KNOWLEDGE GRAPH GENERATING APPARATUS, METHOD, AND NON-TRANSITORY COMPUTER READABLE STORAGE MEDIUM THEREOF

• US 20200133962A1
• Filed: 11/30/2018
• Published: 04/30/2020
• Est. Priority Date: 10/25/2018
• Status: Active Grant

• Current Assignee: Institute For Information Industry
• Original Assignee: Institute For Information Industry
• Inventors: KUO, Hsin-Yi, YANG, You-Cyuan, HUANG, Wen-Fa, WANG, Wen-Nan, CHEN, Ping-I
• Granted Patent: US 11,250,035 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/288 Entity relationship models
• CPC Class Codes: G06F 16/9024 Graphs; Linked lists G06F16/9027 takes precedence
• CPC Class Codes: G06F 16/906 Clustering; Classification
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 40/154 Tree transformation for tree-structured or markup documents, e.g. XSLT, XSL-FO or stylesheets
• CPC Class Codes: G06F 40/169 Annotation, e.g. comment data or footnotes
• CPC Class Codes: G06F 40/186 Templates
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06N 5/041 Abduction

Abstract

A knowledge graph generating apparatus, method and non-transitory computer readable storage medium thereof are provided. The apparatus marks an entity-relationship of the template of goods information in the template of webpage according to the operating signal and generates an extraction rule set, wherein the template of webpage is one of multiple goods webpages and has a template format. The apparatus extracts a plurality of first product information of the first goods webpages according to the extraction rule set, wherein the first goods webpages have the template format and are selected from the goods webpages. The apparatus generates a classified goods information result through a product information classification model, wherein the product information classification model is generated based on the first product information and the entity-relationship of the template of goods information. The apparatus converts the classified goods information result into several semantic triples to generate a knowledge graph.

Claims

1. A knowledge graph generating apparatus, comprising:

a network interface, being configured to connect to a plurality of goods webpages;

an operating interface, being configured to generate an operating signal; and

a processor, being electrically connected to the network interface and the operating interface and configured to perform the following operations;

(a) annotating an entity-relationship of a piece of template of goods information in a template of webpage according to the operating signal and generating a first extraction rule set, wherein the template of webpage is one of the goods webpages and the template of webpage has a template format;

(b) connecting to a plurality of first goods webpages through the network interface, and extracting a plurality of pieces of first product information of the first goods webpages according to the first extraction rule set, wherein the first goods webpages have the template format and the first goods webpages are selected from the goods webpages;

(c) generating a first classified goods information result through a product information classification model, wherein the product information classification model is generated based on the first product information and the entity-relationship of the template of goods information; and

(d) converting the first classified goods information result into a plurality of semantic triples to generate a knowledge graph.

2. The knowledge graph generating apparatus of claim 1, wherein the template format is a webpage table or a Cascading Style Sheet (CSS).

3. The knowledge graph generating apparatus of claim 1, wherein the first extraction rule set is related to at least one category of the template format.

4. The knowledge graph generating apparatus of claim 1, wherein the entity-relationship at least contains an entity, a relation and at least one attribute corresponding to the relation, of the template of goods information.

5. The knowledge graph generating apparatus of claim 1, wherein the product information classification model comprises the following operations:

making a word segmentation according to the first product information to generate a word segmentation result; and

extracting a relation result according to the entity-relationship of the template of goods information and the word segmentation result to generate the first classified goods information result, wherein the first classified goods information result indicates the entity-relationship of the first product information.

6. The knowledge graph generating apparatus of claim 1, wherein the processor further executes the following operations:

extracting a piece of second product information of a second goods webpage according to the first extraction rule set, wherein the second goods webpage has the template format;

generating a second classified goods information result through the product information classification model, wherein the second classified goods information result indicates the entity-relationship of the second product information; and

updating the knowledge graph according to the second classified goods information result.

7. The knowledge graph generating apparatus of claim 1, wherein the processor further executes the following operations:

annotating an entity-relationship of a second product in a second template of webpage according to a second operating signal and generating a second extraction rule set, wherein the second template of webpage is one of the goods webpages and the second template of webpage has a second template format; and

generating an extraction rule model according to the first extraction rule set and the second extraction rule set.

8. The knowledge graph generating apparatus of claim 1, wherein the processor further executes the following operations:

re-extracting the first goods webpages according to the first extraction rule set to obtain updated first product information when there is an update version of the first goods webpages;

generating an updated first classified goods information result through the product information classification model; and

updating the knowledge graph according to the updated first classified goods information result.

9. A knowledge graph generating method for use in a knowledge graph generating apparatus, the knowledge graph generating apparatus being configured to generate an operating signal, the knowledge graph generating method being performed by the knowledge graph generating apparatus and comprising:

(a) annotating an entity-relationship of a piece of template of goods information in a template of webpage according to the operating signal and generating a first extraction rule set, wherein the template of webpage is one of a plurality of goods webpages and the template of webpage has a template format;

(b) extracting a plurality of pieces of first product information of a plurality of first goods webpages according to the first extraction rule set, wherein the first goods webpages have the template format and the first goods webpages are selected from the goods webpages;

(c) generating a first classified goods information result through a product information classification model, wherein the product information classification model is generated based on the first product information and the entity-relationship of the template of goods information; and

(d) converting the first classified goods information result into a plurality of semantic triples to generate a knowledge graph.

10. The knowledge graph generating method of claim 9, wherein the template format is a webpage table or a Cascading Style Sheet (CSS).

11. The knowledge graph generating method of claim 9, wherein the first extraction rule set is related to at least one category of the template format.

12. The knowledge graph generating method of claim 9, wherein the entity-relationship at least contains an entity, a relation and at least one attribute corresponding to the relation, of the template of goods information.

13. The knowledge graph generating method of claim 9, wherein the step of generating the first classified goods information result through the product information classification model comprises:

making a word segmentation according to the first product information to generate a word segmentation result; and

extracting a relation result according to the entity-relationship of the template of goods information and the word segmentation result to generate the first classified goods information result, wherein the first classified goods information result indicates the entity-relationship of the first product information.

14. The knowledge graph generating method of claim 9, further comprising:

extracting a piece of second product information of a second goods webpage according to the first extraction rule set, wherein the second goods webpage has the template format;

generating a second classified goods information result through the product information classification model, wherein the second classified goods information result indicates the entity-relationship of the second product information; and

updating the knowledge graph according to the second classified goods information result.

15. The knowledge graph generating method of claim 9, further comprising:

annotating an entity-relationship of a second product in a second template of webpage according to a second operating signal and generating a second extraction rule set, wherein the second template of webpage is one of the plurality of goods webpages and the second template of webpage has a second template format; and

generating an extraction rule model according to the first extraction rule set and the second extraction rule set.

16. The knowledge graph generating method of claim 9, further comprising:

re-extracting the first goods webpages according to the first extraction rule set to obtain updated first product information when there is an update version of the first goods webpages;

generating an updated first classified goods information result through the product information classification model; and

updating the knowledge graph according to the updated first classified goods information result.

17. A non-transitory computer readable storage medium having a computer program stored therein, the computer program executing a knowledge graph generating method after being loaded into an electronic device, the knowledge graph generating method comprising:

(a) annotating an entity-relationship of a piece of template of goods information in a template of webpage according to an operating signal and generating a first extraction rule set, wherein the template of webpage is one of a plurality of goods webpages and the template of webpage has a template format;

(b) extracting a plurality of pieces of first product information of a plurality of first goods webpages according to the first extraction rule set, wherein the first goods webpages have the template format and the first goods webpages are selected from the goods webpages;

(c) generating a first classified goods information result through a product information classification model, wherein the product information classification model is generated based on the first product information and the entity-relationship of the template of goods information; and

(d) converting the first classified goods information result into a plurality of semantic triples to generate a knowledge graph.

18. The non-transitory computer readable storage medium of claim 17, wherein the step of generating the first classified goods information result through the product information classification model comprises:

making a word segmentation according to the first product information to generate a word segmentation result; and

extracting a relation result according to the entity-relationship of the template of goods information and the word segmentation result to generate the first classified goods information result, wherein the first classified goods information result indicates the entity-relationship of the first product information.

19. The non-transitory computer readable storage medium of claim 17, wherein the knowledge graph generating method further comprises:

extracting a piece of second product information of a second goods webpage according to the first extraction rule set, wherein the second goods webpage has the template format;

generating a second classified goods information result through the product information classification model, wherein the second classified goods information result indicates the entity-relationship of the second product information; and

updating the knowledge graph according to the second classified goods information result.

20. The non-transitory computer readable storage medium of claim 17, wherein the knowledge graph generating method further comprises:

re-extracting the first goods webpages according to the first extraction rule set to obtain updated first product information when there is an update version of the first goods webpages;

generating an updated first classified goods information result through the product information classification model; and

updating the knowledge graph according to the updated first classified goods information result.