Visual scheduling method and system for scientific and technological achievement evaluation system

Abstract

The invention relates to the field of data visualization, in particular to a visual scheduling method and system for a scientific and technological achievement evaluation system. The method comprises the following steps: identifying system access behavior data of a user; performing attention hotspot field prediction on system access behavior data of the user, performing deep personalized demand evolution, and constructing a personalized scientific and technological achievement demand graph; real-time data flow matching retrieval is carried out according to the personalized scientific and technological achievement demand graph, semantic feature analysis is carried out, and deep semantic features of each text item and visual semantic features of each image and table item are extracted; and performing dynamic clustering processing according to the deep semantic features of each text item and the visual semantic features of each image and table item, and performing multi-dimensional scientific and technological achievement item aggregation to construct a scientific and technological achievement display library. According to the invention, comprehensive, dynamic, accurate and personalized scientific and technological achievement evaluation visualization is realized.

Description

Technical Field

[0001] The present invention relates to the field of data visualization, and in particular to a visualization scheduling method and system for a scientific and technological achievement evaluation system. Background Art

[0002] With the rapid development of science and technology, the evaluation of scientific and technological achievements has become an important means to promote innovation and optimize resource allocation. In the traditional evaluation process of scientific and technological achievements, the evaluation method usually relies on the subjective judgment and qualitative analysis of experts, which makes it difficult to achieve a comprehensive and accurate evaluation. At the same time, the traditional evaluation system is slow to respond to the dynamic changes of scientific and technological achievements and cannot track emerging achievements or changing technological trends in a timely manner, which seriously affects the timeliness and accuracy of the evaluation system. With the continuous development of big data, artificial intelligence and visualization technology, traditional evaluation methods can no longer meet the needs of modern scientific and technological evaluation, and a more intelligent, dynamic and real-time scientific and technological achievement evaluation system is urgently needed.

[0003] In recent years, the application of scientific and technological achievement evaluation systems in scientific research, technological innovation and other fields has become more and more extensive. However, due to the high diversity and complexity of scientific and technological achievements, how to quickly and accurately identify the most valuable achievements among a large number of scientific and technological achievements and make effective displays and decisions has become an important challenge facing the system. Traditional evaluation methods often rely only on a single data source and fixed standards, resulting in a lack of comprehensiveness, dynamism and personalization in the evaluation results. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a visual scheduling method and system for a scientific and technological achievement evaluation system to solve at least one of the above technical problems.

[0005] To achieve the above object, the present invention provides a visual scheduling method for a scientific and technological achievement evaluation system, comprising the following steps: Step S1: Identify the user's system access behavior data; predict the hot areas of concern for the user's system access behavior data, and perform in-depth personalized demand evolution to build a personalized scientific and technological achievement demand map; Step S2: Perform real-time data stream matching retrieval according to the personalized scientific and technological achievement demand graph, and perform semantic feature analysis to extract the deep semantic features of each text item and the visual semantic features of each image and table item; Step S3: Perform dynamic clustering processing based on the deep semantic features of each text item and the visual semantic features of each image and table item, and then perform multi-dimensional scientific and technological achievement project aggregation to build a scientific and technological achievement display library; Step S4: extract multiple scientific and technological achievement samples based on the scientific and technological achievement display library, and conduct a dynamic priority comprehensive evaluation to obtain the dynamic priority of each scientific and technological achievement sample; Step S5: design a multi-level visualization framework; and perform dynamic layout management according to the dynamic priority of each scientific and technological achievement sample to build a real-time interactive visualization framework; Step S6: Perform instant dynamic scheduling processing on the real-time interactive visualization framework, and perform iterative reinforcement learning to build an intelligent visualization scheduling management model.

[0006] The present invention analyzes the user's access behavior data (such as clicks, browsing paths, dwell time, search keywords, etc.), and the system accurately identifies the user's interests and needs, and then extracts the user's hot areas of concern, which helps to customize personalized scientific and technological achievements recommendations according to the needs of different users. User needs often change with time, situation or technological progress. Through deep-level demand evolution prediction, the system can adapt to the dynamic changes of user needs and always maintain high relevance. The constructed personalized demand graph can clearly represent the characteristics of each user's scientific and technological achievement demand, and provide the system with a basis for subsequent retrieval and display decisions. Based on the demand graph, the system can obtain relevant scientific and technological achievement data in real time, and realize more immediate and accurate matching retrieval, which means that the system can quickly respond to user needs and provide the latest scientific and technological achievement information. Through deep semantic analysis technology, the system not only understands the literal meaning of the text content, but also identifies the hidden deep semantics. The intentions, innovations, etc. of scientific research articles are deeply analyzed through natural language processing (NLP) algorithms. Through visual semantic feature extraction, image and table items can understand the relationship between patterns, content and data in their images, and enhance the ability to understand visual materials. Clustering is performed based on the similarity of different data features, so that scientific and technological achievements with high relevance are automatically classified together. This dynamic clustering method enables the display library to be flexibly adjusted according to changes in demand and can identify implicit connections between scientific and technological achievements in real time. Through multi-dimensional aggregation methods (such as classification by field, application, technical difficulty, etc.), the construction of the display library is more comprehensive and detailed, providing users with more accurate retrieval and recommendation services. The clustered scientific and technological achievement projects are more efficiently organized and retrieved, reducing the interference of irrelevant information and improving the speed and accuracy of users' access to information. Dynamic priority comprehensive evaluation helps to reflect the relative importance of scientific and technological achievements in real time. Each sample is evaluated based on factors such as the innovation, practicality, and market demand of the results to determine their priority in the display. As new results emerge or user needs change, the priority will be dynamically adjusted to ensure that the most relevant and valuable results are displayed, which enhances the flexibility and adaptability of the system. Through priority evaluation, users can find the results that best match their needs in the display library more quickly, improving user experience. By designing a multi-level visualization framework, the system can display different dimensions of scientific and technological achievements in different views. For example, the hierarchical display gradually unfolds from macro (field) to micro (single project), allowing users to easily browse and gain in-depth understanding of results at different levels. By dynamically adjusting the layout, results with higher priority are always in a prominent position, ensuring that the most important information is presented to users first, which improves the efficiency of users in obtaining important information. Through the real-time interactive visualization framework, users can perform customized operations according to their needs, such as filtering results by subject, field or priority, which improves the interactivity of the system and user participation.Through real-time dynamic scheduling processing, the system can instantly adjust the content of the display framework based on user behavior, feedback, and external changes (such as scientific research progress or market demand). When a certain scientific and technological achievement is frequently viewed, the system will increase its display priority. Based on iterative reinforcement learning, the system will continuously optimize the scheduling strategy so that the display content is increasingly in line with user needs and system goals. This self-learning ability enables the system to continuously improve its performance in long-term operation. The reinforcement learning model can gradually optimize scheduling decisions by continuously learning user feedback and display effects to achieve the best visualization effect and user satisfaction.

[0007] In this specification, a visualization scheduling system for a scientific and technological achievement evaluation system is provided, which is used to execute the visualization scheduling method for a scientific and technological achievement evaluation system as described above, including: The demand evolution module is used to identify the user's system access behavior data; predict the hot areas of concern based on the user's system access behavior data, and conduct in-depth personalized demand evolution to build a personalized scientific and technological achievement demand map; The deep semantic analysis module is used to perform real-time data stream matching and retrieval based on the personalized scientific and technological achievement demand graph, and to perform semantic feature analysis to extract the deep semantic features of each text item and the visual semantic features of each image and table item; Dynamic clustering module, which is used to perform dynamic clustering processing based on the deep semantic features of each text item and the visual semantic features of each image and table item, and then aggregate multi-dimensional scientific and technological achievement items to build a scientific and technological achievement display library; The dynamic priority module is used to extract multiple scientific and technological achievement samples based on the scientific and technological achievement display library, and conduct a comprehensive evaluation of dynamic priorities to obtain the dynamic priority of each scientific and technological achievement sample; The interactive visualization module is used to design a multi-level visualization framework and to perform dynamic layout management according to the dynamic priority of each scientific and technological achievement sample to build a real-time interactive visualization framework. The real-time dynamic scheduling module is used to perform real-time dynamic scheduling processing on the real-time interactive visualization framework, and perform iterative reinforcement learning to build an intelligent visualization scheduling management model.

[0008] The present invention can make in-depth predictions on the needs of each user by systematically analyzing user behavior data, and dynamically adjust them over time, so as to provide users with truly relevant content. According to each user's demand graph, the system can accurately recommend scientific and technological achievements that meet their personal needs, improving user experience and the personalization ability of the system. The demand evolution module enables the system to continuously learn user behavior and preferences. As time accumulates, the recommended scientific and technological achievements are increasingly in line with the actual needs of users. Relevant scientific and technological achievement data are retrieved in real time based on the personalized scientific and technological achievement demand graph to ensure that the results displayed to users are highly consistent with the needs. Natural language processing (NLP) is performed on text data to analyze the deep semantics of the text, including keywords, sentiment tendencies, themes, and intentions; visual semantic analysis is performed on image and table data to extract visual patterns and information in the content (such as object recognition in images or data trends in tables). This module can perform semantic fusion across multiple data types such as text, images, and tables to ensure a comprehensive and accurate understanding of the characteristics of each scientific and technological achievement. Through deep semantic analysis, the system can go beyond the surface content, understand and extract the true meaning of text and visual information, and provide more accurate results. The system can combine the information in text, images and tables to construct a more comprehensive description of scientific and technological achievements, solving the problem that a single data modality cannot fully display information. Comprehensive analysis of deep semantics improves the relevance and accuracy of search results, ensuring that users can obtain scientific and technological achievements that are highly matched to their needs. Dynamic clustering brings related achievements together, reduces information redundancy, makes the structure of the display library clearer, and allows users to quickly find scientific and technological achievements in related fields. Clustering is dynamic, and the system adjusts clustering results in a timely manner based on new data input and user behavior to ensure that the displayed content continues to fit user needs. The multi-dimensional aggregation method enables scientific and technological achievements to be displayed from multiple angles to meet the needs and preferences of different users. The priority module ensures that the most important and relevant scientific and technological achievements are always displayed first, which improves the efficiency of the system and helps users find the achievements they care about most quickly. Users can more easily access the most valuable scientific and technological achievements at present, improving the overall user experience and satisfaction. Because the priority is adjusted dynamically, the system can respond to new scientific research trends or changes in user preferences in a timely manner, ensuring that the display of scientific and technological achievements is always connected with market demand. The multi-level visualization framework can help users understand scientific and technological achievements from different perspectives, reduce information redundancy and overload, and enhance the sense of hierarchy of display. The dynamic layout allows users to see the most relevant and important scientific and technological achievements at any time, which improves the interactivity and user participation of the display. High-priority results can be displayed in a timely manner to ensure that the system always focuses on the most valuable scientific research results. The system can respond to changes in user needs or scientific research progress in real time to ensure that the displayed content is always the most relevant. Through reinforcement learning, the system can continuously optimize its scheduling strategy to improve the display quality and user satisfaction in long-term operation.The reinforcement learning model can help the system make optimal decisions in different situations and improve the efficiency of dynamic adjustment of displayed content. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of the steps of a visual scheduling method for a scientific and technological achievement evaluation system according to the present invention; Figure 2 Detailed implementation flow chart of step S1; Figure 3 Detailed implementation flow chart of step S2; Figure 4 Detailed implementation flow chart of step S3. DETAILED DESCRIPTION

[0010] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0011] The present application example provides a visual scheduling method and system for a scientific and technological achievement evaluation system. The execution subject of the visual scheduling method and system for a scientific and technological achievement evaluation system includes but is not limited to: mechanical equipment, data processing platform, cloud server node, network upload device, etc. equipped with the system can be regarded as a general computing node of the present application, and the data processing platform includes but is not limited to: at least one of an audio and image management system, an information management system, and a cloud data management system.

[0012] See also Figures 1 to 4 The present invention provides a visualization scheduling method for a scientific and technological achievement evaluation system, and the visualization scheduling method for a scientific and technological achievement evaluation system comprises the following steps: Step S1: Identify the user's system access behavior data; predict the hot areas of concern for the user's system access behavior data, and perform in-depth personalized demand evolution to build a personalized scientific and technological achievement demand map; Step S2: Perform real-time data stream matching retrieval according to the personalized scientific and technological achievement demand graph, and perform semantic feature analysis to extract the deep semantic features of each text item and the visual semantic features of each image and table item; Step S3: Perform dynamic clustering processing based on the deep semantic features of each text item and the visual semantic features of each image and table item, and then perform multi-dimensional scientific and technological achievement project aggregation to build a scientific and technological achievement display library; Step S4: extract multiple scientific and technological achievement samples based on the scientific and technological achievement display library, and conduct a dynamic priority comprehensive evaluation to obtain the dynamic priority of each scientific and technological achievement sample; Step S5: design a multi-level visualization framework; and perform dynamic layout management according to the dynamic priority of each scientific and technological achievement sample to build a real-time interactive visualization framework; Step S6: Perform instant dynamic scheduling processing on the real-time interactive visualization framework, and perform iterative reinforcement learning to build an intelligent visualization scheduling management model.

[0013] The present invention analyzes the user's access behavior data (such as clicks, browsing paths, dwell time, search keywords, etc.), and the system accurately identifies the user's interests and needs, and then extracts the user's hot areas of concern, which helps to customize personalized scientific and technological achievements recommendations according to the needs of different users. User needs often change with time, situation or technological progress. Through deep-level demand evolution prediction, the system can adapt to the dynamic changes of user needs and always maintain high relevance. The constructed personalized demand graph can clearly represent the characteristics of each user's scientific and technological achievement demand, and provide the system with a basis for subsequent retrieval and display decisions. Based on the demand graph, the system can obtain relevant scientific and technological achievement data in real time, and realize more immediate and accurate matching retrieval, which means that the system can quickly respond to user needs and provide the latest scientific and technological achievement information. Through deep semantic analysis technology, the system not only understands the literal meaning of the text content, but also identifies the hidden deep semantics. The intentions, innovations, etc. of scientific research articles are deeply analyzed through natural language processing (NLP) algorithms. Through visual semantic feature extraction, image and table items can understand the relationship between patterns, content and data in their images, and enhance the ability to understand visual materials. Clustering is performed based on the similarity of different data features, so that scientific and technological achievements with high relevance are automatically classified together. This dynamic clustering method enables the display library to be flexibly adjusted according to changes in demand and can identify implicit connections between scientific and technological achievements in real time. Through multi-dimensional aggregation methods (such as classification by field, application, technical difficulty, etc.), the construction of the display library is more comprehensive and detailed, providing users with more accurate retrieval and recommendation services. The clustered scientific and technological achievement projects are more efficiently organized and retrieved, reducing the interference of irrelevant information and improving the speed and accuracy of users' access to information. Dynamic priority comprehensive evaluation helps to reflect the relative importance of scientific and technological achievements in real time. Each sample is evaluated based on factors such as the innovation, practicality, and market demand of the results to determine their priority in the display. As new results emerge or user needs change, the priority will be dynamically adjusted to ensure that the most relevant and valuable results are displayed, which enhances the flexibility and adaptability of the system. Through priority evaluation, users can find the results that best match their needs in the display library more quickly, improving user experience. By designing a multi-level visualization framework, the system can display different dimensions of scientific and technological achievements in different views. For example, the hierarchical display gradually unfolds from macro (field) to micro (single project), allowing users to easily browse and gain in-depth understanding of results at different levels. By dynamically adjusting the layout, results with higher priority are always in a prominent position, ensuring that the most important information is presented to users first, which improves the efficiency of users in obtaining important information. Through the real-time interactive visualization framework, users can perform customized operations according to their needs, such as filtering results by subject, field or priority, which improves the interactivity of the system and user participation.Through real-time dynamic scheduling processing, the system can instantly adjust the content of the display framework based on user behavior, feedback, and external changes (such as scientific research progress or market demand). When a certain scientific and technological achievement is frequently viewed, the system will increase its display priority. Based on iterative reinforcement learning, the system will continuously optimize the scheduling strategy so that the display content is increasingly in line with user needs and system goals. This self-learning ability enables the system to continuously improve its performance in long-term operation. The reinforcement learning model can gradually optimize scheduling decisions by continuously learning user feedback and display effects to achieve the best visualization effect and user satisfaction.

[0014] In the embodiment of the present invention, refer to Figure 1 , is a schematic flow chart of the steps of a visual scheduling method for a scientific and technological achievement evaluation system of the present invention. In this example, the steps of the visual scheduling method for a scientific and technological achievement evaluation system include: Step S1: Identify the user's system access behavior data; predict the hot areas of concern for the user's system access behavior data, and perform in-depth personalized demand evolution to build a personalized scientific and technological achievement demand map; In this embodiment, the collected user access behavior data needs to be stored in a database. It is recommended to use a relational database (such as MySQL) or a non-relational database (such as MongoDB) to facilitate subsequent data query and analysis. When storing data, it is necessary to ensure that each record contains a user ID, timestamp, page information and other relevant fields to facilitate subsequent personalized analysis. After data collection, data cleaning and preprocessing are performed. The cleaning steps include removing duplicate records, processing missing values ​​and outliers, etc. For missing values, consider using the mean filling method or deleting missing value records. The Z-score method is used to process outliers to identify and remove data points that are beyond a reasonable range. The cleaned data will provide a reliable basis for subsequent analysis. By analyzing the user's access behavior data, relevant features are extracted, such as access frequency, dwell time, and categories of visited pages. Based on these features, the user's hot areas of concern are identified. Pages with high access frequency and long dwell time usually represent the user's interests. Select a suitable hot area identification method, such as a clustering-based analysis method (such as K-Means) or a weight-based scoring method. The K-Means method divides the user's access behavior data into different clusters, thereby identifying the similarities of various fields and the user's interests. Using the identification method, cluster analysis is performed on the user's access behavior. Set the number of clusters (such as K=5), cluster the user behavior data, observe the characteristics of different clusters, and identify the hot areas of user concern. It is found that some user groups have a high degree of attention to the fields of "artificial intelligence" or "biotechnology". This process is implemented through Python's Scikit-learn library, which facilitates data processing and analysis. After identifying the hot areas of user concern, a personalized demand model is constructed, taking into account the user's historical access behavior, preferences and characteristics. The demand model uses a content-based recommendation system or collaborative filtering algorithm, combined with the user's historical behavior and hot areas, to generate personalized needs. By analyzing the user's access behavior in different time periods, the evolution trend of demand is identified. The user's interest in a certain technical field changes over time. Use time series analysis methods (such as ARIMA model) to predict future changes in user demand. Based on the personalized demand model and evolutionary analysis, a personalized scientific and technological achievement demand graph is constructed. The demand graph should include the user's interest areas, related scientific and technological achievements and their changing trends. Storing the demand graph through a graph database (such as Neo4j) facilitates subsequent query and analysis. The construction of the demand graph not only helps users obtain relevant scientific and technological achievements, but also provides a basis for the recommendation of scientific and technological achievements.

[0015] Step S2: Perform real-time data stream matching retrieval according to the personalized scientific and technological achievement demand graph, and perform semantic feature analysis to extract the deep semantic features of each text item and the visual semantic features of each image and table item; In this embodiment, a real-time data stream acquisition mechanism is built, connected to the scientific and technological achievements database through an API, and the latest scientific and technological achievements data is received in real time. These data include text items (such as research papers, reports, etc.), images (such as experimental results diagrams, schematic diagrams) and tables (such as experimental data tables, statistical data tables). Data stream processing tools such as Apache Kafka are used to ensure the real-time and stability of the data stream. According to the personalized scientific and technological achievements demand graph constructed previously, data stream matching retrieval is implemented. The demand graph contains the user's interest areas and keywords, and these information are used for matching. The keyword matching and similarity calculation methods are used, such as using TF-IDF (term frequency-inverse document frequency) to score the real-time data to ensure that the matching results are highly relevant to the user's needs. In the matching process, the data received in real time is first preprocessed, including removing stop words, punctuation marks, and performing stem extraction. Then, the text similarity algorithm (such as cosine similarity) is used to calculate the relevance of each text item to the keywords in the demand graph. For image and table items, their titles and description information are extracted for matching. Finally, the matching results are stored in a dataset to be processed, ready for subsequent feature analysis. Choose a suitable deep semantic feature extraction method. For text projects, you can use pre-trained language models (such as BERT, GPT-3, etc.) to generate vector representations of text. These models can capture contextual information in the text and extract deeper semantic features. For image projects, use convolutional neural networks (CNN) to extract visual features, while for table projects, use feature embedding technology to obtain them. Perform deep semantic feature extraction on each text project. Input the text into the selected language model to generate an embedding vector for the text. For the BERT model, by passing the text of each project into the model, extract the output of the last layer as the deep semantic features of the text. For the abstract of a research paper, the generated vector represents the theme, research methods, and conclusions of the paper in high dimension. For image projects, use pre-trained CNN models (such as ResNet or Inception) to extract visual features. Input the image into the model to obtain its feature vector, which represents the key content and style of the image. For table projects, use table feature extraction technology to combine the data and structure information in the table to generate a feature representation of the table. Embed the row and column information and key values ​​of the table into the feature vector. Store the extracted deep semantic features in a database for subsequent analysis and query. The features of each item should be associated with its basic information (such as title, author, publication time, etc.) to ensure the integrity and accessibility of the data. Use a vector database (such as Faiss or Pinecone) to manage and retrieve high-dimensional feature vectors to support subsequent similarity search and recommendation. After identifying the hot areas of user concern, build a personalized demand model, taking into account the user's historical access behavior, preferences, and characteristics.The demand model uses a content-based recommendation system or collaborative filtering algorithm, combined with the user's historical behavior and hot areas, to generate personalized demand. By analyzing the user's access behavior in different time periods, the evolution trend of demand can be identified. The user's interest in a certain technical field changes over time. Use time series analysis methods (such as ARIMA model) to predict future changes in user demand. Based on the personalized demand model and evolutionary analysis, a personalized scientific and technological achievement demand graph is constructed. The demand graph should include the user's interest areas, related scientific and technological achievements and their changing trends. The demand graph is stored in a graph database (such as Neo4j) to facilitate subsequent query and analysis. The construction of the demand graph not only helps users obtain relevant scientific and technological achievements, but also provides a basis for the recommendation of scientific and technological achievements.

[0016] Step S3: Perform dynamic clustering processing based on the deep semantic features of each text item and the visual semantic features of each image and table item, and then perform multi-dimensional scientific and technological achievement project aggregation to build a scientific and technological achievement display library; In this embodiment, before dynamic clustering processing is performed, the deep semantic features of each text item are first integrated with the visual semantic features of each image and table item, and these features are processed using a standardized or normalized method to ensure that each feature is on the same scale to avoid excessive influence of a certain feature on the clustering results. Z-score standardization is usually used to ensure that the mean of each feature is 0 and the standard deviation is 1. Select a suitable clustering algorithm for dynamic processing. Considering the multidimensional characteristics of the data, use the K-Means clustering algorithm, hierarchical clustering or DBSCAN (Density-Based Spatial Clustering of Applications with Noise). The K-Means algorithm is suitable for processing larger data sets, while DBSCAN performs well in processing noise and finding clusters of arbitrary shapes. For this project, considering the presence of noise in the data, DBSCAN is recommended. Set the parameters of the clustering algorithm, especially the minimum number of samples (min_samples) and radius (epsilon) in DBSCAN. Set the minimum number of samples to 5 and the radius to 0.5 to ensure effective clustering results. According to the distribution of data, the effects of different parameters are evaluated through visualization (such as Elbow Method or Silhouette Score) to select appropriate parameters. The integrated features are dynamically clustered using the selected clustering algorithm. Clustering is implemented through Python's Scikit-learn library. After clustering, each project is assigned to the corresponding cluster. Each cluster should contain semantically similar projects, thereby providing basic data support for subsequent display and analysis. After completing dynamic clustering, set the aggregation criteria for scientific and technological achievement projects, which include the similarity of projects within the cluster, user evaluation, dynamic priority, etc. The goal of aggregation is to group similar scientific and technological achievement projects together to form a comprehensive collection of scientific and technological achievements for user retrieval and browsing. Select a suitable aggregation method, such as feature-weighted aggregation or similarity-based aggregation. The weighted average method is used to weight the projects within the cluster according to their feature scores to ensure that the aggregation results can reflect the comprehensive characteristics of each project. By analyzing the characteristics and standards of the projects within each cluster, the aggregation of multi-dimensional scientific and technological achievement projects is achieved. For each cluster, the weighted feature values ​​of its member projects are calculated, and the aggregated scientific and technological achievement projects are generated. For text items and image items in a cluster, an aggregated item is generated by integrating their deep semantic features and visual features to facilitate subsequent access and understanding by users. The aggregated scientific and technological achievement items are stored in the scientific and technological achievement display library. The display library should contain the basic information, feature values, list of aggregated members, and relevant user feedback of each aggregated item. Use a database (such as MySQL or MongoDB) for management to ensure the integrity and retrievability of the data.

[0017] Step S4: extract multiple scientific and technological achievement samples based on the scientific and technological achievement display library, and conduct a dynamic priority comprehensive evaluation to obtain the dynamic priority of each scientific and technological achievement sample; In this embodiment, the data source of the scientific and technological achievements display library is confirmed to ensure that a variety of scientific and technological achievements samples can be extracted. These samples should cover different fields and topics for subsequent priority evaluation. Common data sources include database queries, API interfaces, or data export functions. Formulate sample selection criteria, such as the number of samples, fields, publication time, and influence. Set to extract 100 samples to ensure that at least five different scientific and technological fields are covered to improve the representativeness of the evaluation. When setting parameters, consider selecting the latest published results to ensure the timeliness of the data. According to the set standards, extract samples from the scientific and technological achievements display library. Use SQL query language to extract data from a relational database, or obtain data from a non-relational database through a RESTful API. The extracted data should include basic information for each sample, such as title, abstract, author, publication time, number of citations, user evaluation, etc. After sample extraction, data cleaning and preprocessing are performed to ensure the integrity and consistency of the sample data. Remove any duplicate records and missing values, and handle outliers (such as projects with abnormally high or low citations). Use Python's Pandas library for data cleaning to ensure the reliability of subsequent analysis. Determine the indicator system for dynamic priority evaluation and collect data for the above evaluation indicators for each sample. The gold content score is calculated through expert review or historical data; the timeliness score is calculated based on the difference between the publication time and the current date; user evaluation is directly obtained from the display library; and the number of citations is extracted from the database. Select a suitable evaluation model to integrate various indicators. Consider the weighted comprehensive scoring method and set the weight of each indicator (gold content accounts for 40%, timeliness accounts for 30%, user evaluation accounts for 20%, and number of citations accounts for 10%). These weights are obtained through expert consultation or historical data analysis. Record the calculated dynamic priority value in the database to ensure that the priority of each sample is associated with its basic information. Use databases such as MongoDB or MySQL for management to ensure the integrity and queryability of the data. Set up a dynamic priority update mechanism to facilitate the subsequent real-time evaluation of new data. When new scientific and technological achievement data enters the display library, it automatically triggers the priority recalculation to ensure that users always obtain the latest evaluation results.

[0018] Step S5: design a multi-level visualization framework; and perform dynamic layout management according to the dynamic priority of each scientific and technological achievement sample to build a real-time interactive visualization framework; In this embodiment, suitable visualization tools and techniques are selected to realize framework design. Common visualization libraries such as D3.js, Chart.js or Plotly, etc., D3.js is particularly suitable for complex multi-level visualization design due to its flexibility and powerful data binding capabilities. Use HTML5 and CSS3 to build responsive design to ensure that the framework is compatible on various devices. Use prototyping tools (such as Figma or Adobe XD) to create a preliminary prototype of the visualization framework. Invite users to review and collect feedback to optimize the design. User feedback includes the ease of use of visualization, the clarity of information, and the fluency of interaction. Before dynamic layout management, ensure that the dynamic priority value of each scientific and technological achievement sample has been calculated and stored in the database. It is recommended to standardize the dynamic priority value for subsequent layout processing. Select a suitable dynamic layout algorithm to manage the scientific and technological achievement samples in the visualization framework. Consider using force-directed graph layout or hierarchical layout. Force-directed graphs are suitable for showing the correlation between projects, while hierarchical layouts are suitable for showing the hierarchical information of projects. Force-directed graph layout is implemented based on D3.js, and dynamic layout effects are achieved by setting forces and links between nodes. Set the parameters of dynamic layout, such as the spacing between nodes, attraction and repulsion. Set the initial spacing between nodes to 100px to ensure that the items do not overlap during dynamic layout. By adjusting these parameters, optimize the aesthetics and readability of the layout. According to the dynamic priority value of each scientific and technological achievement sample, use the selected layout algorithm to manage the dynamic layout. Samples with high priority should be placed in a prominent position in the visualization frame so that users can quickly identify them. Through the real-time data binding feature of D3.js, the layout can be dynamically updated to ensure that the visualization results are reflected in real time when the data changes. After the layout is completed, design interactive functions, such as hovering the mouse to display detailed information, clicking to expand the detailed view, etc. Use the event handling mechanism of D3.js to implement these functions and enhance the user experience. Through these interactive designs, users can easily obtain specific information about scientific and technological achievements while maintaining a customized visualization view.

[0019] Step S6: Perform instant dynamic scheduling processing on the real-time interactive visualization framework, and perform iterative reinforcement learning to build an intelligent visualization scheduling management model.

[0020] In this embodiment, the dynamic scheduling goals of the real-time interactive visualization framework are clarified, including optimizing the order of data display, ensuring that users obtain the latest information, and improving the user interaction experience. The implementation of dynamic scheduling requires real-time response to user operations and data changes to ensure the relevance and timeliness of the visualization content. Establish a data flow monitoring mechanism to track data changes in the scientific and technological achievements display library in real time (such as the release of new achievements, the update of existing achievements, etc.). Use WebSocket technology to achieve real-time data transmission. When data changes occur, trigger corresponding events to notify the visualization framework to update. Select a suitable scheduling algorithm to manage the dynamic layout in the visualization framework. Consider using a priority scheduling algorithm to sort and display scientific and technological achievements according to dynamic priority values. Set scheduling parameters, such as update frequency (such as checking data changes every 5 seconds) and display quantity (such as displaying the top 10 highest priority achievements each time). Implement dynamic scheduling processing, and regularly check data changes and update the display content of the visualization framework by writing a scheduling processing function. When new achievements are stored, recalculate their dynamic priority and adjust the display order. Use JavaScript and D3.js libraries to combine this process to ensure that the user interface can reflect data changes in real time. Define the goal of the reinforcement learning model, which is to optimize the visualization scheduling strategy through user interaction data. Goals include improving user satisfaction, increasing user click-through rate, and improving the efficiency of information acquisition. The model needs to be able to continuously update and optimize the scheduling strategy based on user feedback. Define the state space and action space of reinforcement learning. The state space includes information such as the currently displayed scientific and technological achievements, user click behavior, and user stay time. The action space includes the selection and arrangement of scientific and technological achievements displayed in the visualization. Define different display strategies (such as by time, by priority, etc.) as actions. Use reinforcement learning frameworks (such as TensorFlow or PyTorch) to train the model. Train with user interaction data and continuously update the strategy. Use Q-learning or deep Q network (DQN) algorithms to optimize the scheduling strategy. The model will learn which display methods can maximize user satisfaction and interaction rate. Regularly evaluate the effect of the trained reinforcement learning model, analyze user feedback and behavior data, and judge the accuracy and effectiveness of the model. Compare the effects of different scheduling strategies through A / B testing, select the optimal strategy for deployment, and continuously iterate and optimize the model based on the evaluation results.

[0021] In this embodiment, refer to Figure 2 , is a flowchart of detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include: Step S11: Identify the user's system access behavior data; Step S12: predicting the hot areas of interest based on the user's system access behavior data to generate the user's hot areas of interest in scientific and technological achievements; Step S13: Retrieve the user's system access behavior data and identify the input keywords to extract the user's search keywords; Step S14: Calculate the time-series search frequency of the user's search keywords to obtain multiple keyword search frequency features; Step S15: mining the user access demand for the user's scientific and technological achievements focus areas based on the multiple keyword search frequency characteristics to generate user access demand characteristics; Step S16: Conduct in-depth personalized demand evolution on user access demand characteristics and construct a personalized scientific and technological achievement demand map.

[0022] In this embodiment, after obtaining the authorization of the system and the user, the source of the user access behavior data is determined, which usually includes user login records, page access logs, click stream data, etc. These data are obtained through the log files of the back-end server or user behavior tracking tools (such as Google Analytics). Embed data capture code in the system to ensure that the details of each user visit can be recorded, including access time, access page, dwell time, click events, etc. Design a suitable database structure (such as a NoSQL database) to store these behavior data for subsequent analysis. Clean the collected data, remove redundant and invalid data, and handle missing values ​​and outliers. Use data processing tools (such as Pandas library) to clean and preprocess the data to ensure the integrity and consistency of the data. Format the cleaned data into a structured format (such as CSV or database table) for subsequent analysis and mining. Ensure that the data contains key information such as user ID, access time, access page, etc. Use a clustering algorithm (such as K-means or DBSCAN) to analyze user access behavior data to identify hot areas of user concern. Select an appropriate number of clusters (such as the K value determined based on the elbow rule) and standardize the data. Extract features from the pages visited by users, such as page categories, access frequency, etc. Based on the clustering results, identify the hot areas in user access behavior. Analyze the cluster centers to determine the topics or fields represented by each cluster, and generate a list of hot areas of scientific and technological achievements that users are concerned about. Extract user search input records from system logs, which usually contain user-entered keywords, search time, and related page information. Use regular expressions or natural language processing (NLP) tools (such as NLTK or spaCy) to analyze the search records and extract keywords. Obtain the core keywords retrieved by users through word segmentation, stop word removal, etc. Store the extracted keywords in a structured format, including user ID, search time, and corresponding keywords, for subsequent analysis and mining. Use time series analysis methods to count the search frequency of each keyword in different time periods. Set time windows (such as hours, days, weeks) to facilitate trend analysis. Use the Pandas library to organize the extracted keywords and calculate the number of occurrences of each keyword in each time window. Integrate the statistical results into a keyword frequency feature matrix for subsequent analysis and mining. Each row represents a time window, each column represents a keyword, and each cell represents frequency. Select appropriate data mining methods (such as decision trees and random forests) to mine users' access needs. Identify user demand patterns by analyzing keyword frequency features. Select keyword frequency features as input variables and combine them with users' basic information (such as occupations and research fields) for model training to improve the accuracy of demand mining. Use machine learning libraries (such as Scikit-learn) to train models and evaluate model performance. Ensure the robustness and generalization ability of the model through cross-validation.Integrate user access demand features into the structure of a demand graph, including user ID, scientific and technological achievement topics of interest, and related keywords. Use graph databases (such as Neo4j) or graph processing libraries (such as NetworkX) to build a personalized scientific and technological achievement demand graph. Represent the relationships between users, keywords, and topics as nodes and edges of the graph. Use visualization tools (such as Gephi or D3.js) to display the demand graph to help researchers understand the user's scientific and technological achievement focus and demand evolution trend. The visualization results should include interactive functions for in-depth analysis.

[0023] In this embodiment, the specific steps of step S12 are: Perform user data access path analysis on the user's system access behavior data and extract the user access path; Calculate the dwell time of the project page according to the user's access path to obtain the dwell time of each project page; Perform frequency distribution statistics of scientific and technological achievement item clicks on the user's system access behavior data to obtain statistical data on item click frequency distribution; Based on the dwell time on each project page and the statistical data of the project click frequency distribution, hot areas of concern are predicted to generate users' hot areas of concern for scientific and technological achievements.

[0024] In this embodiment, the user's access behavior data is collected, including information such as user ID, access time, access page URL, page type, etc. These data are usually collected through website logs, analysis tools (such as Google Analytics) or custom tracking codes. The integrity and accuracy of data collection are crucial, especially to ensure that all users' access records are captured so that user behavior can be fully analyzed. After the data collection is completed, data cleaning and formatting are performed. The steps of data cleaning include removing duplicate records, invalid data (404 error page) and obvious outliers. Next, the timestamp format is unified to ensure that all data is represented by the same standard. URL normalization is also important to avoid the same page being regarded as multiple pages due to different URLs. The cleaned data will be used for subsequent path analysis to ensure the accuracy of the results. In the access path extraction stage, the access path of each user is constructed based on the user's access log. This step involves grouping the data by user ID and generating an access sequence for each user. This sequence records the order of pages visited by the user within a specific time, forming a complete access path. These access paths will provide a basis for analyzing user behavior and help identify user preferences and behavior patterns. Dwell time refers to the time a user stays on a page, from the time the user first visits the page to the time the user leaves the page. If there are multiple visit records for the same page, the last visit time minus the first visit time. Clearly defined dwell time calculation rules can ensure the accuracy of subsequent analysis. The calculation process of dwell time requires analyzing each user's visit records one by one. First, sort by user ID and visit time to ensure that the dwell time of each page is accurately calculated. For each page, record the time of the user's first visit and leave, and calculate the time difference to get the dwell time on the page. The data processing involved in this process requires meticulousness to ensure that the dwell time of each page is accurately recorded. After calculating the dwell time of each page, the next step is to summarize the dwell time. Calculate the average dwell time and total dwell time for each page to get an overall view of the user's dwell time on each project page. Through the statistical results, identify which pages users are more interested in, and then provide a basis for subsequent hot area prediction. Click frequency refers to the number of times a user visits a project page. To obtain the click frequency, we must first ensure that the detailed information of each user click is recorded, including the user ID and the corresponding page URL. The accuracy of this step directly affects the subsequent frequency statistics, so it is crucial to ensure the integrity of the data. By counting the number of clicks on each page, we can get the click frequency distribution of each project page. This process usually involves counting the access records of each page to form a frequency distribution table. In this table, we can clearly see which pages are visited more frequently and which pages are relatively less clicked.In order to more intuitively display the results of click frequency distribution, data visualization tools are used to generate relevant charts, such as bar charts or pie charts, which can help analysts quickly identify the popularity of each project page and provide support for subsequent decision-making. Before predicting the hot areas of attention, it is necessary to first combine the dwell time and click frequency of each project page to build a comprehensive feature data set. The characteristics of each page include average dwell time, total dwell time and click frequency, which will provide the necessary information basis for subsequent hot area prediction. Select a suitable prediction model to analyze the hot areas of attention. Use the weighted scoring method to quantify the user's attention level according to the dwell time and click frequency. By setting weights, adjust the influence of dwell time and click frequency on attention according to actual conditions. Combined with the constructed feature data set, score the project page. Define attention as the weighted sum of dwell time and click frequency. In this way, it is possible to identify the areas of scientific and technological achievements that users are concerned about, and then provide guidance for related research or product development.

[0025] In this embodiment, refer to Figure 3 , is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include: Step S21: performing real-time data stream matching and retrieval on the scientific and technological achievement evaluation system according to the personalized scientific and technological achievement demand graph, and extracting multi-dimensional scientific and technological achievement project data; Step S22: Identify abnormal noise points on the multi-dimensional scientific and technological achievement project data and mark all abnormal data noise points; Step S23: performing adaptive data cleaning on the abnormal data noise points to obtain multi-dimensional cleaning optimization project data, wherein the multi-dimensional cleaning optimization project data includes text data, images and tables; Step S24: performing deep semantic analysis on the text data to extract deep semantic features of each text item; Step S25: Perform visual semantic depth recognition on the image and table to extract the visual semantic features of each image and table item.

[0026] In this embodiment, before performing real-time data stream matching retrieval, it is first necessary to build a personalized scientific and technological achievement demand graph. This demand graph should be based on the user's historical behavior, preferences and needs, covering different scientific and technological achievement fields, keywords and related attributes. It is established through data analysis tools (such as cluster analysis or classification algorithms) to ensure that the demand graph accurately reflects the user's personalized needs. Connect to the scientific and technological achievement evaluation system through API or data stream technology (such as Apache Kafka) to obtain the latest scientific and technological achievement data in real time. These data usually include multi-dimensional information such as project title, abstract, research field, author information, etc. Ensure the stability and real-time nature of the data stream in order to respond to user needs in a timely manner. By comparing the personalized demand graph with the data obtained in real time, data matching retrieval is implemented. This process uses keyword matching, fuzzy matching and other technologies to improve the flexibility and accuracy of retrieval. The scientific and technological achievement projects that meet the user's needs are extracted and stored in the data set to be processed. Before identifying abnormal noise points in the multi-dimensional scientific and technological achievement project data, it is first necessary to evaluate the quality of the data, which includes checking the integrity, accuracy and consistency of the data. Set a certain threshold (missing value percentage exceeding 20% ​​is considered abnormal) to help identify potential abnormal data. Choose an appropriate anomaly detection algorithm, such as Isolation Forest, Z-score method, or density-based clustering (DBSCAN), which can effectively identify outliers in the data set. When using the Z-score method, calculate the Z-score of each data point, and mark it as an outlier if its absolute value exceeds 3. Mark the identified outliers to ensure that they can be accurately identified in the subsequent data cleaning and optimization process. This process requires accurate recording of the type and location of the anomaly for subsequent analysis and processing. Before data cleaning, you first need to formulate cleaning rules, which should include how to deal with missing values, outliers, and duplicate data. For missing values, choose to fill (such as using the mean or median) or delete (if the missing values ​​exceed a set proportion). For outliers, choose to replace or delete. Use programming tools (such as Python's Pandas library) to implement adaptive data cleaning. During the cleaning process, text data, image data, and table data are processed separately according to the set rules. Text data cleaning involves removing stop words, spelling correction, and format standardization; image cleaning includes denoising and format conversion; and table data needs to ensure data consistency and integrity. The cleaned multi-dimensional data should be recorded as structured data, including text, image, and table information for each item, to ensure the convenience and effectiveness of subsequent analysis. To perform deep semantic analysis on text data, first select appropriate natural language processing (NLP) tools and libraries, such as SpaCy, NLTK, or BERT, which can help extract deep semantic features in text. By performing word segmentation, part-of-speech tagging, and entity recognition on the text, the deep semantic features of each text item can be extracted.Use the BERT model to generate vector representations of text to capture contextual information in the text. For scientific and technological achievements in specific fields, further combine domain knowledge for semantic enhancement. The extracted deep semantic features should be stored in a structured manner to facilitate subsequent analysis and model training. These features will provide a basis for content understanding and association analysis of technical achievements. For deep visual semantic recognition of images and tables, first select a suitable computer vision model, such as a convolutional neural network (CNN) or a more advanced visual Transformer model, which can effectively extract visual features of images and tables. Input the image and table into the selected model to extract their visual semantic features. For images, it involves object detection, classification, and feature embedding; for tables, it is necessary to extract the structural information and content information of the table. Ensure that the model is properly trained to improve recognition accuracy. Record the extracted visual semantic features as structured data and integrate them into the multi-dimensional cleaning optimization project data. These features will provide important support for subsequent analysis and applications (such as recommendation systems or decision support).

[0027] In this embodiment, refer to Figure 4 , is a flowchart of detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include: Step S31: performing latent semantic association mining based on the deep semantic features of each text item and the visual semantic features of each image and table item to extract latent semantic association features between items; Step S32: dynamically clustering the multi-dimensional cleansing and optimization project data according to the latent semantic association features between the projects to obtain a plurality of project clusters; Step S33: defining cluster keywords for each of the multiple project clusters one by one to generate a feature label for each cluster; Step S34: Perform multi-dimensional scientific and technological achievement project aggregation based on the feature labels of each cluster to build a scientific and technological achievement display library.

[0028] In this embodiment, before performing latent semantic association mining, it is first necessary to integrate the deep semantic features of the text items with the visual semantic features of the image and table items. These features should be stored in a structured form for subsequent analysis. Text features may include word vectors, sentence vectors, etc., while image features are feature vectors extracted by CNN, and table features may be vector representations of table structure information. Select a suitable association mining method, such as latent semantic analysis (LSA), principal component analysis (PCA), or a graph-based relationship mining method. Latent semantic analysis (LSA) is particularly suitable for processing the correlation between text and image features. By constructing a co-occurrence matrix of features, the relationship between each feature is analyzed to mine potential semantic associations. The selected method is applied to extract the latent semantic association features between projects from the integrated features. This process involves dimensionality reduction of the features to extract the most representative latent features. By analyzing the similarity between the features, projects that are semantically related to each other are identified, thereby establishing potential relationships between projects. Select a suitable clustering algorithm, such as K-Means, hierarchical clustering, or DBSCAN. For multi-dimensional cleaning and optimization project data, DBSCAN performs well in processing noise and identifying clusters of arbitrary shapes, so it is more suitable. The extracted latent semantic association features are used for dynamic clustering. Ensure that the data is normalized to eliminate the dimensional differences between different features. Use the Z-score normalization method to make the feature mean 0 and the standard deviation 1 to ensure the validity of the clustering results. Apply the selected clustering algorithm for dynamic clustering. Divide the project data into multiple clusters by setting appropriate parameters (such as the minimum number of samples and radius in DBSCAN). The clustering process will classify projects with similar features into one category based on similarity, thereby identifying multiple project clusters. After obtaining multiple project clusters, analyze the project features in each cluster one by one. Statistically calculate the feature distribution of the projects in each cluster and identify the most significant features in the cluster. Choose an appropriate keyword extraction method, such as TF-IDF (Term Frequency-Inverse Document Frequency), TextRank, or LDA (Latent Dirichlet Allocation). The TF-IDF method can effectively identify keywords that are highly frequent in a specific cluster but uncommon in other clusters. Based on the selected method, extract key features from each cluster and generate feature labels. Count the text features and visual features in each cluster, and comprehensively determine the most representative keywords, so that each cluster will have one or more keyword tags that can clearly indicate the theme and characteristics of the cluster. Select a suitable aggregation method to build a scientific and technological achievement display library based on the feature tags of each cluster. Use a tag-based system or combine it with a recommendation algorithm to achieve project aggregation. Aggregate projects based on cluster feature tags. Aggregate all projects with the same or similar feature tags into a display library. Each aggregated project should contain information such as its deep semantic features, visual features, and its position in the cluster.Finally, a scientific and technological achievement display library is built, which contains multiple aggregated projects. It is ensured that the display library has a good visual interface, is convenient for users to browse and search, and can be dynamically updated and maintained according to user needs.

[0029] In this embodiment, step S4 includes the following steps: Step S41: extracting multiple scientific and technological achievement samples based on the scientific and technological achievement display library; Step S42: conducting expert review and quantification of the gold content of multiple scientific and technological achievement samples one by one, so as to generate a gold content quantification value of each scientific and technological achievement sample; Step S43: performing timeliness analysis on multiple scientific and technological achievement samples to generate timeliness characteristics of each sample; Step S44: performing application field influence analysis on multiple scientific and technological achievement samples to obtain the application field influence of each sample; Step S45: Perform a comprehensive dynamic priority assessment based on the quantitative value of the gold content of each scientific and technological achievement sample, the timeliness characteristics of each sample, and the influence of each sample in the application field, so as to obtain the dynamic priority of each scientific and technological achievement sample.

[0030] In this embodiment, before extracting the scientific and technological achievement samples, the data source and data structure of the scientific and technological achievement display library are first confirmed. These samples should cover different fields and themes to ensure the diversity and representativeness of the samples. Formulate sample selection criteria, including the number, field, research depth, and influence of samples. Set to extract 100 samples to ensure that at least five different scientific and technological fields are covered to facilitate the comprehensiveness of subsequent analysis. According to the set standards, multiple scientific and technological achievement samples are extracted from the scientific and technological achievement display library. Use random sampling or stratified sampling methods to ensure the balance of samples. The extracted samples should contain basic information of each project, such as title, abstract, author, publication time, etc., for subsequent evaluation. Before quantifying the gold content, first define the standard of "gold content". Factors to be considered include the innovation of the research, the practicality of the technology, the impact factor of the published journal, the number of citations, etc. Based on the defined standards, establish a quantitative indicator system for gold content. Innovation score (1-10 points), journal impact factor, number of citations, practicality score, organize field experts to review each scientific and technological achievement sample one by one, and score each sample according to the set quantitative indicators. Expert review is conducted through questionnaires or face-to-face to ensure the objectivity and accuracy of the review results. After collecting the review results, the quantitative value of the gold content of each sample is calculated, usually using the weighted average method. Before conducting a timeliness analysis, first define the standard of "timeliness". This includes the publication time of scientific and technological achievements, the frequency of technology updates, and industry development trends. Collect timeliness data related to each scientific and technological achievement sample, especially the publication time and the latest development of related technologies. If the latest research results of a certain technology appear shortly after the sample is published, it will reduce the timeliness of the sample. Based on the collected data, analyze the timeliness characteristics of each sample. Set a timeliness score and use a time window (achievements within the last three years have higher scores, and achievements over three years have lower scores). Record the timeliness characteristics for subsequent comprehensive evaluation. The definition of the influence of the application field should take into account factors such as the actual application of scientific and technological achievements in related fields, industry recognition, and user feedback. Collect application data related to each scientific and technological achievement sample, such as industry application cases, technology transfer, market feedback, etc., which are obtained through industry reports, market research or user surveys. Set the influence scoring standard based on the collected application data. Use a 1-10 point scoring system, combined with quantitative indicators such as breadth of application and user satisfaction, to evaluate the influence of each sample. Record the scoring results for subsequent dynamic priority evaluation. Before conducting a dynamic priority comprehensive evaluation, first integrate the gold content quantitative value, timeliness characteristics and application field influence of each scientific and technological achievement sample to form a comprehensive scoring system. These data should be stored in a structured form for subsequent analysis. Choose an appropriate comprehensive evaluation method, such as the weighted sum method or the analytic hierarchy process (AHP).In the weighted sum method, a weight is set for each indicator, and the weight value is obtained through expert consultation or historical data analysis. The dynamic priority calculation applies a comprehensive evaluation method to calculate the dynamic priority of each scientific and technological achievement sample. Priority=w1×GoldContent+w2×Timeliness+w3×Impact, where w1, w2 and w3 are the weights of gold content, timeliness and influence respectively. The calculated priority value will reflect the relative importance of each scientific and technological achievement sample at the current point in time.

[0031] In this embodiment, the specific steps of step S5 are: Step S51: designing a multi-level visualization framework; Step S52: dynamically layout and manage multiple scientific and technological achievement samples on a multi-level visualization framework according to the dynamic priority of each scientific and technological achievement sample, thereby constructing a multi-level scientific and technological achievement visualization framework; Step S53: Perform real-time interactive visualization scheduling on the multi-level scientific and technological achievement visualization framework to construct a real-time interactive visualization framework.

[0032] In this embodiment, the goal of designing a multi-level visualization framework is clearly defined. The goals include: intuitively displaying the dynamic priority of scientific and technological achievements, facilitating user interaction and in-depth exploration of the details of each achievement, and supporting information retrieval at different levels. The ultimate goal of designing a framework is to improve users' understanding and utilization efficiency of scientific and technological achievements. According to the characteristics of scientific and technological achievements and user needs, the hierarchical structure of the multi-level visualization framework is determined. It can usually be divided into the following levels: S First level: display an overall overview of scientific and technological achievements, including the dynamic priority of all projects. Second level: display basic information of each scientific and technological achievement, such as title, abstract, gold content, timeliness, etc. Third level: provide more detailed information, such as specific application cases of the project, citations, etc. Select appropriate visualization tools and technologies to implement framework design. Commonly used visualization libraries such as D3.js, Plotly or Tableau can support dynamic data display and interactive design. The JavaScript-based D3.js library is particularly suitable for complex multi-level visualization design due to its flexibility and powerful data binding capabilities. Use prototyping tools (such as Figma or Sketch) to design a preliminary visualization framework prototype and invite potential users and experts to review it. Iterate and optimize based on the collected feedback to ensure that the framework can meet user needs. Before dynamic layout management, ensure that the dynamic priority calculation results of each scientific and technological achievement sample have been completed and sorted by priority. Store this data in a structured form, such as JSON format, so that the data is easy to access and dynamically update. Select a suitable dynamic layout algorithm to achieve layout management of a multi-level visualization framework. Consider using force-directed graph layout or tree diagram layout. The former is suitable for displaying the association between projects, and the latter is suitable for hierarchical information display. According to the dynamic priority of each sample, the scientific and technological achievement samples are dynamically laid out using the selected layout algorithm. Place samples with high priority in a prominent position in the visualization framework to ensure that users can quickly identify key achievements. When implementing dynamic layout through visualization tools such as D3.js, use data binding and transition effects to enhance the user experience, making layout changes smoother and more natural. After the layout is completed, evaluate the visualization framework to ensure that information at different levels can be clearly displayed. Collect feedback through user testing and observe the user's interactive feedback and information acquisition efficiency during use. Before building a real-time interactive visualization framework, analyze users' real-time interactive needs, such as data update frequency, user query latency, etc. Ensure that the visualization framework can support dynamic updates of real-time data and user interaction. Design interactive elements, such as floating prompts, click-to-expand, and filtering functions, so that users can actively explore various levels of information about scientific and technological achievements. Implement these interactive functions through JavaScript and D3.js to ensure that users can smoothly obtain information when interacting with the visualization framework. Establish a real-time data update mechanism to ensure that the framework can reflect immediately when new scientific and technological achievement data is added or existing data is updated.Use technologies such as WebSocket to implement real-time data transmission to ensure that users can always see the latest information. After implementing the real-time interactive visualization framework, evaluate its effectiveness through user testing. Observe how users react to real-time data updates during use, as well as the ease of use of the interactive design. Further optimize the interactive design based on user feedback to ensure that users can efficiently obtain the information they need.

[0033] In this embodiment, the specific steps of step S6 are: Step S61: acquiring multi-user evaluation information of each scientific and technological achievement project in real time based on the scientific and technological achievement evaluation system; Step S62: Identify the urgency of the project based on the multi-user evaluation information, and generate an urgency value for each scientific and technological achievement project; Step S63: performing an average calculation of the technical complexity according to the multi-user evaluation information to obtain the multi-user complexity average value of each scientific and technological achievement project; Step S64: making a dynamic visualization scheduling decision based on the urgency value of each scientific and technological achievement project and the multi-user complexity average value of each scientific and technological achievement project, thereby generating a dynamic visualization scheduling strategy; Step S65: Perform instant dynamic scheduling processing on the real-time interactive visualization framework according to the dynamic visualization scheduling strategy, and perform iterative reinforcement learning to build an intelligent visualization scheduling management model.

[0034] In this embodiment, the data source of the scientific and technological achievement evaluation system is confirmed to ensure that multi-user evaluation information of each scientific and technological achievement project can be obtained in real time, including user ratings, comment content, suggestions and feedback. A mechanism for obtaining user evaluation information in real time is designed. Use the API interface to capture data from the evaluation platform (such as user feedback system, social media, etc.) regularly or in real time. To ensure the frequency and accuracy of data acquisition, it is recommended to use WebSocket technology to achieve the reception of real-time data streams. After obtaining the evaluation information, the data format needs to be standardized to ensure that each evaluation information contains necessary fields, such as user ID, project ID, score, timestamp, etc. The data is stored in JSON format to make subsequent processing more convenient. A database (such as MySQL or MongoDB) is established to store the user evaluation information obtained in real time to ensure the security and accessibility of the data. The data is backed up regularly, and data cleaning and update strategies are set to maintain data quality. Before identifying the degree of urgency, the standard of "urgency" is first defined. Factors considered include the urgency of user feedback, changes in market demand for projects, and frequency of technology updates. Based on the features extracted from user evaluations, a weighted scoring method is used to calculate the urgency value of each scientific and technological achievement project. The following indicators are set: the positivity of user feedback (such as the proportion of positive comments), the volatility of ratings (such as the standard deviation of ratings), and the market dynamics related to the project (such as industry news). By analyzing the acquired feature data, the urgency value of each project is calculated using the following formula: Urgency = w1 × Feedback Score + w2 × Rating Variance + w3 × Market Dynamics, where w1, w22, and w3 are the set weights. This calculation is implemented through data analysis tools (such as Python's Pandas library). The calculated urgency values ​​are stored in the database, and a regular update mechanism is set to facilitate subsequent analysis and decision-making. Before calculating the average technical complexity, the standard of "technical complexity" is first defined. Factors considered include the technical details of the project, the difficulty of implementation, the required resources, etc. By analyzing the user's evaluation information, features related to technical complexity are extracted. Analyze the user's comments on the technical details and the feedback on the difficulty of project implementation, etc. This information is extracted through natural language processing technology (such as sentiment analysis). Calculate the average technical complexity of each scientific and technological achievement project. Record the calculated average value of technical complexity in the database and ensure that it is updated synchronously with data such as the urgency value. Design a dynamic visual scheduling decision model, combining the urgency value and the average value of technical complexity of each scientific and technological achievement project. Use weighted scoring method or multi-objective decision analysis (such as AHP) to generate scheduling strategies. Set comprehensive evaluation indicators, including the following: urgency, technical complexity, resource requirements, user feedback, and use comprehensive evaluation indicators to generate dynamic scheduling strategies.Calculate the decision score for each project. The higher the score, the higher the priority. Design the framework of the intelligent visual scheduling management model, combined with the dynamic visual scheduling strategy, to ensure that the system can respond to external changes in real time and optimize internal resource allocation. Perform instant dynamic scheduling processing on the real-time interactive visualization framework according to the dynamic scheduling strategy. Use scheduling algorithms (such as genetic algorithms or particle swarm optimization) to optimize the scheduling process and ensure the effective use of resources. Build an iterative reinforcement learning mechanism to continuously optimize the scheduling model through the feedback of historical data and the input of new data. Use a reinforcement learning framework (such as TensorFlow or PyTorch) to train the model to improve the accuracy of scheduling decisions. Regularly evaluate the performance of the intelligent visual scheduling management model and analyze the effectiveness and efficiency of scheduling decisions. Optimize the model based on the evaluation results to ensure that the system can adapt to the rapidly changing environment of scientific and technological achievements.

[0035] In this embodiment, a visualization scheduling system for a scientific and technological achievement evaluation system is provided, which is used to execute the visualization scheduling method for a scientific and technological achievement evaluation system as described above, including: The demand evolution module is used to identify the user's system access behavior data; predict the hot areas of concern based on the user's system access behavior data, and conduct in-depth personalized demand evolution to build a personalized scientific and technological achievement demand map; The deep semantic analysis module is used to perform real-time data stream matching and retrieval based on the personalized scientific and technological achievement demand graph, and to perform semantic feature analysis to extract the deep semantic features of each text item and the visual semantic features of each image and table item; Dynamic clustering module, which is used to perform dynamic clustering processing based on the deep semantic features of each text item and the visual semantic features of each image and table item, and then aggregate multi-dimensional scientific and technological achievement items to build a scientific and technological achievement display library; The dynamic priority module is used to extract multiple scientific and technological achievement samples based on the scientific and technological achievement display library, and conduct a comprehensive evaluation of dynamic priorities to obtain the dynamic priority of each scientific and technological achievement sample; The interactive visualization module is used to design a multi-level visualization framework and to perform dynamic layout management according to the dynamic priority of each scientific and technological achievement sample to build a real-time interactive visualization framework. The real-time dynamic scheduling module is used to perform real-time dynamic scheduling processing on the real-time interactive visualization framework, and perform iterative reinforcement learning to build an intelligent visualization scheduling management model.

[0036] The present invention can make in-depth predictions on the needs of each user by systematically analyzing user behavior data, and dynamically adjust them over time, so as to provide users with truly relevant content. According to each user's demand graph, the system can accurately recommend scientific and technological achievements that meet their personal needs, improving user experience and the personalization ability of the system. The demand evolution module enables the system to continuously learn user behavior and preferences. As time accumulates, the recommended scientific and technological achievements are increasingly in line with the actual needs of users. Relevant scientific and technological achievement data are retrieved in real time based on the personalized scientific and technological achievement demand graph to ensure that the results displayed to users are highly consistent with the needs. Natural language processing (NLP) is performed on text data to analyze the deep semantics of the text, including keywords, sentiment tendencies, themes, and intentions; visual semantic analysis is performed on image and table data to extract visual patterns and information in the content (such as object recognition in images or data trends in tables). This module can perform semantic fusion across multiple data types such as text, images, and tables to ensure a comprehensive and accurate understanding of the characteristics of each scientific and technological achievement. Through deep semantic analysis, the system can go beyond the surface content, understand and extract the true meaning of text and visual information, and provide more accurate results. The system can combine the information in text, images and tables to construct a more comprehensive description of scientific and technological achievements, solving the problem that a single data modality cannot fully display information. Comprehensive analysis of deep semantics improves the relevance and accuracy of search results, ensuring that users can obtain scientific and technological achievements that are highly matched to their needs. Dynamic clustering brings related achievements together, reduces information redundancy, makes the structure of the display library clearer, and allows users to quickly find scientific and technological achievements in related fields. Clustering is dynamic, and the system adjusts clustering results in a timely manner based on new data input and user behavior to ensure that the displayed content continues to fit user needs. The multi-dimensional aggregation method enables scientific and technological achievements to be displayed from multiple angles to meet the needs and preferences of different users. The priority module ensures that the most important and relevant scientific and technological achievements are always displayed first, which improves the efficiency of the system and helps users find the achievements they care about most quickly. Users can more easily access the most valuable scientific and technological achievements at present, improving the overall user experience and satisfaction. Because the priority is adjusted dynamically, the system can respond to new scientific research trends or changes in user preferences in a timely manner, ensuring that the display of scientific and technological achievements is always connected with market demand. The multi-level visualization framework can help users understand scientific and technological achievements from different perspectives, reduce information redundancy and overload, and enhance the sense of hierarchy of display. The dynamic layout allows users to see the most relevant and important scientific and technological achievements at any time, which improves the interactivity and user participation of the display. High-priority results can be displayed in a timely manner to ensure that the system always focuses on the most valuable scientific research results. The system can respond to changes in user needs or scientific research progress in real time to ensure that the displayed content is always the most relevant. Through reinforcement learning, the system can continuously optimize its scheduling strategy to improve the display quality and user satisfaction in long-term operation.The reinforcement learning model can help the system make optimal decisions in different situations and improve the efficiency of dynamic adjustment of displayed content.

[0037] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0038] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A visual scheduling method for a scientific and technological achievement evaluation system, characterized in that: The following steps are involved: Step S1: Identify the user's system access behavior data; predict the hot areas of concern for the user's system access behavior data, and perform in-depth personalized demand evolution to build a personalized scientific and technological achievement demand map; Step S2: Perform real-time data stream matching retrieval according to the personalized scientific and technological achievement demand graph, and perform semantic feature analysis to extract the deep semantic features of each text item and the visual semantic features of each image and table item; Step S3: Perform dynamic clustering processing based on the deep semantic features of each text item and the visual semantic features of each image and table item, and then aggregate the multi-dimensional scientific and technological achievement items to build a scientific and technological achievement display library; Step S4: extract multiple scientific and technological achievement samples based on the scientific and technological achievement display library, and conduct a dynamic priority comprehensive evaluation to obtain the dynamic priority of each scientific and technological achievement sample; Step S5: design a multi-level visualization framework; And according to the dynamic priority of each scientific and technological achievement sample, dynamic layout management is carried out to build a real-time interactive visualization framework; Step S6: Perform instant dynamic scheduling processing on the real-time interactive visualization framework, and perform iterative reinforcement learning to build an intelligent visualization scheduling management model.

2. The visual scheduling method for a scientific and technological achievement evaluation system according to claim 1, characterized in that: The specific steps of step S1 are: Step S11: Identify the user's system access behavior data; Step S12: predicting the hot areas of interest based on the user's system access behavior data to generate the user's hot areas of interest in scientific and technological achievements; Step S13: Retrieve the user's system access behavior data and identify the input keywords to extract the user's search keywords; Step S14: Calculate the time-series search frequency of the user's search keywords to obtain multiple keyword search frequency features; Step S15: mining the user access demand for the user's scientific and technological achievements focus areas based on the multiple keyword search frequency characteristics to generate user access demand characteristics; Step S16: Conduct in-depth personalized demand evolution on user access demand characteristics and construct a personalized scientific and technological achievement demand map.

3. The visual scheduling method for a scientific and technological achievement evaluation system according to claim 2 is characterized in that: The specific steps of step S12 are: Perform user data access path analysis on the user's system access behavior data and extract the user access path; Calculate the dwell time of the project page according to the user's access path to obtain the dwell time of each project page; Perform frequency distribution statistics of scientific and technological achievement item clicks on the user's system access behavior data to obtain statistical data on item click frequency distribution; Based on the dwell time on each project page and the statistical data of the project click frequency distribution, hot areas of concern are predicted to generate users' hot areas of concern for scientific and technological achievements.

4. The visual scheduling method for a scientific and technological achievement evaluation system according to claim 1, characterized in that: The specific steps of step S2 are: Step S21: performing real-time data stream matching and retrieval on the scientific and technological achievement evaluation system according to the personalized scientific and technological achievement demand graph, and extracting multi-dimensional scientific and technological achievement project data; Step S22: Identify abnormal noise points on the multi-dimensional scientific and technological achievement project data and mark all abnormal data noise points; Step S23: performing adaptive data cleaning on the abnormal data noise points to obtain multi-dimensional cleaning optimization project data, wherein the multi-dimensional cleaning optimization project data includes text data, images and tables; Step S24: performing deep semantic analysis on the text data to extract deep semantic features of each text item; Step S25: Perform visual semantic depth recognition on the image and table to extract the visual semantic features of each image and table item.

5. The visual scheduling method for a scientific and technological achievement evaluation system according to claim 1, characterized in that: The specific steps of step S3 are: Step S31: performing latent semantic association mining based on the deep semantic features of each text item and the visual semantic features of each image and table item to extract latent semantic association features between items; Step S32: dynamically clustering the multi-dimensional cleansing and optimization project data according to the latent semantic association features between the projects to obtain a plurality of project clusters; Step S33: defining cluster keywords for each of the multiple project clusters one by one to generate a feature label for each cluster; Step S34: Perform multi-dimensional scientific and technological achievement project aggregation based on the feature labels of each cluster to build a scientific and technological achievement display library.

6. The visual scheduling method for a scientific and technological achievement evaluation system according to claim 1, characterized in that: The specific steps of step S4 are: Step S41: extracting multiple scientific and technological achievement samples based on the scientific and technological achievement display library; Step S42: conducting expert review and quantification of the gold content of multiple scientific and technological achievement samples one by one, so as to generate a gold content quantification value of each scientific and technological achievement sample; Step S43: performing timeliness analysis on multiple scientific and technological achievement samples to generate timeliness characteristics of each sample; Step S44: performing application field influence analysis on multiple scientific and technological achievement samples to obtain the application field influence of each sample; Step S45: Perform a comprehensive dynamic priority assessment based on the quantitative value of the gold content of each scientific and technological achievement sample, the timeliness characteristics of each sample, and the influence of each sample in the application field, so as to obtain the dynamic priority of each scientific and technological achievement sample.

7. The visual scheduling method for a scientific and technological achievement evaluation system according to claim 1, characterized in that: The specific steps of step S5 are: Step S51: designing a multi-level visualization framework; Step S52: dynamically layout and manage multiple scientific and technological achievement samples on a multi-level visualization framework according to the dynamic priority of each scientific and technological achievement sample, thereby constructing a multi-level scientific and technological achievement visualization framework; Step S53: Perform real-time interactive visualization scheduling on the multi-level scientific and technological achievement visualization framework to construct a real-time interactive visualization framework.

8. The visual scheduling method for a scientific and technological achievement evaluation system according to claim 1, characterized in that: The specific steps of step S6 are: Step S61: acquiring multi-user evaluation information of each scientific and technological achievement project in real time based on the scientific and technological achievement evaluation system; Step S62: Identify the urgency of the project based on the multi-user evaluation information, and generate an urgency value for each scientific and technological achievement project; Step S63: performing an average calculation of the technical complexity according to the multi-user evaluation information to obtain the multi-user complexity average value of each scientific and technological achievement project; Step S64: making a dynamic visualization scheduling decision based on the urgency value of each scientific and technological achievement project and the multi-user complexity average value of each scientific and technological achievement project, thereby generating a dynamic visualization scheduling strategy; Step S65: Perform instant dynamic scheduling processing on the real-time interactive visualization framework according to the dynamic visualization scheduling strategy, and perform iterative reinforcement learning to build an intelligent visualization scheduling management model.

9. A visual scheduling system for scientific and technological achievement evaluation system, characterized in that: The method for executing the visual scheduling method for the scientific and technological achievement evaluation system according to claim 1 comprises: The demand evolution module is used to identify the user's system access behavior data; predict the hot areas of concern based on the user's system access behavior data, and conduct in-depth personalized demand evolution to build a personalized scientific and technological achievement demand map; The deep semantic analysis module is used to perform real-time data stream matching and retrieval based on the personalized scientific and technological achievement demand graph, and to perform semantic feature analysis to extract the deep semantic features of each text item and the visual semantic features of each image and table item; Dynamic clustering module, which is used to perform dynamic clustering processing based on the deep semantic features of each text item and the visual semantic features of each image and table item, and then aggregate multi-dimensional scientific and technological achievement projects to build a scientific and technological achievement display library; The dynamic priority module is used to extract multiple scientific and technological achievement samples based on the scientific and technological achievement display library, and conduct a comprehensive evaluation of dynamic priorities to obtain the dynamic priority of each scientific and technological achievement sample; The interactive visualization module is used to design a multi-level visualization framework and to perform dynamic layout management according to the dynamic priority of each scientific and technological achievement sample to build a real-time interactive visualization framework. The real-time dynamic scheduling module is used to perform real-time dynamic scheduling processing on the real-time interactive visualization framework, and perform iterative reinforcement learning to build an intelligent visualization scheduling management model.

Images