CN111966821A - Knowledge graph visualization method based on mechanics principle - Google Patents

Abstract

The invention discloses a knowledge map visualization method based on mechanical principles, and belongs to the field of knowledge map visualization. The invention converts the triple data of the knowledge graph into a relationship set and an entity set; wherein, the relationship set includes edge data, and the entity set includes node data; then the edge data and the node data are marked; and the marked edge data is marked. Deduplication and node data respectively; calculate the length of the edge, the intersection coordinates of the node and the edge and the coordinates of the edge name according to the node data and edge data; finally use d3.js to render the nodes and edges to obtain a visual knowledge graph. The invention overcomes the deficiencies in the prior art that the knowledge graph triplet cannot be converted into a visualized relational graph. The present invention can convert the knowledge graph triplet into a relational graph, and can realize the ability to analyze problems intuitively through the relational graph. , which more intuitively and completely expresses the information contained in the knowledge graph.

Description

A Knowledge Graph Visualization Method Based on Mechanical Principles

technical field

The invention belongs to the technical field of knowledge map visualization, and more particularly, relates to a knowledge map visualization method based on mechanical principles.

Background technique

In November 2012, Google first proposed the concept of Knowledge Graph (KG), indicating that it would add the function of Knowledge Graph to its search results. Its original intention is to improve the capabilities of search engines and enhance users' search quality and search experience. According to statistics in January 2015, the KG built by Google has 500 million entities and about 3.5 billion entity relationship information, and has been widely used to improve the search quality of search engines. Although the concept of Knowledge Graph is relatively new, it is not a new research field. As early as 2006, BernersLee proposed the idea of linked data, calling for the promotion and improvement of related technical standards such as URI (Uniform resource identifier), RDF (resource discription framework), OWL (Webontology language), ready for the arrival of the Semantic Web. Subsequently, a wave of semantic network research was launched. The knowledge graph technology is based on the relevant research results, which is a sublation and sublimation of the existing semantic network technology.

代表知识库中的三元组集合。三元组的基本形式主要包括实体1、关系、实体2和概念、属性、属性值等，实体是知识图谱中的最基本元素，不同的实体间存在不同的关系。概念主要指集合、类别、对象类型、事物的种类，例如人物、地理等；属性主要指对象可能具有的属性、特征、特性、特点以及参数，例如国籍、生日等；属性值主要指对象指定属性的值，例如中国、1988-09-08等。每个实体(概念的外延)可用一个全局唯一确定的ID来标识，每个属性-属性值对可用来刻画实体的内在特性，而关系可用来连接两个实体，刻画它们之间的关联。Knowledge graphs are usually represented in the form of triples, that is, G=(E, R, S), where E={e1, e2, e3,...,en} is the entity set in the knowledge base, including |E| different entities; R={r1,r2,…,rn} is the relation set in the knowledge base, which contains |R| different relations;

Represents a collection of triples in the knowledge base. The basic forms of triples mainly include entity 1, relationship, entity 2, concept, attribute, attribute value, etc. Entity is the most basic element in the knowledge graph, and different entities have different relationships. Concepts mainly refer to collections, categories, object types, and types of things, such as people, geography, etc. Attributes mainly refer to the attributes, characteristics, characteristics, characteristics and parameters that objects may have, such as nationality, birthday, etc. Attribute values mainly refer to the specified attributes of objects , such as China, 1988-09-08, etc. Each entity (the extension of the concept) can be identified by a globally unique ID, each attribute-attribute value pair can be used to describe the internal characteristics of the entity, and the relationship can be used to connect two entities and describe the association between them.

For the visualization of knowledge graphs, some solutions have also been proposed in the prior art. For example, the name of the invention is: A big data knowledge graph visualization method and device (application date: February 27, 2020; application number: 202010123185.0) This solution improves the Provided is a big data knowledge graph visualization method and device, wherein the method includes: obtaining data to be visualized; for node data and edge data in the data to be visualized, based on a preset D3.js library, obtaining formatted data. Node data and formatted edge data; based on the preset rendering method, the formatted node data and formatted edge data are generated to generate corresponding nodes to be displayed and edges to be displayed; real-time calculation of each The to-be-displayed position of the node to be displayed on the page, and the to-be-displayed position of each to-be-displayed edge on the page; based on the to-be-displayed position of each to-be-displayed node on the page, and the to-be-displayed edge on the page Display position, render the node to be displayed and the edge to be displayed into the page, and obtain a visual knowledge graph. However, the shortcoming of this scheme is that the operation of the knowledge graph only includes display and dragging, and cannot operate the nodes of the knowledge graph and cannot display the relationship between the nodes.

To sum up, how to convert the knowledge graph into a visualized relationship graph is an urgent problem to be solved in the prior art.

SUMMARY OF THE INVENTION

1. The problem to be solved

The present invention overcomes the deficiencies in the prior art that the knowledge graph triples cannot be converted into visualized relational graphs, and provides a knowledge graph visualization method based on mechanical principles, which can convert the knowledge graph triples into relational graphs , which can realize the ability to analyze problems intuitively through the relationship graph, and express the information contained in the knowledge graph more intuitively and completely.

2. Technical solutions

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

A method for visualizing knowledge graph based on mechanics principle of the present invention is characterized in that, it comprises the following steps:

Convert the triple data of the knowledge graph into a relationship set and an entity set; wherein, the relationship set includes edge data, and the entity set includes node data; the edge data and the node data are marked; the marked edge data and node data are respectively Carry out deduplication; calculate the length of the edge, the coordinates of the intersection of the node and the edge and the coordinates of the edge name according to the node data and the edge data; according to the length of the edge, the coordinates of the intersection of the node and the edge and the coordinates of the edge name, and use d3.js to compare the node and edge Render to get a visual knowledge graph.

Further, the specific process of marking the edge data and the node data is as follows: perform an associated query on the node data, and mark the queried node and the associated edge, wherein the queried node set is an idlist, and the edge and the node are marked. The collection is labellist.

Further, the length of the edge is calculated by the following formula:

link.length=link.name.length*fontSize+n

Among them, link.length represents the length of the side, link.name.length represents the length of the side name, fontSize represents the font size, and n represents a constant.

Furthermore, the specific process of calculating the coordinates of the intersection point between the node and the edge is as follows: the edge data includes the start node and the end node, and the coordinates of the intersection point of the node and the edge are calculated according to the center coordinates of the start node and the end node.

Further, when the edge is a line connecting the center of the circle between the start node and the end node, the intersection coordinates of the node and the edge are calculated by the following formula:

When link.target.x is equal to link.source.x:

y1=link.target.y-link.source.y>0? link.source.y+r: link.source.y–r;

y2=link.target.y-link.source.y>0? link.target.y-r: link.target.y+r;

x1 = link.source.x;

x2 = link.target.x;

Among them, (link.source.x, link.source.y) is the center coordinate of the starting node, (link.target.x, link.target.y) is the center coordinate of the ending node, and r is the starting node and the ending node. The radius value of the node, (x1, y1) is the intersection coordinates of the edge and the start node, (x2, y2) is the intersection coordinates of the edge and the end node;

When link.target.y is equal to link.source.y:

y1 = link.source.y;

y2 = link.target.y;

x1=link.target.x-link.source.x>0? link.source.x+r: link.source.x-r;

x2=link.target.x-link.source.x>0? link.target.x-r: link.target.x+r;

When link.target.x is not equal to link.source.x and link.target.y is not equal to link.source.y:

为起始节点和终止节点之间的圆心角tan值。in,

is the central angle tan value between the start node and the end node.

Further, when the edge is located on both sides of the line connecting the center of the circle between the start node and the end node, the intersection coordinates of the node and the edge are calculated by the following formula:

Calculate the intersection coordinates (xm, ym) of the vertical line connecting the center of the circle through the intersection of the edge and the starting node:

Among them, a=r*link.linknum/maxLinkNumber, link.linknum is the layer number mark of the edge, and maxLinkNumber is the maximum number of edges on both sides of the center of the circle;

Calculate the intersection coordinates (xn, yn) of the vertical line connecting the center of the circle through the intersection of the edge and the end node:

Further, the edge name coordinates are calculated by the following formula; let sy=link.source.y, sx=link.source.x, tx=link.target.x, ty=link.target.y;

dx=(distance-link.name.length*fontSize-2*r)/2

Then use d3.js to calculate dx to get dy, (dx, dy) is the coordinate of the edge name.

Further, the specific process of rendering using d3.js is: verifying the node data and edge data, and then using d3.js to render to obtain a visual knowledge graph after verification.

Further, the specific process of verifying the node data and the edge data is as follows: verifying whether the start node and the end node of the edge exist, if any of the start node and the end node does not exist, then the edge to filter.

3. Beneficial effects

Compared with the prior art, the beneficial effects of the present invention are:

A knowledge graph visualization method based on the mechanical principle of the present invention takes the entities of the knowledge graph of structured data as nodes and the relationships as edges, thereby realizing the visual expression of the knowledge graph. Further, each relationship is expressed independently based on the mechanical principle, thereby realizing the complete expression of the information in the knowledge graph. In addition, through the scaling of nodes, the nodes that do not need to be displayed are put away, so as to realize the ability to analyze problems through the relationship graph, and express the information contained in the knowledge graph more intuitively and completely.

Description of drawings

Fig. 1 is the method flow schematic diagram of the present invention;

FIG. 2 is a schematic diagram of the coordinates of the intersection of an edge and a node according to the present invention.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of the embodiments of the present invention, not all of the embodiments; moreover, each embodiment is not relatively independent, and can be combined with each other according to needs, so as to achieve better effects. Thus, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but is merely representative of selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

In order to further understand the content of the present invention, the present invention will be described in detail with reference to the accompanying drawings and embodiments.

Example 1

As shown in FIG. 1, a knowledge graph visualization method based on mechanical principles of the present invention includes the following steps:

1. Data conversion

Convert the triplet data of the knowledge graph into a relationship set and an entity set; it is worth noting that the relationship set includes edge data, and the entity set includes node data; the converted data is the data structure that can be processed by d3.js. The structure looks like this:

Among them, nodes is a collection of entities, nodes record node data, each node has at least two attributes: name and id, name is the display name of the node, and id is the unique identifier of the node. Links records the edge data for the relationship set. The edge data represents the relationship between nodes with an associated relationship. Each edge records the source, target and id. The source is the start node of the edge, and the target is the end node of the edge. The id is Unique ID of the edge.

2. Tag data

Mark the edge data of the relationship set and the node data in the entity set; specifically, perform an associated query on the node data, and mark the queried node and the associated edge, wherein the queried node set is idlist, edge and node label The collection is labellist. The specific process includes:

2-1) Origin-related data query: In the initialization state, it contains an origin and other nodes associated with the origin

Mark the origin and mark it as origin; mark other nodes, the id marked as the origin is the origin id;

Mark the edge, and the id marked as the starting point is the origin id; add the origin id to the global variable idList.

2-2) Other node related data query: expansion of other nodes other than the origin

For node labeling, add the label of the expanded node plus the id of the expanded node; for the edge label, add the label of the expanded node to the id of the expanded node; add the id of the expanded node to the global variable idList.

3. Data deduplication

The marked edge data and node data are deduplicated respectively. Since the two nodes may be associated with the same node after expansion, it is necessary to compare the data before marking with the data after marking. The specific steps are as follows:

3-1) Do double traversal of the nodes data before marking and after marking, if there are existing nodes in the marked nodes data, that is, the id is the same to remove duplicate nodes.

3-2) Do double traversal of the links data before and after the mark, if there are existing edges in the links data after the mark, that is, the id is the same, and the duplicate edges are removed

4. Calculate the length of the side

Calculate the length of the edge according to the node data and edge data; specifically, calculate the length of the edge by the following formula:

link.length=link.name.length*fontSize+n

Among them, link.length represents the length of the side, link.name.length represents the length of the side name, which is the string length of the side name, and fontSize represents the font size, which is the font size of the text set at the front end, represented by a positive integer; n represents Constant, the n is the length of the space on both sides of the edge name, and the value range of n is (0, 10).

5. Calculate the intersection coordinates of nodes and edges

As shown in Figure 2, the intersection coordinates of the node and the edge are calculated according to the node data and the edge data. Specifically, the edge data includes the start node and the end node, and the intersection of the node and the edge is calculated according to the center coordinates of the start node and the end node. coordinate. The calculation steps include:

5-1) Group the edges; specifically, according to the start node and the end node of the edge, the direction of the connecting line is not distinguished, as long as the edge connects the same pair of nodes, it is considered to be the same group:

Calculate the hash value of the start node (source) and end node (target) of the edge:

m(source)=link.source string length; m(target)=link.target string length

Compare the hashes of source and target:

key=link.source.hash<link.target.hash? link.source+':'+link.target:link.target+':'+

link.source;

Where m is the length of the source or target string, and ASCII (target[i]/source[i]) is the ASCII code of each character of the target or source.

Traverse all the edges in the links, if the keys of the edges are the same, they are in the same group.

5-2) Distribute the edges of the same group evenly on both sides passing through the center of the two nodes

maxLinkNumber=group.length%2==0? group.length/2:(group.length-1)/2

Among them, maxLinkNumber is the maximum number of edges on both sides, and group.length is the number of current edge groups.

Each edge is marked with the number of layers. The bottom edge is marked as -maxLinkNumber, and the number of layers from the bottom to the top is incremented by 1, that is, the number of layers of the center line between the start node and the end node is marked as 0, and the upper layer is marked as 0. The layer number marks of the edges of 1 are sequentially increased by 1, and the layer number marks of the lower layers are decreased by 1 in turn.

5-3) Determine whether the layer number mark link.linknum of the edge is equal to 0

5-4) If link.linknum=0, the intersection coordinates (x1, y1), (x2, y2) of the node and the edge are calculated by the following formula:

When link.target.x is equal to link.source.x, that is, when the start and end nodes are on a vertical line:

y1=link.target.y-link.source.y>0? link.source.y+r: link.source.y–r;

y2=link.target.y-link.source.y>0? link.target.y-r: link.target.y+r;

x1 = link.source.x;

x2 = link.target.x;

Among them, (link.source.x, link.source.y) is the center coordinate of the starting node, (link.target.x, link.target.y) is the center coordinate of the ending node, and r is the starting node and the ending node. The radius value of the node, (x1, y1) is the intersection coordinates of the edge and the start node, (x2, y2) is the intersection coordinates of the edge and the end node;

When link.target.y is equal to link.source.y, i.e. when the start and end nodes are on a horizontal line:

y1 = link.source.y;

y2 = link.target.y;

x1=link.target.x-link.source.x>0? link.source.x+r: link.source.x-r;

x2=link.target.x-link.source.x>0? link.target.x-r: link.target.x+r;

When link.target.x is not equal to link.source.x and link.target.y is not equal to link.source.y, that is, when the start node and end node are neither on the same horizontal line nor on the same vertical line :

为起始节点和终止节点之间的圆心角tan值。in,

is the central angle tan value between the start node and the end node.

5-5) If link.linknum is not equal to 0, calculate the intersection coordinates of nodes and edges by the following formula:

Make a vertical line on the line connecting the center of the circle, and calculate the distance a between the intersection point and the line connecting the center of the circle according to the number of layers marked on the side:

a=r*link.linknum/maxLinkNumber

Among them, link.linknum is the layer number mark of the edge, and maxLinkNumber is the maximum number of edges on both sides of the center of the circle;

Calculate the intersection coordinates (xm, ym) of the vertical line connecting the center of the circle through the intersection of the edge and the starting node:

When the start and end nodes are on a vertical line:

Calculate the intersection coordinates (xn, yn) of the vertical line connecting the center of the circle through the intersection of the edge and the end node:

Where (xn, yn) is the coordinate of the intersection point of the line connecting the center of the circle with the intersection of the edge and the end node

When the start and end nodes are on a vertical line:

Draw a vertical line through the intersection of the edge and the node and the line connecting the center of the circle, and the slope of the vertical line and the x-axis:

tanω=(x1-x2)/(y2-y1)

The x-axis distance relative to the vertical point (xm, ym):

The y-axis distance relative to the vertical point (xm, ym):

if (y2-y1)=0

kx=0

ky=a

If a>0:

xs=a>0? xm-kx:xm+kx

ys=ym-ky

xt=a>0? xn-dx:xn+dx

yt=yn-ky

Where (xs, ys) are the coordinates of the intersection of the edge and the starting node, and (xt, yt) are the coordinates of the intersection of the edge and the ending node. It is worth mentioning that, considering the way of displaying multiple relationships between nodes, multiple relationships between nodes are not combined into one edge, so that each relationship can be expressed independently, thus realizing the complete expression of information in the knowledge graph.

6. Calculate the coordinates of the edge name

Calculate the edge name coordinates according to the node data and edge data; calculate the edge name coordinates by the following formula; let sy=link.source.y, sx=link.source.x, tx=link.target.x, ty=link.target. y;

dx=(distance-link.name.length*fontSize-2*r)/2

Then use d3.js to calculate dx to get dy, (dx, dy) is the coordinate of the edge name.

7. Rendering with d3.js

According to the length of the edge, the intersection coordinates of the node and the edge, and the coordinates of the edge name, and use d3.js to render the node and edge to obtain a visual knowledge graph. The specific process is: check the node data and the edge data, specifically, check whether the start node and the end node of the edge exist, if any of the start node and the end node does not exist, then the edge Filtering; after verification, use d3.js to render to obtain a visual knowledge graph.

It is worth noting that in the process of visualization using d3.js rendering, the specific steps of node expansion and node contraction are as follows:

7-1) Node expansion, the steps include:

7-1-1) First judge whether the id of the node already exists in the idList. If it does, it means that the node has been expanded and no additional query is needed, and the node is currently in a shrinking state, just the label of the node Remove from labelList and re-render.

7-1-2) If the id of the node does not exist in the idList, it indicates that the node has not been expanded. It is necessary to query the associated nodes and relationships of the node, and add the id of the node to the idList and re-render it.

7-2) The node shrinks, specifically, the label of the node is added to the labelList, and then re-rendered. It is worth noting that, through the scaling of nodes, the nodes that do not need to be displayed are put away, thereby realizing the ability to analyze problems through the relationship graph, and expressing the information contained in the knowledge graph more intuitively and completely.

A knowledge graph visualization method based on the mechanical principle of the present invention takes the entities of the knowledge graph of structured data as nodes and the relationships as edges, thereby realizing the visual expression of the knowledge graph. Further, each relationship is expressed independently based on the mechanical principle, thereby realizing the complete expression of the information in the knowledge graph. In addition, through the scaling of nodes, the nodes that do not need to be displayed are put away, so as to realize the ability to analyze problems through the relationship graph, and express the information contained in the knowledge graph more intuitively and completely.

The present invention has been described in detail above with reference to specific exemplary embodiments. However, it should be understood that various modifications and variations can be made without departing from the scope of the present invention as defined by the appended claims. The detailed description and drawings are to be regarded in an illustrative rather than a restrictive sense, and if any such modifications and variations exist, they will fall within the scope of the invention described herein. In addition, the background art is intended to illustrate the research and development status and significance of the present technology, and is not intended to limit the present invention or the application and application fields of the present invention.

Claims (9)

1. A knowledge graph visualization method based on a mechanical principle is characterized by comprising the following steps:

converting triple data of the knowledge graph into a relation set and an entity set; the relationship set comprises edge data, and the entity set comprises node data;

marking the node data of the side data sum;

respectively carrying out duplicate removal on the marked edge data and node data;

calculating the length of the edge, the intersection point coordinate of the node and the edge name coordinate according to the node data and the edge data;

and rendering the nodes and the edges by using d3.js according to the length of the edges, the coordinates of the intersection points of the nodes and the edges and the coordinates of the edge names to obtain the visual knowledge graph.

2. The knowledge graph visualization method based on the mechanical principle as claimed in claim 1, wherein the specific process of marking the edge data and the node data is as follows: and performing association query on the node data, and marking the queried node and the associated edge, wherein the queried node set is idlist, and the edge and node marking set is labellist.

3. A method of knowledge-graph visualization based on mechanics principles according to claim 1 characterized by that, the length of the edge is calculated by the following formula:

link.length＝link.name.length*fontSize+n

where, link.length represents the length of the edge, link.name.length represents the length of the edge name, fontSize represents the font size, and n represents a constant.

4. The method for visualizing the knowledge graph based on the mechanics principle of claim 1, wherein the specific process of calculating the coordinates of the intersection points of the nodes and the edges is as follows: the edge data comprises an initial node and an end node, and intersection point coordinates of the nodes and the edges are obtained through calculation according to circle center coordinates of the initial node and the end node.

5. The method of claim 4, wherein when the edge is a circle center connecting line between the start node and the end node, the coordinates of the intersection point of the node and the edge are calculated by the following formula:

when link.target.x is equal to link.source.x:

y1＝link.target.y-link.source.y>0？link.source.y+r:link.source.y–r；

y2＝link.target.y-link.source.y>0？link.target.y-r:link.target.y+r；

x1＝link.source.x；

x2＝link.target.x；

wherein (link.source.x, link.source.y) is the center coordinates of the start node, (link.target.x, link.target.y) is the center coordinates of the end node, r is the radius values of the start node and the end node, (x1, y1) is the intersection coordinates of the edge and the start node, and (x2, y2) is the intersection coordinates of the edge and the end node;

when link.target.y is equal to link.source.y:

y1＝link.source.y；

y2＝link.target.y；

x1＝link.target.x-link.source.x>0？link.source.x+r:link.source.x-r；

x2＝link.target.x-link.source.x>0？link.target.x-r:link.target.x+r；

when link.target.x is not equal to link.source.x and link.target.y is not equal to link.source.y:

wherein,

is the value of the central angle tan between the start node and the end node.

6. The method of claim 4, wherein when the edges are located on both sides of the circle-center connecting line between the start node and the end node, the coordinates of the intersection point of the node and the edge are calculated by the following formula:

calculating the intersection point coordinates (xm, ym) of the intersection point of the passing edge and the initial node, which is perpendicular to the circle center connecting line:

link, number is the number of layers of the side, and maxlink number is the maximum number value of two sides of the circle center;

and (3) calculating the intersection point coordinates (xn, yn) of the intersection point of the passing edge and the termination node to the circle center connecting line as the perpendicular line:

7. the method for visualizing the knowledge graph based on the mechanical principle of any one of claims 1 to 6, wherein the edge name coordinates are calculated by the following formula; let sy be link.source.y, sx be link.source.x, tx be link.target.x, ty be link.target.y;

dx＝(distance-link.name.length*fontSize-2*r)/2

and then calculating dx by using d3.js to obtain dy (dx, dy) as an edge name coordinate.

8. The knowledge graph visualization method based on mechanical principle as claimed in claim 7, wherein the specific process of rendering by using d3.js is:

and verifying the node data and the edge data, and rendering by using d3.js after verification to obtain a visual knowledge graph.

9. The method for knowledge graph visualization based on mechanics principle of claim 8, wherein the specific process of checking the node data and the edge data is as follows:

and checking whether the starting node and the ending node of the edge exist or not, and if any node does not exist, filtering the edge.

Images