Science popularization content generation method based on joint analysis of emotional resistance and knowledge gap

Abstract

The invention belongs to the technical field of popular science content generation, and particularly discloses a popular science content generation method based on emotional resistance and knowledge gap conjoint analysis, which comprises the following steps: acquiring a comment text published by a user from a social media platform or a museum message board in real time, and extracting keywords in the comment text; using a pre-constructed mixed emotion model to quantify an emotion resistance value of the user comment, determining an emotion state of the user, analyzing the user comment through pre-constructed knowledge graph traversal and dependency syntax, detecting missing concepts and wrong associations in the user comment, and determining a knowledge gap type of the user; and associating the emotional resistance value with the knowledge gap type through dynamic mapping of a dynamic strategy matrix, matching a proper content template from a multi-modal resource library according to an association result, and rendering the content template into final output content to be pushed to a user. According to the invention, the acceptance of the user to the complex scientific probability can be effectively improved.

Description

Technical Field

[0001] The present application belongs to the technical field of popular science content generation, and more specifically, relates to a popular science content generation method based on the joint analysis of emotional resistance and knowledge gaps. Background Art

[0002] In the technology of scientific knowledge dissemination, especially in the process of disseminating complex scientific knowledge such as dinosaur fossils, cognitive misunderstandings (such as "pterosaurs belong to the order Dinosauria") are often easily solidified. This misunderstanding stems from the lack of systematic popularization of science and in-depth academic research.

[0003] However, current popular science technologies usually rely on knowledge level grading (such as user testing to divide knowledge into beginner / intermediate / advanced levels) or interest tag recommendations (such as collaborative filtering based on browsing history). Although these technologies can achieve basic content adaptation, they ignore the interference effect of emotional factors on knowledge reception, causing users to have cognitive resistance in highly controversial scenarios (such as ethical discussions on commercial fossil excavation), resulting in low user acceptance.

[0004] Therefore, how to improve users' acceptance of complex scientific concepts is a difficult problem in current research. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the purpose of this application is to provide a method for generating popular science content based on the joint analysis of emotional resistance and knowledge gaps, which can effectively improve users' acceptance of complex scientific probabilities.

[0006] To achieve the above objectives, in a first aspect, this application provides a method for generating popular science content based on a joint analysis of emotional resistance and knowledge gaps, comprising the following steps: S10, obtaining in real time the comment texts posted by users on social media platforms or museum message boards, and extracting keywords from the comment texts; S20 uses a pre-built hybrid sentiment model to quantify the emotional resistance value of user comments and determine the user's emotional state. It also analyzes user comments through pre-built knowledge graph traversal and dependency syntax to detect missing concepts and incorrect associations in user comments and determine the type of user knowledge gap. S30, dynamically mapping the emotional resistance value and the knowledge gap type through a dynamic strategy matrix, matching a suitable content template from a multimodal resource library based on the association result, and then rendering the content template into final output content and pushing it to the user.

[0007] Compared to existing technologies, the proposed solution innovatively integrates sentiment resistance analysis with knowledge gap detection to construct a dynamic strategy matrix to drive multimodal content generation, significantly improving the dissemination efficiency and user acceptance of complex scientific knowledge. Addressing the challenges of traditional science popularization systems, which often overlook sentiment interference, employ a monotonous content format, and suffer from inefficient error correction, this breakthrough solution proposes the following: Using a domain sentiment dictionary and deep learning models to quantify user sentiment resistance, identify cognitive errors (e.g., "pterosaur misclassification") in real time, and integrate knowledge graphs to construct a dynamic strategy matrix to prioritize relevant content (e.g., legal case videos and 3D skeleton comparison animations). Furthermore, through lightweight knowledge graphs and multimodal rendering technology, accurate error correction and optimized user engagement are achieved in low-resource scenarios. Compared to traditional approaches, this solution demonstrates significant advantages in sentiment adaptation, cognitive error correction efficiency, and cross-scenario adaptability. It effectively addresses user resistance, fragmented knowledge dissemination, and cross-platform compatibility challenges in highly controversial scenarios, providing a scalable and highly interactive solution for museums, educational institutions, and science popularization platforms, with significant application value in science communication, cultural heritage preservation, and public education.

[0008] As a further preferred embodiment, in step S10, the comment texts posted by users from social media platforms or museum message boards are obtained in real time through an API interface or crawler technology.

[0009] As a further preferred embodiment, in step S10, the step of extracting keywords from the comment text is specifically as follows: Remove irrelevant information from the comment text, including advertisements and URLs; Use Jieba word segmentation tool to extract keywords.

[0010] As a further preference, in step S20, the hybrid sentiment model is constructed by using a domain dictionary and a fine-tuned BERT model.

[0011] As a further preferred embodiment, the BERT model is fine-tuned using the Transformers library of Hugging Face; The mixed sentiment model is used to quantify the emotional resistance value ER of user reviews. The formula is: ER=0.3×S+0.7×P Where S is the dictionary score, which is based on 200+ domain words; P is the sentiment probability predicted by the fine-tuned BERT model.

[0012] As a further preference, in step S20, the knowledge graph is constructed using Neo4j.

[0013] As further preferred, in step S20, the emotional state includes anger, suspicion, neutrality, curiosity, excitement and sadness.

[0014] As a further preference, in step S30, the output content includes video, 3D animation or text, and the multimodal resource library includes a video library, a 3D model library and a legal case library.

[0015] As a further preference, in step S30, the dynamic strategy matrix is: when the emotional resistance value is lower than the set value and the knowledge gap type is concept missing, the legal case video is pushed to the user; otherwise, a 3D animation or text is pushed to the user.

[0016] In a second aspect, the present application provides a system for implementing any of the above-mentioned methods for popularizing science content based on joint analysis of emotional resistance and knowledge gaps, comprising a data storage layer, a processing logic layer, an algorithm model layer, and an output interface layer; The data storage layer uses relational databases, object storage, and graph databases to achieve efficient storage and management of multi-source heterogeneous data. The processing logic layer uses sentiment analysis and knowledge detection services to deeply process user comments and accurately identify emotional resistance and knowledge gaps; The algorithm model layer uses BERT fine-tuning technology for sentiment analysis and combines it with a dynamic strategy matrix to achieve intelligent optimization of content generation strategies; The output interface layer presents the processing results in a multimodal form that adapts to users' emotions and knowledge needs through dynamic content generation, interactive feedback and Web front-end rendering modules, thereby improving the acceptance rate and dissemination effect of popular science content.

[0017] It can be understood that the beneficial effects of the second aspect mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flowchart of a method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps provided in an embodiment of the present application; Figure 2 This is an architecture diagram of a popular science content generation system based on joint analysis of emotional resistance and knowledge gaps provided by a specific embodiment of the present application; Figure 3 This is a flowchart of a method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps provided in a specific embodiment of the present application; Figure 4 This is a structural diagram of a popular science content generation system based on joint analysis of emotional resistance and knowledge gaps provided in a specific embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of this application.

[0020] After research, this application found that traditional popular science technology has obvious shortcomings in sentiment analysis and content generation, as follows: Traditional popular science methods and systems rely heavily on knowledge graphs for content recommendations, but fail to fully integrate sentiment analysis technology. This results in low user acceptance in highly controversial or emotionally sensitive scenarios. In addition, the generation strategy for popular science content is relatively fixed and lacks the ability to dynamically adjust based on the user's real-time emotional state. For example, when the user's emotions fluctuate, the content format cannot be flexibly switched, affecting the user experience. At the same time, traditional error correction methods only provide standard definitions and lack emotionally adapted content formats, such as case videos or animations. This results in low error correction efficiency and insufficient user understanding and acceptance of complex scientific concepts. These problems limit the personalized service capabilities of the popular science system and also affect user engagement and learning outcomes.

[0021] To address the above-mentioned issues, especially the solidification of cognitive misunderstandings (such as "pterosaurs belong to the order Dinosauria") and emotional polarization in the dissemination of complex scientific knowledge such as dinosaur fossils, this application proposes a dynamic generation method based on the joint analysis of emotional resistance (ER) and knowledge gap (KG): quantifying the intensity of user emotions through a hybrid sentiment model (domain dictionary + BERT fine-tuning); constructing a knowledge graph, and detecting concept omissions (such as "mineralization is not mentioned") and incorrect associations (such as "wrong evolutionary path of pterosaurs-dinosaurs") through knowledge graph traversal technology; constructing an ER-KG dynamic strategy matrix (high-resistance scenarios prioritize pushing authoritative legal case videos, and low-resistance scenarios generate 3D bone comparison animations), and generating multimodal content based on the ER-KG mapping.

[0022] This application realizes the systematic integration of emotional adaptation and cognitive correction in complex science communication for the first time, which can significantly improve error correction efficiency, increase user engagement, break through the limitations of emotion-knowledge separation of traditional technologies, and realize the systematic integration of dynamic adaptation and precise correction in complex science communication.

[0023] Figure 1This is a flow chart of a method for generating popular science content based on a joint analysis of emotional resistance and knowledge gaps provided in an embodiment of the present application. This method is used for popularizing complex scientific knowledge such as dinosaur fossils, and specifically includes steps S10 to S30, which are described in detail as follows: Step S10: obtaining in real time the comment text posted by the user from the social media platform or the museum message board, and extracting keywords from the comment text.

[0024] In this embodiment, the user's comment text can be obtained in real time through an API interface or crawler technology.

[0025] Step S20 uses a pre-built hybrid sentiment model to quantify the emotional resistance of user comments and determine the user's emotional state (anger, suspicion, neutrality, curiosity, excitement, sadness). The pre-built knowledge graph traversal and dependency syntax are then used to analyze the user comments, detecting missing concepts and incorrect associations in the comments and determining the type of knowledge gap.

[0026] In this embodiment, the hybrid sentiment model is constructed by using a domain dictionary and a fine-tuned BERT model. The knowledge graph can be constructed using Neo4j.

[0027] Specifically, the BERT model is fine-tuned using Hugging Face’s Transformers library.

[0028] In step S30, the emotional resistance value and the knowledge gap type are dynamically mapped by the dynamic strategy matrix, and a suitable content template is matched from the multimodal resource library according to the association result. The content template is then rendered into the final output content and pushed to the user.

[0029] In this embodiment, the output content may include video, 3D animation or text. The multimodal resource library may include a video library, a 3D model library and a legal case library.

[0030] Compared to existing technologies, the proposed solution innovatively integrates sentiment resistance analysis with knowledge gap detection to construct a dynamic strategy matrix to drive multimodal content generation, significantly improving the dissemination efficiency and user acceptance of complex scientific knowledge. Addressing the challenges of traditional science popularization systems, which often overlook sentiment interference, employ a monotonous content format, and suffer from inefficient error correction, this breakthrough solution proposes the following: Using a domain sentiment dictionary and deep learning models to quantify user sentiment resistance, identify cognitive errors (e.g., "pterosaur misclassification") in real time, and integrate knowledge graphs to construct a dynamic strategy matrix to prioritize relevant content (e.g., legal case videos and 3D skeleton comparison animations). Furthermore, through lightweight knowledge graphs and multimodal rendering technology, accurate error correction and optimized user engagement are achieved in low-resource scenarios. Compared to traditional approaches, this solution demonstrates significant advantages in sentiment adaptation, cognitive error correction efficiency, and cross-scenario adaptability. It effectively addresses user resistance, fragmented knowledge dissemination, and cross-platform compatibility challenges in highly controversial scenarios, providing a scalable and highly interactive solution for museums, educational institutions, and science popularization platforms, with significant application value in science communication, cultural heritage preservation, and public education.

[0031] The present application is described in detail below based on specific embodiments.

[0032] Figure 2 This is the architecture diagram of the popular science content generation system based on the joint analysis of emotional resistance and knowledge gaps provided by this specific embodiment. Figure 2 As shown, the system architecture proposed in this embodiment includes the following three parts: (1) Input layer data collection: User review text collection: We collect user review text from social media platforms (such as Weibo and Zhihu) and museum message boards through APIs or crawler technology. The collected text data is saved in CSV file format using UTF-8 encoding to ensure data integrity and readability.

[0033] Establishing an authoritative knowledge base: This database provides structured storage of dinosaur taxonomy (e.g., Pterosaurs, Theropods), the fossil formation process (death - burial - mineralization - exposure), and conservation laws (e.g., the Regulations on the Protection of Paleontological Fossils). Data sources include authoritative textbooks (e.g., Paleontology) and academic papers (e.g., Nature abstracts), ensuring accuracy and authority.

[0034] Establish a strategy effect library: This records historical A / B test results, including indicators such as content ID, audience type, reading completion rate, and sharing rate. This is used for subsequent strategy optimization and effect evaluation to ensure that the system can make intelligent adjustments based on historical data.

[0035] (2) Data processing at the processing layer: (a) Sentimental resistance analysis module: Text cleaning: Remove irrelevant information such as advertisements and URLs from comments, and use Jieba word segmentation tools to extract keywords (such as "looting" and "magical").

[0036] Calculation of emotional resistance (ER): The emotional resistance (ER) is calculated using the mixed model, using the formula: ER=0.3×S+0.7×P Among them, S is the dictionary score, which is based on 200+ domain words; P is the sentiment probability predicted by the fine-tuned BERT model.

[0037] Dictionary score: Based on 200+ domain words (e.g., “looting” has a strength of 0.95, “fossil” has a neutrality of 0.3).

[0038] BERT model: A fine-tuned Chinese BERT-base model that outputs six emotion labels (angry, suspicious, neutral, curious, excited, and sad).

[0039] (b) Knowledge gap detection module: Knowledge graph construction: Use Neo4j to build a knowledge graph containing nodes (such as dinosaurs, fossils, and regulations) and relationships (such as belongs to, leads to, and steps).

[0040] Dependency parsing: Perform dependency parsing on user comments to detect missing concepts (e.g., “mineralization is not mentioned”) and incorrect associations (e.g., “pterosaur → dinosaur”).

[0041] (3) Output layer content generation: Dynamic Content Generation Engine: Based on the ER-KG strategy matrix, it draws on multimodal resource libraries (video library, 3D model library, legal case library) to generate adaptive content. Content formats include video, 3D animation, and text, dynamically selecting the optimal content format based on the user's emotional state and knowledge gap type.

[0042] Figure 3 This is a flowchart of the method for generating popular science content based on the joint analysis of emotional resistance and knowledge gaps provided by this specific embodiment. Figure 3 As shown, the specific implementation process of this application is as follows: Step 1: Users post comments through social media or museum message boards, and the system collects the comment text in real time.

[0043] Step 2: Remove irrelevant information such as advertisements and URLs from the comments and extract keywords.

[0044] Step 3: Calculate the emotional resistance value (ER) using the hybrid model to determine the user's emotional state.

[0045] Step 4: Detect missing concepts and incorrect associations in user reviews through knowledge graph traversal and dependency syntax analysis.

[0046] Step 5: Generate strategy tags based on ER value and knowledge gap type for subsequent content generation.

[0047] Step 6: Based on the strategy tags, match the appropriate content template from the multimodal resource library.

[0048] Step 7: Render the matched content template into the final output content including video, 3D animation or text, and push it to the user.

[0049] Figure 4 This is a structural diagram of the popular science content generation system based on the joint analysis of emotional resistance and knowledge gaps provided by this specific embodiment. Figure 4 As shown, the system structure design of this embodiment covers the data storage layer, processing logic layer, algorithm model layer and output interface layer.

[0050] Among them, the data storage layer uses relational databases, object storage and graph databases to achieve efficient storage and management of multi-source heterogeneous data.

[0051] The processing logic layer uses sentiment analysis and knowledge detection services to deeply process user comments and accurately identify emotional resistance and knowledge gaps.

[0052] The algorithm model layer uses BERT fine-tuning technology for sentiment analysis and combines it with a dynamic strategy matrix to achieve intelligent optimization of content generation strategies.

[0053] The output interface layer presents the processing results in a multimodal form that adapts to users' emotions and knowledge needs through dynamic content generation, interactive feedback and Web front-end rendering modules, significantly improving the acceptance rate and dissemination effect of popular science content.

[0054] The structure aims to build an intelligent science popularization system that integrates emotion and knowledge, achieves precise correction and dynamic adaptation, and promotes the effective dissemination of complex scientific concepts.

[0055] The key points of the technical solution provided in this embodiment are: Key Point 1: Emotion-Knowledge Joint Modeling. This example innovatively proposes emotion-knowledge joint modeling technology, linking the emotion resistance value (ER) with the knowledge gap type (KG) through ER-KG dynamic mapping. For example, when the ER is <-0.5 (high resistance) and the KG is "concept missing," the system triggers the push of a legal case video. Simultaneously, the system updates the user's emotional state every 10 seconds, optimizing content priority in real time to ensure that users receive the most appropriate content. Compared to traditional methods, this joint modeling technology can more accurately identify user needs, improving user acceptance and engagement.

[0056] Key Point 2: Lightweight Knowledge Graph. This example constructs a lightweight knowledge graph covering multiple core nodes related to dinosaur fossils (e.g., "Tyrannosaurus Rex" and "Sedimentation"), and manually annotates scientific relationships. Furthermore, common misconceptions (e.g., "Pterosaur = Dinosaur") are predefined, and counterfactual reasoning prompts are automatically generated. This knowledge graph not only clearly demonstrates the complex connections between geological concepts but also provides a solid foundation for intelligent analysis and decision support, ensuring the accuracy and usability of knowledge.

[0057] Key Point 3: Multimodal Dynamic Generation. The multimodal dynamic generation module provided in this embodiment can extract clips from a pre-recorded legal case library and overlay dynamic subtitles to generate videos. It can also render skeletal comparison animations (e.g., comparing pterosaur and dinosaur skeletons) based on the Blender model library and support WebGL interaction. This technology enables the diversification and dynamic adaptation of content formats, improving not only the visual quality of the content but also the interactivity of the user experience, providing a more intuitive and effective way to communicate complex geological concepts.

[0058] It is easy for those skilled in the art to understand that the above is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps, characterized in that: The steps include: S10, obtaining in real time the comment texts posted by users on social media platforms or museum message boards, and extracting keywords from the comment texts; S20 uses a pre-built hybrid sentiment model to quantify the emotional resistance value of user comments and determine the user's emotional state. It also analyzes user comments through pre-built knowledge graph traversal and dependency syntax to detect missing concepts and incorrect associations in user comments and determine the type of user knowledge gap. S30, dynamically mapping the emotional resistance value and the knowledge gap type through a dynamic strategy matrix, matching a suitable content template from a multimodal resource library based on the association result, and then rendering the content template into final output content and pushing it to the user.

2. The method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps according to claim 1, characterized in that: In step S10, the comment text posted by the user from the social media platform or the museum message board is obtained in real time through the API interface or crawler technology.

3. The method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps according to claim 1, characterized in that: In step S10, the step of extracting keywords from the comment text is specifically as follows: Remove irrelevant information from the comment text, including advertisements and URLs; Use Jieba word segmentation tool to extract keywords.

4. The method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps according to claim 1, characterized in that: In step S20, a hybrid sentiment model is constructed by using the domain dictionary and the fine-tuned BERT model.

5. The method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps according to claim 4, characterized in that: The BERT model was fine-tuned using Hugging Face's Transformers library. The mixed sentiment model is used to quantify the emotional resistance value ER of user reviews. The formula is: ER=0.3×S+0.7×P Where S is the dictionary score, which is based on 200+ domain words; P is the sentiment probability predicted by the fine-tuned BERT model.

6. The method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps according to claim 1, characterized in that: In step S20, the knowledge graph is constructed using Neo4j.

7. The method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps according to claim 1, characterized in that: In step S20, the emotional states include anger, suspicion, neutrality, curiosity, excitement and sadness.

8. The method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps according to claim 1, characterized in that: In step S30, the output content includes video, 3D animation or text, and the multimodal resource library includes a video library, a 3D model library and a legal case library.

9. The method for generating popular science content based on joint analysis of emotional resistance and knowledge gaps according to claim 1, characterized in that: In step S30, the dynamic strategy matrix is: when the emotional resistance value is lower than the set value and the knowledge gap type is concept missing, the legal case video is pushed to the user; otherwise, the 3D animation or text is pushed to the user.

10. A system for implementing the method for popularizing science content based on joint analysis of emotional resistance and knowledge gaps according to any one of claims 1 to 9, characterized in that: It includes data storage layer, processing logic layer, algorithm model layer and output interface layer; The data storage layer uses relational databases, object storage, and graph databases to achieve efficient storage and management of multi-source heterogeneous data. The processing logic layer uses sentiment analysis and knowledge detection services to deeply process user comments and accurately identify emotional resistance and knowledge gaps; The algorithm model layer uses BERT fine-tuning technology for sentiment analysis and combines it with a dynamic strategy matrix to achieve intelligent optimization of content generation strategies; The output interface layer presents the processing results in a multimodal form that adapts to users' emotions and knowledge needs through dynamic content generation, interactive feedback and Web front-end rendering modules, thereby improving the acceptance rate and dissemination effect of popular science content.

Images