keywords related to science knowledge point k, and α, β, γ, and δ are preset weight coefficients; Based on the user interest matrix and the browsing behavior data of all users, the collaborative filtering algorithm is used to calculate the predicted interest of users in popular science knowledge points that they have not yet encountered. Science knowledge points with predicted interest levels higher than a preset threshold are marked as potential interest points for users and displayed as icons in the science content interface or recommended to users through push notifications. The method also includes the following steps to achieve personalized recommendations of popular science content: Collect user interest and behavior data, analyze the user interest and behavior data using machine learning algorithms, extract user interest topics, attention levels and learning progress characteristics in the field of popular science, and build a user profile model. The nodes in the science knowledge graph are labeled in multiple dimensions, covering the difficulty level of the knowledge points, the subject area they belong to, the applicable age group, and related hot events. The recommendation algorithm is designed based on the user profile model and node annotation information. Formula 6 is used to calculate the matching degree between the user and each node in the graph. Formula 6: M(U,N)=α*P(N|U)+β*D(N)+γ*R(N); Where M(U,N) represents the matching degree between user U and graph node N, P(N|U) is the probability of predicting that the user is interested in node N based on user behavior data, D(N) is the fitness score of the knowledge point difficulty of node N and the user's learning progress, R(N) is the degree to which node N is associated with the current hot events, and α, β, γ are weight coefficients. Based on the matching degree ranking results, the top-K nodes are selected as popular science articles or article lists as recommended content. Personalized recommendation results are displayed in the popular science content interface, and the recommendation list is dynamically updated to respond to changes in user behavior and knowledge graph updates.

2. The method for constructing and displaying popular science content based on a graph, as described in claim 1, is characterized in that... The method also includes the following steps to implement knowledge-based Q&A for popular science content: Design and build a knowledge question answering system. The knowledge question answering system receives questions from users related to popular science content. The question input is parsed by a natural language processing module, extracts key information, and is converted into standardized query statements. Formula 5 is used to calculate the semantic similarity between the standardized query statement and the tags of each node in the popular science knowledge structure model. Formula 5: sim(Q,T)=Σw_i*cos(vec(q_i),vec(t_i)),i=1,2,…,n; Where sim(Q,T) represents the semantic similarity between query statement Q and node label T, Q consists of n words q_1,q_2,...,q_n, T consists of m words t_1,t_2,...,t_m, w_i is the weight of word q_i in query statement Q, vec(q_i) and vec(t_i) are the vectorized representations of words q_i and t_i respectively, and cos() is the cosine similarity calculation function; Nodes with semantic similarity higher than a preset threshold are marked as relevant to the question and sorted in descending order of similarity; Based on the sorting results, the system extracts answer information from the science articles or article lists corresponding to the relevant nodes, generates answer text, returns the answer text to the user, and highlights the nodes related to the question on the science content interface.

3. The method for constructing and displaying popular science content based on a graph, as described in claim 1, is characterized in that... The method also includes the following steps to enable community interaction and Q&A for popular science content: Establish a user interaction module to enable users to perform social interaction operations within the science popularization content interface and record relevant interaction data; The system integrates an online Q&A system, allowing users to ask questions about popular science articles. The system automatically identifies the correlation between questions and knowledge graph nodes, assigns appropriate tags to questions, and pushes them to community members or experts with relevant professional knowledge backgrounds. It receives and displays answers from community members or experts to user questions. The answers can be directly linked to relevant graph nodes. It analyzes user-submitted questions, answers, and comments to extract keywords and sentiment information to enrich user profiles, and also to optimize the retrieval and recommendation of community content. We regularly compile high-value Q&A content from the community and highlight it on the science popularization content interface through system recommendations or manual editing.

4. The method for constructing and displaying popular science content based on a graph, as described in claim 1, is characterized in that... The method also includes the following steps to achieve personalized delivery of popular science content and planning of learning paths: Establish a user profiling system to collect and analyze multi-dimensional data on users' browsing history, interaction behavior, interests and preferences, and knowledge level, and form a personalized user characteristic model; By combining user profiles and knowledge graph structures, a push algorithm based on content similarity, knowledge dependency, and user interest matching is designed to recommend popular science articles most relevant to the user's current learning status and interests in real time. It offers a user-customizable subscription feature, allowing users to select a watchlist of specific topics, fields, or experts, and the system will then regularly push customized science popularization content accordingly. A learning path planning engine is built to intelligently generate personalized learning path diagrams based on the logical connections between the user's existing knowledge nodes, learning objectives, and knowledge graphs. These diagrams include recommended learning order, key knowledge points, and supplementary learning resources. Implement a learning progress tracking and feedback mechanism to record users' reading and comprehension of the pushed content, obtain user satisfaction and learning effect feedback through interactive tests and questionnaires, and dynamically adjust the push strategy and learning path planning.

5. A system for constructing and displaying popular science content based on graphs, characterized in that, include: The popular science knowledge structure model construction module is used to construct a popular science knowledge structure model by using a three-dimensional force-guided diagram as a display carrier for popular science content and by using graph technology. The popular science knowledge structure model contains multiple nodes, each node is a popular science knowledge point, and the relationship between knowledge points is represented by lines between nodes. The popular science content interface creation module is used to use 3D technology to render the popular science knowledge structure model in three dimensions, generating a popular science content interface with an intuitive and vivid display effect. The popular science content display module allows users to access popular science articles or a list of articles corresponding to the clicked node by clicking any node on the popular science content interface. The knowledge structure model building module is also used for: Each node is assigned a corresponding attribute value (importance_i, relations_i) based on the importance of the scientific knowledge points and their logical relationships. Based on the attribute value of each node, the position coordinates of each node in three-dimensional space are calculated using Formula 1. The curvature of the connecting lines reflects the tightness or complexity of the relationship between the nodes, which is represented by the visual characteristics of the connecting lines, including their length, curvature, color, and thickness. Formula 1: (x_i,y_i,z_i)=f(importance_i,relations_i); Where (x_i, y_i, z_i) are the three-dimensional coordinates of the i-th node, importance_i is the importance coefficient of the popular science knowledge point represented by the i-th node, and relations_i represents the logical relationship parameters between the i-th node and other nodes. Formula 2 was used to statistically analyze the keywords and their frequencies of popular science knowledge points. The keywords with high TF-IDF values ​​were then associated with the corresponding popular science knowledge point nodes and displayed as tags and annotations on the 3D force map. Formula 2; TF-IDF(k_i,d_j)=tf(k_i,d_j)*log(N / df(k_i)); Where TF-IDF(k_i,d_j) represents the weight of keyword k_i in document d_j, tf(k_i,d_j) is the number of times keyword k_i appears in document d_j, N is the total number of documents, and df(k_i) is the number of documents containing keyword k_i; The science popularization content interface also includes a search box, which is used to calculate the similarity between the user's query keywords and the tags of each node using Formula 3 after the user enters query keywords, and displays relevant nodes in sorted order by similarity: Formula 3; similarity(q,t)=cosine(vec(q),vec(t)); Where, similarity(q,t) represents the similarity between query keyword q and node label t, vec(q) and vec(t) are the vectorized representations of query keyword q and node label t, respectively, and cosine() is the cosine similarity calculation function; The system is also used to implement personalized recommendations for nodes: Collect user browsing behavior data during the use of the science popularization content interface, process the collected browsing behavior data, and calculate the user's interest in each science popularization knowledge point based on Formula 4; Formula 4; interest_u(k)=α*click_u(k)+β*read_u(k)+γ*dwell_time_u(k)+δ*search_freq_u(k); Where, interest_u(k) represents user u's interest in science knowledge point k, click_u(k) is the number of times user u clicks on science knowledge point k, read_u(k) is the number of times user u reads articles related to science knowledge point k, dwell_time_u(k) is the time user u stays on the interface related to science knowledge point k, search_freq_u(k) is the frequency of user u searching for keywords related to science knowledge point k, and α, β, γ, and δ are preset weight coefficients; Based on the user interest matrix and the browsing behavior data of all users, the collaborative filtering algorithm is used to calculate the predicted interest of users in popular science knowledge points that they have not yet encountered. Science knowledge points with predicted interest levels higher than a preset threshold are marked as potential interest points for users and displayed as icons in the science content interface or recommended to users through push notifications. The system is also used to enable personalized recommendations of science popularization content: Collect user interest and behavior data, analyze the user interest and behavior data using machine learning algorithms, extract user interest topics, attention levels and learning progress characteristics in the field of popular science, and build a user profile model. The nodes in the science knowledge graph are labeled in multiple dimensions, covering the difficulty level of the knowledge points, the subject area they belong to, the applicable age group, and related hot events. The recommendation algorithm is designed based on the user profile model and node annotation information. Formula 6 is used to calculate the matching degree between the user and each node in the graph. Formula 6: M(U,N)=α*P(N|U)+β*D(N)+γ*R(N); Where M(U,N) represents the matching degree between user U and graph node N, P(N|U) is the probability of predicting that the user is interested in node N based on user behavior data, D(N) is the fitness score of the knowledge point difficulty of node N and the user's learning progress, R(N) is the degree to which node N is associated with the current hot events, and α, β, γ are weight coefficients. Based on the matching degree ranking results, the top-K nodes are selected as popular science articles or article lists as recommended content. Personalized recommendation results are displayed in the popular science content interface, and the recommendation list is dynamically updated to respond to changes in user behavior and knowledge graph updates.

Images