Information visualization method and system, medium, equipment and program product

Abstract

The invention provides an information visualization method and system, a medium, equipment and a program product, and relates to the technical field of information processing, and the method comprises the steps: obtaining heterogeneous input information; performing category detection on the heterogeneous input information by utilizing the classification model, and determining a category corresponding to the heterogeneous input information; calculating entity relevancy, information relevancy and information weight corresponding to the heterogeneous input information according to the category; inputting structured label information corresponding to the heterogeneous input information according to the entity relevancy, the information relevancy and the information weight; constructing an atlas data structure corresponding to the structured label information; and mapping the map data structure into visual elements to obtain visual information. According to the method, heterogeneous input information is converted into visual information, so that leap-type improvement from bottom-layer data quantification to top-layer cognitive perception is realized.

Description

Technical Field

[0001] This application relates to the field of information processing technology, and in particular to an information visualization method, system, medium, device and program product. Background Technology

[0002] Existing systems typically display information as independent entries, making it difficult for users to quickly identify hidden relationships between different pieces of information or entities. This necessitates extensive manual contextual comparison and reasoning. Faced with thousands of pieces of information, there is a lack of a multi-dimensional, adaptive scoring mechanism to determine the importance, timeliness, and relevance to specific entities. The mixing of important and redundant information leads to low analytical efficiency.

[0003] Therefore, improving the ability to filter information is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] The purpose of this application is to provide an information visualization method, system, computer-readable storage medium, electronic device, and computer program product that can quickly identify hidden relationships between different information or entities by realizing information visualization.

[0005] To address the aforementioned technical problems, this application provides an information visualization method, the specific technical solution of which is as follows:

[0006] Acquire heterogeneous input information;

[0007] A classification model is used to perform category detection on the heterogeneous input information to determine the category corresponding to the heterogeneous input information;

[0008] Calculate the entity relevance, information relevance, and information weight corresponding to the heterogeneous input information based on the category;

[0009] The structured label information corresponding to the heterogeneous input information is input based on the entity relevance, the information relevance, and the information weight;

[0010] Construct the graph data structure corresponding to the structured label information;

[0011] The graph data structure is mapped to visualization elements to obtain visualization information.

[0012] Optionally, before using a classification model to perform category detection on the heterogeneous input information and determine the category corresponding to the heterogeneous input information, the method further includes:

[0013] Read the heterogeneous input information and determine the entities of interest and information content contained in the heterogeneous input information;

[0014] Discard non-standard information that does not conform to the preset category or whose information content does not conform to the information content annotation, and obtain standard entity information;

[0015] Accordingly, classifying the heterogeneous input information using a classification model includes:

[0016] The standard entity information is classified using a classification model.

[0017] Optionally, classifying the heterogeneous input information using a classification model includes:

[0018] The standard entity information is fed into a deep text classifier, and the category parameters for the information category recognition task in the deep text classifier are adjusted to obtain the category corresponding to the heterogeneous input information.

[0019] Optionally, calculating the entity relevance, information relevance, and information weight corresponding to the heterogeneous input information based on the category includes:

[0020] Calculate the entity relevance based on the semantic matching degree between the information content and the entity of interest;

[0021] Calculate the information relevance based on the information distance between the information content and the processed information;

[0022] The entity of interest and the information content are input into a preset weighted linear model, which then outputs information weights based on the information content's timeliness, source, and polarity parameters.

[0023] Optionally, constructing the graph data structure corresponding to the structured label information includes:

[0024] A clustering algorithm is used to establish the first logical relationship between each entity and the information content in the structured tag information;

[0025] Clustering algorithms are used to extract information about similar content or entities to obtain a second logical relationship.

[0026] The graph data structure is generated by using the entity as a node, the first logical relationship as the first type of edge between nodes, and the second logical relationship as the second type of edge between nodes.

[0027] Optionally, the graph data structure can be mapped to visualization elements to obtain visualization information, including:

[0028] The node radius of the node corresponding to each entity is calculated based on the cumulative value of the entity relevance. The cumulative value is positively correlated with the node radius.

[0029] The linear thickness of the edges is mapped based on the information relevance; the edges include the first type of edges and the second type of edges;

[0030] The color of the edge is mapped based on the information weight;

[0031] Visual information is output based on the node radius, the linear thickness of the edge, and the color of the edge.

[0032] This application also provides an information visualization system, including:

[0033] The acquisition module is used to acquire heterogeneous input information;

[0034] The category detection module is used to perform category detection on the heterogeneous input information using a classification model, and determine the category corresponding to the heterogeneous input information.

[0035] The information parameter calculation module is used to calculate the entity relevance, information relevance, and information weight corresponding to the heterogeneous input information according to the category.

[0036] The structured processing module is used to input structured label information corresponding to the heterogeneous input information based on the entity relevance, the information relevance, and the information weight;

[0037] The graph generation module is used to construct the graph data structure corresponding to the structured label information;

[0038] The visualization module is used to map the graph data structure into visualization elements to obtain visualization information.

[0039] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0040] This application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method described above when it invokes the computer program in the memory.

[0041] This application also provides a computer program product, including a computer program that, when executed, implements the steps of the method described above.

[0042] This application provides an information visualization method, the specific technical solution of which is as follows: acquiring heterogeneous input information; using a classification model to perform category detection on the heterogeneous input information to determine the category corresponding to the heterogeneous input information; calculating the entity relevance, information relevance, and information weight corresponding to the heterogeneous input information based on the category; inputting structured label information corresponding to the heterogeneous input information based on the entity relevance, the information relevance, and the information weight; constructing a graph data structure corresponding to the structured label information; and mapping the graph data structure to visualization elements to obtain visualization information.

[0043] This application, by acquiring heterogeneous input information, can process complex data from different sources, formats, and types, broadening the applicability of information visualization and enabling diverse data to be incorporated into the visualization processing workflow. Utilizing a classification model to perform category detection on heterogeneous input information accurately determines the category of the information, making information processing more targeted and refined. Information of different categories can be further analyzed and visualized based on its own characteristics and importance, avoiding the problem of inaccurate or incomplete information processing caused by generalizing different types of information, thus improving the efficiency and quality of information processing. By calculating the entity relevance, information relevance, and information weight corresponding to heterogeneous input information according to categories, the inherent connections and importance between information can be deeply explored. By quantifying these relevances and weights, key entities and important information content among numerous pieces of information, as well as the relationships between them, can be more clearly identified. By constructing a graph data structure corresponding to structured labeled information, the complex structures such as hierarchical relationships and associations between information can be intuitively displayed, enabling visualized information to more accurately convey the internal logic and structure of the information, enhancing the readability and understandability of the visualized information. Finally, mapping the graph data structure to visual elements enables users to more easily access and understand information, reducing the cognitive cost of complex information and improving the efficiency and effectiveness of information delivery. Users can easily identify key points, trends, and relationships in information through intuitive visual elements, thereby making better decisions and analyses, fully leveraging the value of information, and enhancing the effectiveness and practicality of information visualization in real-world applications.

[0044] This application also provides an information visualization system, a computer-readable storage medium, an electronic device, and a computer program product, which have the above-mentioned beneficial effects, and will not be elaborated here. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0046] Figure 1 A flowchart illustrating an information visualization method provided in an embodiment of this application;

[0047] Figure 2 This is a schematic diagram of an information visualization system structure provided in an embodiment of this application;

[0048] Figure 3 This is a structural diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0050] See Figure 1 , Figure 1 A flowchart illustrating an information visualization method provided in this application embodiment, the method comprising:

[0051] S101: Obtain heterogeneous input information;

[0052] S102: Use a classification model to perform category detection on the heterogeneous input information and determine the category corresponding to the heterogeneous input information;

[0053] S103: Calculate the entity relevance, information relevance, and information weight corresponding to the heterogeneous input information according to the category;

[0054] S104: Input the structured label information corresponding to the heterogeneous input information according to the entity relevance, the information relevance, and the information weight;

[0055] S105: Construct the graph data structure corresponding to the structured tag information;

[0056] S106: Map the graph data structure into visualization elements to obtain visualization information.

[0057] In step S102, various types of classification models can be used to perform category detection on heterogeneous input information. One feasible implementation is to use machine learning-based classification algorithms, such as Support Vector Machines (SVM). SVM separates data of different categories by finding the optimal separating hyperplane. For heterogeneous input information, it can find suitable boundaries in the feature space based on the features of the input data, thereby accurately determining the category to which the information belongs. For example, when the input information contains multiple types such as text and images, SVM can comprehensively consider multi-dimensional information such as keywords in the text content and pixel features of the image to determine its category.

[0058] In another feasible implementation, the standard entity information corresponding to the embedded vector sequence can be input into a deep text classifier, and the category parameters of the information category recognition task in the deep text classifier can be adjusted to obtain the category corresponding to the heterogeneous input information. Specifically, a pre-trained language model is used to generate the embedded vectors. For example, BERT can be used to map text to vectors in a high-dimensional space. The standard entity information is input into the pre-trained language model, and the model outputs the corresponding embedded vector sequence. Each embedded vector represents the semantic information of a word or phrase in the text, and the vector sequence as a whole reflects the semantic structure of the standard entity information.

[0059] Embedding vectors can also be generated using custom word embedding methods. This involves training a word embedding model, such as Word2Vec or GloVe, using a large amount of pre-collected text data. During training, the model learns the vector representation of each word based on the contextual relationships within the text, and then converts the words in the standard entity information into corresponding embedding vectors, forming a sequence of embedding vectors.

[0060] Deep text classifiers can be built using attention mechanisms, such as the Transformer architecture. The Transformer architecture assigns different attention weights to each element in the embedded vector sequence, allowing the model to focus more on key information within the text. During construction, a multi-head attention mechanism is designed, enabling the model to capture semantic relationships within the text from different perspectives. Supervised learning or transfer learning methods can be used to adjust the class parameters, improving the classifier's accuracy in classifying heterogeneous input information.

[0061] Once the class parameters of the deep text classifier are tuned, the embedding vector sequences corresponding to new heterogeneous input information can be fed into the classifier. The classifier will identify the category of the input information based on its learned features and class parameters. For each input embedding vector sequence, the classifier will output a class probability distribution, representing the probability that the input information belongs to each category.

[0062] In one feasible implementation, before step S102 is executed, the heterogeneous input information can be read first to determine the entities of interest and information content contained in the heterogeneous input information, and then discard non-standard information that the entities of interest do not conform to the preset category or the information content does not conform to the information content labeling, so as to obtain standard entity information.

[0063] Specifically, various methods can be used to access heterogeneous data from different sources. For example, web crawling technology can be used to scrape data from various websites and social media platforms on the Internet; database connection tools can be used to read stored data from internal relational or non-relational databases; for locally stored files, such as text files, Excel spreadsheets, and JSON files, file reading interfaces can be used. Appropriate parsing strategies are employed for different input formats. For structured data, such as tabular data in a database, SQL queries or database programming interfaces can be used to parse it according to fields and data types; for semi-structured data, such as XML and JSON data, specialized parsing libraries can be used to parse it based on its tag, key-value pair, and other structural characteristics; for unstructured data, such as plain text files, preprocessing operations such as text segmentation and sentence splitting can be performed before further extracting its information content.

[0064] When identifying entities of interest and information content, named entity recognition algorithms from natural language processing can be used. Based on pre-trained models, these algorithms can identify various entities in the text, such as names of people, places, organizations, and dates. Furthermore, a rule base can be built using domain knowledge, and regular expressions and other rule-based matching methods can be used to identify entities of interest within a specific domain. For non-textual data such as images, object detection algorithms can be used to identify entities within the images.

[0065] For text data, in addition to extracting entity-related content through entity recognition, text mining techniques, such as keyword extraction and topic modeling, can be used to extract key information from the text. For multimedia data, such as audio and video, transcription or subtitle extraction can be performed first, and then information can be extracted from the transcribed text content. For image data, image recognition technology can be combined to extract textual information or feature information describing the image content.

[0066] For entities of interest belonging to predefined categories, a category system can be established, and definitions can be made based on this system according to actual application scenarios and needs. When determining whether an entity fits a predefined category, the identified entity can be matched against the category system. Methods such as string matching and semantic similarity calculation can be used to determine whether the entity's category falls within the predefined range. For example, if the predefined category is entities related to a specific industry domain, an industry knowledge graph can be used to compare the entity with nodes in the graph to determine whether it belongs to that industry domain.

[0067] The labeling standards for information content can be formulated based on specific application scenarios and data quality requirements. For information labeled as non-standard, judgment can be made using pre-defined rules or models. For example, for text information, its grammatical structure can be checked for correctness, whether it contains specific keywords or formats, etc.; for numerical information, it can be checked whether it is within a reasonable range, whether it conforms to specific calculation rules, etc. Information content that does not meet the labeling standards is discarded, thus obtaining standard entity information.

[0068] After determining the categories of heterogeneous input information, step S103 requires calculating the associated entity relevance, information relevance, and information weight.

[0069] Entity relevance can be calculated based on the semantic matching degree between the information content and the entities of interest. For example, for text-based input information, word embedding techniques such as Word2Vec or BERT can be used to convert words in the text into vector representations. Then, the semantic relevance between different entities can be measured by calculating metrics such as cosine similarity between vectors. If the input information contains multiple entities, such as people, places, and events, the degree of semantic association between these entities can be determined, thereby obtaining the entity relevance.

[0070] For calculating information relevance, the contextual relationships between information can be considered. Specifically, information relevance can be calculated based on the information distance between the information content and the processed information. Taking text information as an example, sentence structure, paragraph structure, etc., can be analyzed to determine the logical relationships and semantic connections between different sentences or paragraphs. For example, the relevance between sentences can be judged by analyzing factors such as co-occurring words and topic consistency. For heterogeneous input information containing images and text, information relevance can also be calculated by analyzing the degree of matching between image content and text description. For example, key elements in an image can be extracted using image recognition technology and then compared with the description in the text; the degree of matching can determine the information relevance between the image and text. Information relevance is used to identify the hidden connection path between information segment A and processed information B. Its calculation method is mainly based on the distance between the semantic vectors of A and B. If the semantic distance between A and B is close, the relevance is strong. In addition, the score can be improved by identifying secondary entities they share. For example, if two information segments simultaneously mention an unnoticed third-party entity, the information relevance score of information segment A can be improved, indicating that there is an indirect connection between the two.

[0071] The calculation of information weights involves inputting the entity of interest and the information content into a preset weighted linear model, which then outputs the information weights based on the information content's timeliness parameters, source parameters, and information polarity parameters.

[0072] Information weights are used to measure the intrinsic value of the information itself. The timeliness parameter uses an exponential decay function to handle time differences, ensuring that newer information receives a higher score. The source parameter characterizes the pre-defined credit rating of the information source, while the information polarity parameter characterizes whether the information is positive, negative, or neutral; for example, significant negative information receives a higher importance score. A pre-defined weighted linear model can output estimated scores or ratings for each of the timeliness, source, and information polarity parameters.

[0073] This step can be implemented by calling a scoring model, which can adopt a multi-task learning (MTL) architecture. All scoring tasks share a high-performance underlying text encoder (such as LSTM or Transformer) to capture the general semantic features of the information segment. However, after the encoder, the scoring model is differentiated into three independent scoring heads: entity relevance, information relevance, and information weight. Each scoring head is responsible for calculating the score or grade of a specific dimension, which can ensure the efficiency, professionalism, and parallelism of the scoring process.

[0074] After obtaining key information such as entity relevance, information relevance, and information weight, step S104 requires integrating this information to generate corresponding structured label information. This process can be achieved by designing a specific label generation algorithm. For example, a rule-based method can be used to transform entity relevance, information relevance, and information weight into specific labels based on predefined rules and thresholds. For instance, if the entity relevance is higher than a certain threshold and the information weight is also high, then a high-priority label can be assigned to the entity or information fragment, indicating that it has an important position and value in the input information.

[0075] Simultaneously, machine learning or deep learning methods can be combined to generate structured label information. For example, a neural network model can be trained, using entity relevance, information relevance, and information weights as input features. Through model learning and training, the corresponding structured labels can be automatically output. This approach can better handle complex feature relationships and nonlinear mappings, generating more accurate and reasonable structured label information.

[0076] When generating structured label information, the hierarchical structure and semantic meaning of the labels also need to be considered. For example, labels can be divided into different levels, such as first-level labels, second-level labels, etc., to describe the features of the input information in more detail. Step S104 can transform complex heterogeneous input information into label information with clear semantics and structure, so that the features and value of the input information can be presented in a standardized, understandable, and operable way. This provides clear and accurate data input for subsequent map construction and visualization, enabling the entire technical solution to process and display input information more efficiently and provide users with more valuable information services.

[0077] After obtaining the structured label information, the next step is to construct the corresponding graph data structure. The purpose of constructing the graph data structure is to organize the structured label information in a more intuitive, scalable, and queryable way, facilitating subsequent visualization and analysis.

[0078] In one feasible implementation, a graph database can be used to store and manage graph data. A graph database is a database system specifically designed for storing and querying graph-structured data. It can efficiently handle graph structure elements such as nodes, edges, and the relationships between them. For example, Neo4j is a popular graph database that provides a flexible data model and the efficient query language Cypher, which can easily transform entities and relationships in structured label information into nodes and edges in the graph database for storage and management. Through graph databases, graph data can be quickly queried and analyzed, such as querying the neighbor nodes of a specific entity or finding specific types of paths.

[0079] When constructing a graph data structure, it is necessary to clearly define the nodes and edges in the graph. In one feasible implementation, a clustering algorithm can be used to establish a first logical relationship between each entity and information content in the structured label information. The clustering algorithm then extracts information content with similar content or entities to obtain a second logical relationship. Finally, using the entities as nodes, the first logical relationship as a first type of edge between nodes, and the second logical relationship as a second type of edge between nodes, the graph data structure is generated.

[0080] It can be seen that the first logical relationship is the relationship between entities and information content, while the second logical relationship is used to characterize the association between information content.

[0081] In practical applications, deep semantic analysis is performed on the scored information segments. This model typically employs a Joint Entity and Relation Extraction (JER) architecture, capable of simultaneously identifying entities (such as people, events, things, and locations) in the text and the relationships between these entities and the information. For structured inputs (such as database records), this process is replaced by direct metadata mapping, which determines the first logical relationship.

[0082] For the second logical relationship, relationships between information with similar content or that mention the same secondary entities can be established through clustering algorithms or semantic matching (such as "reference relationship" or "conflict relationship"), thereby representing the second logical relationship.

[0083] Nodes can represent entities in structured tag information, such as people, places, events, and concepts. Each node can contain attributes such as the entity's name, type, and weight. Edges represent relationships between nodes, such as associations, causal relationships, and hierarchical relationships. Edges can also have attributes such as the type and strength of the relationship. For example, in a graph about the field of science and technology, different technological concepts can be used as nodes, and their dependencies and competition relationships can be used as edges, thus constructing a complete graph data structure.

[0084] In addition, graph data modeling tools and techniques can be used to assist in constructing graph data structures. For example, metadata management tools such as Apache Atlas can be used, which provide rich metadata modeling capabilities, helping users define the types, attributes, and relationships between nodes and edges in the graph. These tools allow for a more systematic construction and management of graph data structures, ensuring the accuracy and consistency of the graph.

[0085] Step S105 transforms the structured label information into a graph data structure with rich semantics and structure, enabling the features and relationships of the input information to be presented in a more intuitive and operable way. This provides strong data support for subsequent visualization, allowing users to more easily understand and analyze the inherent features and relationships of the input information, while also laying the foundation for advanced applications such as graph analysis and mining.

[0086] In step S106, the constructed graph data structure is mapped to visualization elements to obtain intuitive and easy-to-understand visual information. One feasible implementation is to use a general visualization library, such as D3.js. D3.js is a JavaScript-based visualization library that provides rich visualization components and tools, which can easily map elements such as nodes and edges in the graph data structure to visual elements such as graphics, lines, and colors. For example, nodes can be mapped to graphics such as circles or rectangles, and edges can be mapped to lines connecting these graphics. By setting different attributes such as color, size, and shape, the different characteristics and attributes of nodes and edges can be represented, thereby generating intuitive visual charts.

[0087] Another feasible implementation may include the following steps:

[0088] Step 1: Calculate the node radius of the node corresponding to each entity based on the cumulative value of the entity relevance; the cumulative value is positively correlated with the node radius.

[0089] The second step is to map the linear thickness of the edges based on the information relevance; the edges include the first type of edges and the second type of edges;

[0090] Third step: Map the color of the edge based on the information weight;

[0091] Step 4: Output visualization information based on the node radius, the linear thickness of the edge, and the color of the edge.

[0092] The relevance between entities is determined using specific algorithms or methods. Text similarity-based algorithms, such as cosine similarity, can be used to measure the similarity of entities in text descriptions; alternatively, graph neural networks can be used to mine the strength of associations between entities in complex network structures. The relevance values ​​of all entities related to each entity are summed to obtain a cumulative value for each entity. Based on the range of the cumulative value, different intervals are defined, each interval corresponding to a range of node radii. For example, the cumulative value can be divided into high, medium, and low intervals, corresponding to larger, medium, and smaller node radii, respectively. Alternatively, a mapping function between the cumulative value and the node radius can be established to ensure that a larger cumulative value corresponds to a larger node radius.

[0093] For each edge, its information relevance needs to be determined. This can be achieved by analyzing the strength of the relationship between the two entities connected by the edge. The first and second types of edges need to be clearly defined. First-type edges can represent direct relationships between entities, such as parent-child relationships or causal relationships; second-type edges can represent indirect or weaker relationships, such as association relationships or similarity relationships. Based on the edge type, the information relevance should be appropriately adjusted or weighted to more accurately reflect the actual importance of the edge.

[0094] Based on the degree of information relevance, the linear thickness of the edges is divided into multiple levels. For example, the information relevance can be divided into several levels such as very strong, strong, medium, and weak, corresponding to very thick, relatively thick, medium thickness, and relatively thin lines, respectively. The information relevance of the edges is mapped to a visual linear thickness, allowing users to intuitively perceive the importance of the edges.

[0095] Information weights are determined by analyzing the information content carried by the edges. Information weights can be comprehensively evaluated based on multiple factors, such as the type of relationship the edge represents, the amount of data flow, and the business value. For example, an edge representing a high-value business transaction relationship will have a higher information weight; if it is merely a simple data transmission relationship, its information weight will be relatively lower.

[0096] Color mapping rules are established based on the weight of information. Attributes such as brightness and warmth of colors can be used to distinguish different weight levels. For example, edges with high information weight are represented by bright, vivid colors, such as red or yellow; edges with low information weight are represented by darker colors, such as gray or blue. Furthermore, different color combinations can be selected based on specific application scenarios and user needs to achieve the best visualization effect.

[0097] According to the established color mapping rules, the information weight of each edge is mapped to a corresponding color. During the mapping process, it is ensured that the colors are sufficiently distinguishable so that users can clearly identify edges with different weights, thereby better understanding the importance of the information represented by the edges.

[0098] The final generated visualization is a highly dynamic and interactive relationship graph. The visual attributes (size, thickness, color, transparency) of entity nodes (circles or squares) and connecting lines (dashed or solid lines) in the graph reflect the multidimensional scoring of the underlying data in real time and intuitively. For example, a larger entity indicates that it is the center of the current analysis, and the thick red line connecting it indicates that the relationship is driven by a highly important and strongly related piece of information. Users can use interactive operations such as dragging, zooming, and filtering to leverage the visual cues of the graph, achieving a leap from massive amounts of data to key insights, and efficiently identifying core entities and the most critical connection paths in the information graph.

[0099] In visualized information, entity relevance measures the semantic match between the content of an information segment and entities (1 through N) that the user is interested in. For example, the frequency of mentioning entity names in the information, and the match between the information topic and entity tags. Information relevance measures the potential association between the information segment and other information or entities already received and processed by the system. For example, by calculating the distance between the semantic vector of the information segment and the semantic vectors of other information segments. Information importance characterizes the value of the information content itself.

[0100] When mapping graph data structures to visualization elements, user interaction needs must also be considered. For example, interactive features can be added to visualizations, such as displaying detailed information on mouse hover, performing detailed queries by clicking nodes or edges, and zooming and panning. Through these interactive features, users can explore the information within the graph data structure more deeply and better understand the characteristics and relationships of the input information. For instance, when a user hovers their mouse over a node, detailed attribute information about that node can be displayed, such as the entity's name, type, and weight; when a user clicks on an edge, detailed information about the relationship represented by that edge can be displayed, such as the relationship's type and strength.

[0101] This step achieves a leapfrog improvement from bottom-level data quantification to top-level cognitive perception by finely mapping three scoring dimensions onto the visual attributes of the graph: the size / color depth of an entity is linked to the cumulative value of entity relevance to reflect the centrality of the entity; the thickness of the connecting line is linked to the information relevance to intuitively show the strength of the association; and the color of the connecting line is linked to the importance of the information. The heatmap color gradient is used to convey the crisis or value level of the information to the user in real time, thereby achieving a leapfrog improvement from bottom-level data quantification to top-level cognitive perception.

[0102] This application's embodiments, by acquiring heterogeneous input information, can process complex data from different sources, formats, and types, broadening the applicability of information visualization and enabling diverse data to be incorporated into the visualization processing flow. Utilizing a classification model to perform category detection on heterogeneous input information accurately determines the category corresponding to the information, making information processing more targeted and refined. Information of different categories can be further analyzed and visualized based on its own characteristics and importance, avoiding the problem of inaccurate or incomplete information processing caused by generalizing different types of information, thus improving the efficiency and quality of information processing. By calculating the entity relevance, information relevance, and information weight corresponding to heterogeneous input information according to categories, the inherent connections and importance between information can be deeply explored. By quantifying these relevances and weights, key entities and important information content among numerous pieces of information, as well as their interrelationships, can be more clearly identified. By constructing a graph data structure corresponding to structured tag information, the complex structures such as hierarchical relationships and associations between information can be intuitively displayed, enabling visualized information to more accurately convey the internal logic and structure of the information, enhancing the readability and understandability of the visualized information. Finally, mapping the graph data structure to visual elements enables users to more easily access and understand information, reducing the cognitive cost of complex information and improving the efficiency and effectiveness of information delivery. Users can easily identify key points, trends, and relationships in information through intuitive visual elements, thereby making better decisions and analyses, fully leveraging the value of information, and enhancing the effectiveness and practicality of information visualization in real-world applications.

[0103] See Figure 2 , Figure 2 This is a schematic diagram of an information visualization system structure provided in an embodiment of this application. The system includes:

[0104] The acquisition module is used to acquire heterogeneous input information;

[0105] The category detection module is used to perform category detection on the heterogeneous input information using a classification model, and determine the category corresponding to the heterogeneous input information.

[0106] The information parameter calculation module is used to calculate the entity relevance, information relevance, and information weight corresponding to the heterogeneous input information according to the category.

[0107] The structured processing module is used to input structured label information corresponding to the heterogeneous input information based on the entity relevance, the information relevance, and the information weight;

[0108] The graph generation module is used to construct the graph data structure corresponding to the structured label information;

[0109] The visualization module is used to map the graph data structure into visualization elements to obtain visualization information.

[0110] Based on the above embodiments, as a preferred embodiment, it further includes:

[0111] The information filtering module is used to read the heterogeneous input information, determine the entities of interest and information content contained in the heterogeneous input information, discard non-standard information such as entities of interest that do not conform to a preset category or information content that does not conform to information content annotation, and obtain standard entity information.

[0112] This application also provides an embodiment of a computer-readable storage medium and a computer program product. Both the computer-readable storage medium and the computer program product may store a computer program that, when executed by a processor, implements the steps of the method described in the above method embodiments.

[0113] It is understood that if the methods in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and executes all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0114] The computer-readable storage medium provided in this embodiment includes the method mentioned above, and has the same effect.

[0115] This application also provides an electronic device, see [link to document]. Figure 3 The present application provides a structural diagram of an electronic device, such as... Figure 3 As shown, it may include a processor 1410 and a memory 1420.

[0116] The processor 1410 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 1410 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 1410 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 1410 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 1410 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0117] The memory 1420 may include one or more computer-readable storage media, which may be non-transitory. The memory 1420 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In this embodiment, the memory 1420 is used to store at least the following computer program 1421, which, after being loaded and executed by the processor 1410, is capable of implementing the relevant steps in the methods executed by the electronic device side as disclosed in any of the foregoing embodiments. In addition, the resources stored in the memory 1420 may also include an operating system 1422 and data 1423, etc., and the storage method may be temporary storage or permanent storage. The operating system 1422 may include Windows, Linux, Android, etc.

[0118] In some embodiments, the electronic device may further include a display screen 1430, an input / output interface 1440, a communication interface 1450, a sensor 1460, a power supply 1470, and a communication bus 1480.

[0119] certainly, Figure 3 The structure of the electronic device shown does not constitute a limitation on the electronic device in the embodiments of this application. In practical applications, the electronic device may include more than [other components]. Figure 3 More or fewer components as shown, or combinations of certain components.

[0120] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. As the system provided in the embodiments corresponds to the method provided in the embodiments, the description is relatively simple; relevant parts can be found in the method section.

[0121] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. It should be noted that those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

[0122] It should also be noted that, in this specification, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. An information visualization method, characterized in that, include: Acquire heterogeneous input information; A classification model is used to perform category detection on the heterogeneous input information to determine the category corresponding to the heterogeneous input information; Calculate the entity relevance, information relevance, and information weight corresponding to the heterogeneous input information based on the category; The structured label information corresponding to the heterogeneous input information is input based on the entity relevance, the information relevance, and the information weight; Construct the graph data structure corresponding to the structured label information; The graph data structure is mapped to visualization elements to obtain visualization information.

2. The information visualization method according to claim 1, characterized in that, Before using a classification model to perform category detection on the heterogeneous input information and determine the category corresponding to the heterogeneous input information, the method further includes: Read the heterogeneous input information and determine the entities of interest and information content contained in the heterogeneous input information; Discard non-standard information that does not conform to the preset category or whose information content does not conform to the information content annotation, and obtain standard entity information; Accordingly, classifying the heterogeneous input information using a classification model includes: The standard entity information is classified using a classification model.

3. The information visualization method according to claim 2, characterized in that, Using a classification model to perform category detection on the heterogeneous input information includes: The standard entity information is fed into a deep text classifier, and the category parameters for the information category recognition task in the deep text classifier are adjusted to obtain the category corresponding to the heterogeneous input information.

4. The information visualization method according to claim 2, characterized in that, The calculation of entity relevance, information relevance, and information weight corresponding to the heterogeneous input information based on the category includes: Calculate the entity relevance based on the semantic matching degree between the information content and the entity of interest; Calculate the information relevance based on the information distance between the information content and the processed information; The entity of interest and the information content are input into a preset weighted linear model, which then outputs information weights based on the information content's timeliness, source, and polarity parameters.

5. The information visualization method according to claim 1, characterized in that, Constructing the graph data structure corresponding to the structured label information includes: A clustering algorithm is used to establish the first logical relationship between each entity and the information content in the structured tag information; Clustering algorithms are used to extract information about similar content or entities to obtain a second logical relationship. The graph data structure is generated by using the entity as a node, the first logical relationship as the first type of edge between nodes, and the second logical relationship as the second type of edge between nodes.

6. The information visualization method according to claim 5, characterized in that, Mapping the aforementioned graph data structure to visualization elements yields the following visualization information: The node radius of the node corresponding to each entity is calculated based on the cumulative value of the entity relevance. The cumulative value is positively correlated with the node radius. The linear thickness of the edges is mapped based on the information relevance; the edges include the first type of edges and the second type of edges; The color of the edge is mapped based on the information weight; Visual information is output based on the node radius, the linear thickness of the edge, and the color of the edge.

7. An information visualization system, characterized in that, include: The acquisition module is used to acquire heterogeneous input information; The category detection module is used to perform category detection on the heterogeneous input information using a classification model, and determine the category corresponding to the heterogeneous input information. The information parameter calculation module is used to calculate the entity relevance, information relevance, and information weight corresponding to the heterogeneous input information according to the category. The structured processing module is used to input structured label information corresponding to the heterogeneous input information based on the entity relevance, the information relevance, and the information weight; The graph generation module is used to construct the graph data structure corresponding to the structured label information; The visualization module is used to map the graph data structure into visualization elements to obtain visualization information.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the method as claimed in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, implements the steps of the method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, Includes a computer program, which, when executed, implements the steps of the method as described in any one of claims 1 to 6.

Images