CN-115248866-B - Knowledge graph-based steel resource retrieval method, system and device

CN115248866BCN 115248866 BCN115248866 BCN 115248866BCN-115248866-B

Abstract

The invention relates to a steel resource retrieval method, system and device based on a knowledge graph, which comprises the steps of constructing and storing a retrieval knowledge graph for retrieving steel commodity codes according to the purpose and/or characteristics of steel, constructing the association relation between the commodity codes of the steel resource and entities, wherein the entities comprise the purpose and the characteristics, acquiring the purpose and/or the characteristics of the steel to be retrieved, and inquiring all commodity codes associated with the purpose and/or the characteristic words from the retrieval knowledge graph. Compared with the prior art, the invention has the advantages of high retrieval efficiency and accuracy and the like.

Inventors

YU ZHIYANG
CHEN MAOJIAN
SHEN HAILUN
LUO XIONG
HUANG ZIYANG

Assignees

欧冶云商股份有限公司
欧冶云商股份有限公司

Dates

Publication Date: 20260421
Application Date: 20220616
Priority Date: 20220616

Claims (8)

1. A steel resource retrieval method based on a knowledge graph is characterized by comprising the following steps: Constructing and storing a retrieval knowledge graph for retrieving the commodity codes of the steel according to the purpose and/or the characteristics of the steel, wherein the retrieval knowledge graph constructs the association relationship between the commodity codes of the steel resource and the entity, and the entity comprises the purpose and the characteristics; Acquiring the purpose and/or characteristics of the steel to be searched, and inquiring all commodity codes associated with the purpose and/or characteristic words from a search knowledge graph, wherein the construction mode of the search knowledge graph comprises the following steps: S1, extracting entities from steel standards, including purposes, features, standard numbers and standard templates, and determining association relations among the entities; s2, constructing a steel use characteristic knowledge graph taking a standard number as a core based on the association relation between the steel standard and the brand, wherein the steel use characteristic knowledge graph associates the standard number, the brand, the use, the characteristics and the standard template; s3, acquiring a steel SKU knowledge graph, wherein the SKU knowledge graph correlates a unique commodity code with the SKU, and the SKU comprises a brand; S4, fusing the steel use characteristic knowledge graph and the SKU knowledge graph based on the brands, and associating the standard numbers, the brands, the uses, the characteristics and the standard templates with the unique commodity codes to form a retrieval knowledge graph, wherein the step S1 comprises the following steps: s11, collecting steel standards, and taking standard models, standard ranges and normative reference files as original data; S12, performing word segmentation and part-of-speech tagging on the raw data in the standard range, using the raw data as input data for training a named entity recognition model, extracting short words by using the named entity recognition model, wherein the short words comprise characteristics and purposes, further constructing a real table of the short words based on the short words, splicing part of the short words into long words to obtain a real table of the long words, and constructing an association relationship between the long words and the short words; s13, extracting entities from the standard model and the normative reference file to obtain a standard template and a version number, and combining the standard template and the version number to obtain an entity table of the standard number; S14, associating the entity table of the short word, the entity table of the long word and the entity table of the standard number to construct an association relationship among the standard number, the long word and the short word; s15, clustering the standard numbers with the same standard templates, and establishing the association relation between the standard numbers and the standard templates.
2. The steel resource retrieval method based on the knowledge graph of claim 1, wherein the named entity recognition model is a named entity recognition model based on BERT.
3. The steel resource retrieval method based on the knowledge graph as claimed in claim 1, wherein the knowledge graph of the steel SKU in the step S3 is a pre-stored knowledge graph, and the SKU further comprises a plating layer amount, a plating layer type, a surface structure, surface treatment and a side shape.
4. The steel resource searching method based on the knowledge graph according to claim 1, wherein the query is performed by adopting a graph query language Cypher when the query is performed in the searching knowledge graph.
5. The method for searching steel resources based on the knowledge graph of claim 1, further comprising displaying all the steel resources with the commodity codes obtained by the query after the query is completed.
6. A steel resource retrieval system based on a knowledge graph, which is characterized in that the system adopts the steel resource retrieval method based on the knowledge graph as set forth in claim 1, and the system comprises: the storage module is used for storing a search knowledge graph for searching the commodity codes of the steel according to the purpose and/or the characteristics of the steel, the search knowledge graph constructs the association relationship between the commodity codes of the steel resource and the entity, and the entity comprises the purpose and the characteristics; And the searching and inquiring module is used for acquiring the purpose and/or the characteristic of the steel to be searched, searching all commodity codes associated with the purpose and/or the characteristic word from the searching knowledge graph, and displaying the resource with the commodity code label.
7. The system for searching for steel resources based on a knowledge graph according to claim 6, further comprising a display module, wherein the display module displays all steel resources with commodity codes obtained by inquiry.
8. A steel resource retrieval device based on a knowledge graph, comprising a memory and a processor, wherein the memory is used for storing a computer program, and the processor is used for realizing the steel resource retrieval method based on the knowledge graph according to any one of claims 1 to 5 when executing the computer program.

Description

Knowledge graph-based steel resource retrieval method, system and device Technical Field The invention relates to the technical field of information retrieval, in particular to a steel resource retrieval method, system and device based on a knowledge graph. Background The stock level units (Stock Keeping Unit, SKU) currently used in the search of the steel business platform comprise 6 dimensions of branding, plating level, plating type, surface structure, surface treatment and edge shape. The chain can be continuously expanded, the granularity of the steel products is continuously refined, and even the fuzzy search can provide more accurate results for users. Steel SKUs can be stored by a variety of means, and if stored using a relational database such as MySQL, there are the following disadvantages: 1. the granularity is too fine, so that the data size is too large, for example, in the current 6 dimensions, 100 values are assumed in each dimension, the length of the table is 100 times of 6, and the data size can be exponentially increased along with the later-stage additional dimension, so that the searching efficiency is greatly reduced; 2. The excessive data volume causes iron and steel specialists responsible for maintaining the SKU data to consume a great deal of effort to check; 3. the clustering relation among the SKUs is displayed through SQL language query, and the clustering relation can only be presented in a numerical form, so that the clustering relation is not intuitive. Therefore, the conventional relational database is used for storage, and the method has no small limitation in terms of query efficiency, maintenance cost and clustering relation query. Disclosure of Invention The invention aims to overcome the defects of the prior art and provide a steel resource retrieval method, system and device based on a knowledge graph, which greatly improve retrieval efficiency and accuracy. The aim of the invention can be achieved by the following technical scheme: a steel resource retrieval method based on a knowledge graph comprises the following steps: Constructing and storing a retrieval knowledge graph for retrieving the commodity codes of the steel according to the purpose and/or the characteristics of the steel, wherein the retrieval knowledge graph constructs the association relationship between the commodity codes of the steel resource and the entity, and the entity comprises the purpose and the characteristics; And acquiring the purpose and/or characteristics of the steel to be searched, and inquiring all commodity codes associated with the purpose and/or characteristic words from the searching knowledge graph. Preferably, the construction method of the search knowledge graph comprises the following steps: S1, extracting entities from steel standards, including purposes, features, standard numbers and standard templates, and determining association relations among the entities; s2, constructing a steel use characteristic knowledge graph taking a standard number as a core based on the association relation between the steel standard and the brand, wherein the steel use characteristic knowledge graph associates the standard number, the brand, the use, the characteristics and the standard template; s3, acquiring a steel SKU knowledge graph, wherein the SKU knowledge graph correlates a unique commodity code with the SKU, and the SKU comprises a brand; S4, fusing the steel use characteristic knowledge graph and the SKU knowledge graph based on the brands, and associating the standard numbers, the brands, the uses, the characteristics and the standard templates with the unique commodity codes to form the retrieval knowledge graph. Preferably, step S1 comprises: s11, collecting steel standards, and taking standard models, standard ranges and normative reference files as original data; S12, performing word segmentation and part-of-speech tagging on the raw data in the standard range, using the raw data as input data for training a named entity recognition model, extracting short words by using the named entity recognition model, wherein the short words comprise characteristics and purposes, further constructing a real table of the short words based on the short words, splicing part of the short words into long words to obtain a real table of the long words, and constructing an association relationship between the long words and the short words; s13, extracting entities from the standard model and the normative reference file to obtain a standard template and a version number, and combining the standard template and the version number to obtain an entity table of the standard number; S14, associating the entity table of the short word, the entity table of the long word and the entity table of the standard number to construct an association relationship among the standard number, the long word and the short word; s15, clustering the standard numbers with the same standard templates, and establishing th