CN109446247B - Scientific and technological innovation data visual analysis and display method - Google Patents

Abstract

The invention discloses a method for visual analysis and display of scientific and technological innovation data. The method realizes the visualization scalability and reusability of scientific and technological innovation data by establishing a visual representation abstract model and an information polyhedron data model. With the help of technological innovation data feature analysis, a nested circle visualization algorithm for hierarchical data and a fisheye moment diagram visualization algorithm for mesh data are designed. This method has certain generality for large-scale data and data with complex relationship.

Description

Scientific and technological innovation data visualization analysis and display method

Technical field

This patent application relates to the field of data visualization, in particular to a visualization analysis and display method for scientific and technological innovation data.

Background technique

Data visualization is the foundation of existing data analysis, and the current data visualization is gradually popularized. With the increase in the amount of existing data in science and technology government departments and the increasingly complex data associations, the existing data visualization systems for scientific and technological innovation have been unable to satisfy the visualization of large-scale data and information polyhedrons. Algorithms in data visualization have strong professional and personalized needs, so how to develop a visualization system that is scalable, reusable, and professional for scientific and technological innovation data is the main problem now.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is to provide a visual analysis and display method for scientific and technological innovation data, so as to solve the problems of scalability, reusability and professionalism of the scientific and technological innovation data visualization system.

In order to solve the above problems, the technical scheme adopted in the present invention is:

A scientific and technological innovation data visualization analysis and display method, comprising the following steps:

1) Abstract the data visualization process, establish an abstract model of visualization representation, and realize the unified modeling of data components and the expansibility of visualization components;

2) Abstract the information polyhedron data in the scientific and technological innovation data, establish an information polyhedron data model for the information side and user tasks, and realize the visualization of the information side and user tasks;

3) Combined with fisheye moment diagram interaction technology, perform fisheye moment diagram layout algorithm for scientific and technological innovation data to realize the visualization of data with complex relationship in scientific and technological innovation data; at the same time, combined with Venn visualization algorithm, use improved Venn visualization algorithm to realize scientific and technological innovation Visualization of nested circles in data for large scale data.

Preferably, the data component includes tabular data, and the tabular data will eventually be converted into JSON format data for unified configuration and mapping during use: tabular data includes table header, numeric value and data type, data type includes numeric data, date type data, textual data;

Visual components include bar charts (BarChart), line charts (LineChart), pie charts (PieChart), and force-oriented charts (ForceGraph).

Preferably, the information polyhedron data model includes three parts: a user task model, an information side model and a user preference model, and the user task model includes the identifier of the user's established task, the task name, the task type, and the importance of the task;

The information side model includes data item collection, data item association table collection, data item time dimension collection, data item regional dimension collection, data item category classification, data item collection includes data item, data item keyword collection, wherein data item includes data table data category name, scientific and technological innovation data category attribute; data item association table set includes source table code, target data table code, data table association weight, data table association type>, data item category classification includes There are four categories of business data detailed data, business data statistical data, yearbook data statistical data, and yearbook data detailed data;

The data item keyword set refers to the ontology set with the same concept;

User preference model: User preference Preferences=F1(Sum(API), tbi(API)), indicating the importance of the task of verifying the i-th data table. Among them, Sum(API) represents the sum of the task weights of the tables used by the user, that is, Sum(API)=tb1API+tb2API+tb3API+...+tbiAPI; and tbiAPI represents the task weights of the ith data table, tbiAPI=F2( TaskType, Level), where TaskType represents the task type of the data table, and Level represents the task value of the data table.

Preferably, the process of establishing the information polyhedron data model is as follows:

From the perspective of information side, establish a unified data model for information side;

From the perspective of user tasks, establish the relationship between the user task model and the information side;

Combined with user tasks, establish user preference analysis to realize data analysis of user task history records.

Preferably, the fisheye moment diagram layout algorithm is:

Suppose the coordinates of the focus part to be enlarged in the graph are C=(x, y), and the coordinates in the view are C1=(x1, y1), then the corresponding relationship between C1 and C is

C ₁ =F _finsheye (C) (1)

C1点与X轴正向夹角为θ，则上式(1)可以变化为Let the angle between point C and the positive X axis be

The positive angle between point C1 and the X axis is θ, then the above formula (1) can be changed to

Assuming that the radius of the circle in the deformation area of the figure is R, it can be expressed as

R=(x ^m +y ^m ) ⁿ (3)

The radius of the circular area to be deformed is proportional to the parameters m, n. According to the above conditions, the circular enlargement formula of the final deformation area is given as

Preferably, the steps for improving the Venn visualization algorithm are as follows:

Step 1: First determine the radius of the node represented by each circle, the size of the leaf node in the hierarchical data is uniformly set to 1, and the size of the intermediate node is the sum of the number of sub-nodes contained;

Step 2: Select the center point O in the visualization area as the center point of the nested circle arrangement;

Step 3: Use the above improved Veen visualization algorithm to iteratively arrange the order from the root node to the leaf node;

Step 4: End when the last leaf node is arranged.

The further improvement of the technical solution of the present invention is:

Due to the adoption of the above technical solution, the technical progress achieved by the present invention is: the present invention is based on the characteristics of information dissemination, multi-channel and multi-level, and combined with the SIR model commonly used in epidemiology to construct a two-layer SIR with unidirectional influence. Information dissemination model. At the same time, the two-layer SIR information propagation model is improved by combining the characteristics of subjective heterogeneity and memory effect heterogeneity. In order to rank the importance of nodes in a two-layer network, three evaluation indicators of degree centrality, betweenness centrality and proximity centrality are used to comprehensively evaluate the nodes in the network. Using the TOPSIS (Multiple Attribute Decision Analysis) method and combining the index weights and network layer weights, a method for selecting important nodes in a two-layer network is presented.

The present invention mainly solves three problems:

(1) Based on the multi-path and multi-level characteristics of information in the network, combined with the SIR model commonly used in the study of information dissemination models, a two-layer SIR information dissemination model with one-way influence is given. The model has a separate propagation process in each layer, and in this model there is the influence of radiation propagating on the line to the line propagating off the line.

(2) The information dissemination process does not have the characteristics of heterogeneity. Therefore, in the process of information dissemination, the two-layer SIR information dissemination model is improved by combining the subjective heterogeneity and memory effect of the information in the dissemination process. A heterogeneous online and offline information dissemination model.

(3) Through the TOPSIS (Multiple Attribute Decision Analysis) method, the three evaluation indicators of degree centrality, betweenness centrality and proximity centrality are used for comprehensive calculation to evaluate the importance of each node in the network layer, and a A method for selecting important nodes in a two-layer network based on TOPSIS. And the nodes selected by this method are immune nodes as an immune strategy of the multi-layer network.

In order to truly reflect the information dissemination process in the real network, the present invention predicts the information flow by studying the information dissemination problem, controls the information dissemination and realizes the monitoring of public opinion, extracts the main factors in the network for research, and seeks to dominate the complex system in the two-layer network. The simple law of information dissemination is the most effective, convenient and universal way to study node behavior and strategy methods in different types of environments.

The research on the network behavior strategy in the present invention is only related to the network node itself that produces the behavior, regardless of its specific implementation mode, shielding the irrelevant details of the bottom layer, and making the research on the network behavior strategy integrated and simplified It has the characteristics of adaptability, universality and portability to network evolution.

Description of drawings

Figure 1 is a visual component structure diagram;

Figure 2 is a data component structure diagram;

Figure 3 is a structural diagram of an information polyhedron;

Fig. 4 is the traditional Venn visualization algorithm diagram;

Fig. 5 is a nested circle visualization algorithm diagram;

Figure 6 is a time chart of the node arrangement of the nested circle visualization algorithm.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and examples.

1. Definition of basic elements

(1) Visual representation abstract model

The visual representation abstract model describes the visual representation at multiple levels. The visual representation abstract model inherits some features of the user interface description language, and uses a modular description method to describe the interface. The description of the interface mainly includes three parts: visual component (VCMD), data component (DMD), and data Mapping section (VDMD) with displayables.

Each component can be described as:

1. The visualization component is a description of the STUIMD visualization view, and the visualization component can configure and display common visualization methods. There is a many-to-one relationship between visual components and visual views, that is, a user functional view can configure multiple visual components, and visual components can be reused, which increases the reusability of visual components. The visualization components mainly include basic visualization components such as BarChart, LineChart, PieChart, and ForceGraph, as shown in Figure 1.

2. The data component is the description of the data in the STUIMD model, which can configure and describe the data for visualization. The visualization of scientific and technological innovation data is mainly divided into two types of data, including tabular data and network structure data. However, in the data component, the network structure data is converted, so that the network structure data is converted into data in the form of a table. The formal description of table structure data is given, and the table structure description includes three definitions of table header, value, and data type. At the same time, the data types mainly include numeric data, date data, and text data. In use, the table data will eventually be converted into JSON format data for unified configuration and mapping, as shown in Figure 2.

3. The mapping component is to realize the associated configuration function of the visualization component and the data component, mainly for the configuration association record of the data table attribute and the visualization component attribute.

(2) Information polyhedron data model

The information polyhedron data model includes three parts: user task model, information side model and user preference model, see Figure 3 for details.

1. User task model Task=<taskID, taskName, taskType, level>, taskID represents the identifier of the task created by the user, taskName represents the task name, that is, the English name, taskType represents the task type, and level represents the importance of the task. Among them, taskType mainly includes four kinds: browse, extract, move up in parallel, and print.

2. Scientific and technological innovation data model STInfoFacet=<DataItemSet, DataItemRelationSet, DataTimeSet, DataRegionSet, Class>, data item set DataItemSet=<DataItem, KeyWordSet>, where data item DataItem=<DICode, AttriName, MetaData>, where DICode represents data table , AttriName represents the category name of scientific and technological innovation data, and MetaData represents the attribute of scientific and technological innovation data category. KeyWordSet represents a set of data item keywords (ontologies with the same concept). The data item association table set DataItemRelationSet=<SourceDICode, TargetDICode, Degree, Relationship>, where SourceDICode represents the source table code, TargetDICode represents the target data table code, Degree represents the data table correlation weight, and Relationship represents the data table correlation type. DataTimeSet represents a collection of data item time dimensions. DataRegionSet represents a collection of data item regional dimensions. Class represents the classification of data item categories, including four categories of business data detailed data, business data statistical data, yearbook data statistical data, and yearbook data detailed data.

3. User preference Preferences=F1(Sum(API), tbi(API)), indicating the importance of the task of verifying the i-th data table. The Sum(API) represents the sum of the task weights of the tables used by the user, that is, Sum(API)=tb1API+tb2API+tb3API+...+tbiAPI. And tbiAPI represents the task weight of the i-th data table, tbiAPI=F2(TaskType, Level), where TaskType represents the task type of the data table, and Level represents the task value of the data table.

(3) Nested circle visualization algorithm

Nested circles use circles or ellipses to represent the nodes of the tree structure, and the child nodes are inside the circle of the parent node. Nested circles have certain advantages in space utilization and in representing parent and child nodes.

1. Traditional Venn visualization algorithm, see Figure 4 for details:

Step 1: First select a point O in the visual interface as the center of the nested circle arrangement. Circles C1, C2, C3 are arranged around point O. As shown in Figure 4(a), the outermost circles {C1, C2, C3} are arranged after recording, and the arrangement between the three is in the counterclockwise direction {C1->C2->C3->C1}.

Step 2: When a new circle Ci is added, let the radius of the Ci circle be Ri, the radius of the circle is the same as the size of the node, let Cm be the circle closest to the center O, the circle to the left of Cm is Cm-1, and the circle to the right of Cm is Cm-1. The circle is Cm-1, and Cj is any circle of the outermost circle.

Step 3: According to the condition that Ci, Cm and Cm-1 are tangent and the radius of Ci, calculate the center of the Ci circle, and take the center position outside Cm as the center of the i-th node circle.

Step 4: Traverse whether all outer circles in the record intersect with this circle.

Step 5: If there is no outer circle intersecting it, as shown in Figure 4(b), add a new circle. The outer circle set {C1, C2, C3, Ci} is updated and recorded, and the update row is irregular according to the counterclockwise direction {C1->Ci->C2->C3->C1}.

Step 6: If the outer circle intersects with the newly added circle, reset the circle tangent to the new circle to be Cm+1 and Cm, and return to step 3.

Step 7: Assuming that when the new circle is tangent to Cm+p, the outer circle will not intersect it for the first time, then update the outer circle set and the arrangement rule set.

Step 8: Until the last node arrangement is completed, end.

5. Improved algorithm pseudocode:

The visualization algorithm for nested circles is equivalent to the iterative algorithm for the Veen node algorithm. Proceed as follows:

Step 1: First, determine the radius of the node represented by each circle. The size of the leaf node in the hierarchical data is uniformly set to 1, and the size of the intermediate node is the sum of the number of sub-nodes contained.

Step 2: Select the center point O in the visualization area as the center point of the nested circle arrangement.

Step 3: Use the above improved Veen visualization algorithm to iteratively arrange the order from the root node to the leaf node.

Step 4: End when the last leaf node is arranged.

⑷ Fisheye moment diagram layout algorithm

1. Set the coordinates of the focus part to be enlarged in the graph as C=(x, y), and the coordinates in the view as C1=(x1, y1), then the corresponding relationship between C1 and C is

C ₁ =F _finsheye (C) (1)

C1点与X轴正向夹角为θ，则上式(1)可以变化为Let the angle between point C and the positive X axis be

The positive angle between point C1 and the X axis is θ, then the above formula (1) can be changed to

Assuming that the radius of the circle in the deformation area of the figure is R, it can be expressed as

R=(x ^m +y ^m ) ⁿ (3)

The radius of the circular area to be deformed is proportional to the parameters m, n. According to the above conditions, the circular enlargement formula of the final deformation area is given as

2. Experimental verification and analysis

(1) Verification and analysis of nested circle visualization algorithm

1. Figure 6 shows: the experimental data uses 10,000 to 50,000 random nodes of size and size to arrange the nodes using the traditional Venn visualization algorithm and the improved algorithm, and count the time spent by each algorithm. Since the time spent in the outer circle arrangement cannot be accurately obtained in the experiment, but the two algorithms use the same experimental data and the same method for finding the shortest distance to the center node, the experiment is set from the beginning to the completion of the node arrangement as the two Experiment statistics time.

experiment analysis:

1. The time growth rate of the traditional Venn visualization algorithm increases with the increase of the number of nodes. As the node data increases, the outer nodes also increase, and each new node joins needs to determine whether all the outer circles intersect with it, so it will take more and more time to arrange large-scale hierarchical nodes. The improved Venn algorithm defines that only 20 adjacent nodes are judged each time whether they intersect with it. When the size of the nodes differs greatly, the judgment value can be appropriately increased. Since only 20 nodes are judged each time, the time taken by the improved algorithm will be much smaller than the traditional Venn algorithm, showing linear complexity. The larger the number of nodes, the more obvious the advantages of the improved algorithm, so the improved algorithm has higher efficiency than the traditional Venn algorithm.

(2) Experimental evaluation of fisheye moment diagram

1. Table 1 shows: the experimenter observes and learns how to use the three experimental methods within 20 minutes. After mastering the method, the experimenter completed two specific tasks and recorded the time spent. The use of third-party plugins or the aid of other search tools is not allowed in the experiment. Finally, the time spent by the experimenter was used to calculate the mean and standard deviation.

Table 1 Comparison of task completion time

2. Table 2 shows that the survey is carried out from the four aspects of ease of learning, ease of use, effectiveness and reliability of the three experimental tools. At the same time, the full score of each survey item is set to 10 points, and finally the evaluation results of the experimenter are averaged value and standard deviation.

Table 2 Comparison of questionnaire results

experiment analysis:

1. The time used by the experimenter to complete the task using the fisheye moment diagram is less than that of the other two methods. From the standard deviation of the data, the difference in the time between the experimenters completing the task is not large, indicating that the effect of the fisheye moment diagram is stable and reliable. When the user looks for a specific node, first determine the approximate range of the node according to the color, then determine the node position according to the name of the node, and finally use the fisheye view to zoom in on the node position, observe and record the number of node associations. 20 experimenters also reported that the fisheye view has obvious advantages in recording the number of node associations. At the same time, when the user adopts the fisheye view, the overall effect of the moment graph remains unchanged, and the user can quickly find the next node.

2. The ease of learning, ease of use, effectiveness and reliability of the fisheye moment diagram is obviously better than that of the other two methods, followed by the traditional moment diagram, and the method of using the Excel table to complete the task is the worst. The experimenter feedback that the fisheye moment diagram is the best in terms of ease of use, which greatly improves the user's work efficiency. At the same time, according to the standard deviation comparison, it is found that the feedback situation of the experimenter is stable and there is no obvious difference, indicating that the fisheye moment diagram has obvious advantages over the other two methods.

Claims (2)

1. A scientific and technological innovation data visualization analysis and display method, characterized in that: 1) Abstract the data visualization process, establish an abstract model of visualization representation, and realize the unified modeling of data components and the expansibility of visualization components; 2) Abstract the information polyhedron data in the scientific and technological innovation data, establish an information polyhedron data model for the information side and user tasks, and realize the visualization of the information side and user tasks; 3) Combined with fisheye moment diagram interaction technology, perform fisheye moment diagram layout algorithm for scientific and technological innovation data to realize the visualization of data with complex relationship in scientific and technological innovation data; at the same time, combined with Venn visualization algorithm, use improved Venn visualization algorithm to realize scientific and technological innovation Visualization of nested circles in data for large-scale data; The data component includes table data, which will eventually be converted into JSON format data for unified configuration and mapping during use: table data includes table headers, values and data types, and data types include numeric data, date data, and text. type data; Visual components include bar chart (BarChart), line chart (LineChart), pie chart (PieChart), force-directed chart (ForceGraph); The information polyhedron data model includes three parts: the user task model, the information side model and the user preference model. The user task model includes the identifier of the user's established task, the task name, the task type, and the importance of the task; The information side model includes data item collection, data item association table collection, data item time dimension collection, data item regional dimension collection, data item category classification, data item collection includes data item, data item keyword collection, wherein data item includes data table data category name, scientific and technological innovation data category attribute; data item association table set includes source table code, target data table code, data table association weight, data table association type>, data item category classification includes There are four categories of business data detailed data, business data statistical data, yearbook data statistical data, and yearbook data detailed data; The data item keyword set refers to the ontology set with the same concept; User preference model: User preference Preferences=F1(Sum(API), tbi(API)), indicating the importance of the task of verifying the i-th data table, where Sum(API) represents the sum of the task weights of the tables used by the user, namely Sum(API)=tb1API+tb2API+tb3API+...+tbiAPI; and tbiAPI represents the task weight of the ith data table, tbiAPI=F2(TaskType, Level), where TaskType represents the task type of the data table, and Level represents the data The task value of the table; The process of establishing the information polyhedron data model is as follows: From the perspective of information side, establish a unified data model for information side; From the perspective of user tasks, establish the relationship between the user task model and the information side; Combine user tasks, establish user preference analysis, and realize data analysis of user task history; The fisheye moment diagram layout algorithm is: Suppose the coordinates of the focus part to be enlarged in the graph are C=(x, y), and the coordinates in the view are C1=(x1, y1), then the corresponding relationship between C1 and C is: C ₁ =F _finsheye (C) (1) Let the angle between point C and the positive X axis be

The positive angle between point C1 and the X axis is θ, then the above formula (1) can be changed to:

Assuming that the radius of the circle in the deformation area of the graph is R, the corresponding transformed fisheye radius can be expressed as formula (3): R=(x ^m +y ^m ) ⁿ (3) The radius of the circular area to be deformed is proportional to the parameters m, n. According to the above conditions, the circular enlargement formula of the final deformed area is given as:

2. technology innovation class data visualization analysis and display method according to claim 1, is characterized in that: the step of improving Venn visualization algorithm is as follows: Step1: First determine the radius of the node represented by each circle, the size of the leaf node in the hierarchical data is uniformly set to 1, and the size of the intermediate node is the sum of the number of sub-nodes contained; Step2: Select the center point O in the visualization area as the center point of the nested circle arrangement; Step3: Use the above improved Veen visualization algorithm to iteratively arrange the order from the root node to the leaf node; Step4: End when the last leaf node is arranged.

Images