Structuring incoherent nodes by superimposing on a base knowledge graph

Patent metadata

Field	Value
Publication number	US-10740398-B2
Application number	US-201715822653-A
Country	US
Kind code	B2
Filing date	Nov 27, 2017
Priority date	Nov 27, 2017
Publication date	Aug 11, 2020
Grant date	Aug 11, 2020

Abstract

A computer program product, system, and method for building a knowledge graph may include receiving a plurality of new nodes, receiving a base knowledge graph having existing nodes selectively connected by existing edges, and superimposing the new nodes onto selected ones of the existing nodes of the base knowledge graph. The method may further include connecting the new nodes by creating a new edge with a new weight between at least two of the new nodes if corresponding existing nodes in the underlying base knowledge graph have a connection via zero or a predetermined maximum number of existing edges, wherein the new weight is determined based on the existing weights of the existing edges of connections between the corresponding existing nodes, and detaching the new nodes with the new edges from the base knowledge graph.

First claim

What is claimed is:

1. A method for building a knowledge graph, the method comprising:
first receiving, by a processor of a computing system, a plurality of new nodes;
second receiving, by the processor, a base knowledge graph comprising existing nodes selectively connected by existing edges, each edge of the existing edges having an existing weight;
superimposing, by the processor, the plurality of new nodes onto selected nodes of the existing nodes of the base knowledge graph, building pairs of new nodes and corresponding existing nodes;
connecting, by the processor, the plurality of new nodes by creating a new edge with a new weight between at least two nodes of the plurality of new nodes if corresponding existing nodes in the underlying base knowledge graph have a connection via zero or a predetermined maximum number of existing edges, wherein the new weight is determined based on the existing weights of the existing edges of connections between the corresponding existing nodes; and
detaching, by the processor, the plurality of new nodes with the new edges from the base knowledge graph as a resulting knowledge graph.

2. The method according to claim 1 , wherein superimposing uses a cognitive computing method for a mapping of content of a new node onto content of an existing node of the base knowledge graph.

3. The method according to claim 1 , wherein each node of the plurality of new nodes comprises content at least partially in text form.

4. The method according to claim 1 , wherein a weight of the new edge between two nodes of the plurality of new nodes is determined by a count of edges forming a shortest connection between the two corresponding nodes of the existing knowledge graph.

5. The method according to claim 1 , wherein a weight of the new edge between two nodes of the plurality of new nodes is determined by a reciprocal value of a count of edges forming a shortest connection between the two corresponding nodes of the existing knowledge graph.

6. The method according to claim 1 , wherein a weight of the new edge between two nodes of the plurality of new nodes is determined by a reciprocal value of a sum of weights of edges forming a shortest connection between the two corresponding nodes of the existing knowledge graph.

7. The method according to claim 1 , further comprising assigning a score to the new node.

8. The method according to claim 7 , wherein the score is determined as a sum of the weights of the edges connecting to the new node.

9. The method according to claim 7 , wherein the score is determined as a count of the edges connecting to the new node.

10. The method according to claim 1 , further comprising building a cluster in the resulting knowledge graph.

11. The method according to claim 10 , wherein a center of the cluster is the new node having a highest count of direct edges to other new nodes.

12. The method according to claim 10 , wherein a center of the cluster is the new node having a node score above a predefined threshold value.

13. The method according to claim 10 , wherein a name for the cluster is identical to a name of a center of the cluster.

14. A system comprising
a processor;
a memory device coupled to the processor; and
a computer readable storage device coupled to the processor, wherein the storage device contains program code executable by the processor via the memory device to implement a method for building a knowledge graph, the method comprising:
first receiving, by a processor of a computing system, a plurality of new nodes;

second receiving, by the processor, a base knowledge graph comprising existing nodes selectively connected by existing edges, each edges of the existing edges having an existing weight;
superimposing, by the processor, the plurality of new nodes onto selected nodes of the existing nodes of the base knowledge graph, building pairs of new nodes and corresponding existing nodes;
connecting, by the processor, the plurality of new nodes by creating a new edge with a new weight between at least two nodes of the plurality of new nodes if corresponding existing nodes in the underlying base knowledge graph have a connection via zero or a predetermined maximum number of existing edges, wherein the new weight is determined based on the existing weights of the existing edges of connections between the corresponding existing nodes; and
detaching, by the processor, the plurality of new nodes with the new edges from the base knowledge graph as a resulting knowledge graph.

15. The system according to claim 14 , wherein the superimposing uses a cognitive computing system for a mapping of the content of the plurality of new nodes onto the content nodes of the base knowledge graph.

16. The system according to claim 14 , wherein each of the plurality of new nodes comprises content at least partially in text form.

17. The system according to claim 14 , wherein a weight of a new edge between two nodes of the plurality of new nodes is determined by a count of edges forming a shortest connection between the two corresponding nodes of the existing knowledge graph,
or
wherein a weight of the new edge between two nodes of the plurality of new nodes is determined by a reciprocal value of a count of edges forming a shortest connection between the two corresponding nodes of the existing knowledge graph,
or
wherein a weight of the new edge between two nodes of the plurality of new nodes is determined by a reciprocal value of a sum of a weighted number of edges forming a shortest connection between the two corresponding nodes of the existing knowledge graph.

18. The system according to claim 14 , further comprising assigning a score to the new node.

19. The system according to claim 18 , wherein the score is determined as a sum of the weights of the edges connecting to the new node.

20. The system according to claim 18 , wherein the score is determined as a sum of the edges connecting to the new node.

21. The system according to claim 14 , further comprising building a cluster in the resulting knowledge graph.

22. The system according to claim 21 , wherein a center of the cluster is the new node having a highest number of direct edges to other new nodes.

23. The system according to claim 21 , wherein a center of the cluster is the new node having a node score above a predefined threshold value.

24. The method according to claim 21 , wherein a name for the cluster is identical to a name of a center of the cluster.

25. A computer program product, comprising a computer readable hardware storage device storing a computer readable program code, the computer readable program code comprising an algorithm that when executed by a computer processor of a computing system implements a method for building a knowledge graph, the method comprising: first receiving, by a processor of a computing system, a plurality of new nodes;
second receiving, by the processor, a base knowledge graph comprising existing nodes selectively connected by existing edges, each edges of the existing edges having an existing weight;
superimposing, by the processor, the plurality of new nodes onto selected nodes of the existing nodes of the base knowledge graph, building pairs of new nodes and corresponding existing nodes;
connecting, by the processor, the plurality of new nodes by creating a new edge with a new weight between at least two nodes of the plurality of new nodes if corresponding exi

Assignees

• IBM

Inventors

• Scheideler Tim U
• Ravizza Stefan
• Giovannini Andrea
• Haxhaj Avdyl
• Streit Simon
• Graf Florian

Classifications

• G06N20/00 Primary

  Machine learning · CPC title

• G06N5/022 Primary

  Knowledge engineering; Knowledge acquisition · CPC title

• G06F16/30

  of unstructured textual data (document management systems G06F16/93) · CPC title

• G06F16/907

  Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually · CPC title

• G06F16/9024 Primary

  Graphs; Linked lists (G06F16/9027 takes precedence) · CPC title