Industrial brain data analysis platform based on industrial knowledge graph

Abstract

The invention discloses an industrial brain data analysis platform based on an industrial knowledge graph, which belongs to the technical field of relay protection and comprises an enterprise data collection module, an industrial graph construction module, an upstream and downstream association matching module and a priority recommendation judgment module. The enterprise data collection module collects data information of enterprises entering the platform, the data information comprises enterprise names, enterprise operation licenses, enterprise locations, enterprise invoices, drafts and check data, archives of the enterprises are established through # imgabs0 #, the data information of the enterprises entering the platform is recorded, and the data are transmitted to the industrial map construction module. According to the industrial brain data analysis platform based on the industrial knowledge graph, the whole industrial brain data analysis platform based on the industrial knowledge graph can realize enterprise classification, electronic bill data extraction and upstream and downstream enterprise association, the enterprises entering the platform can conveniently explore potential suppliers and dealers, and the practicability and functionality of the platform are enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of electronic data processing, and specifically is an industrial brain data analysis platform based on industrial knowledge graph. Background Art

[0002] Electronic bills, also known as electronic tickets, are a form of voucher that digitizes traditional paper bills. They use computer technology and network communication methods to record and store bill information in digital form to achieve bill generation, transmission, payment, redemption, settlement, and archiving. With the rapid development of information technology, transactions between enterprises are becoming increasingly frequent and complex. Electronic bills such as invoices, bills of exchange, and checks have become an important part of corporate financial management. Electronic invoices not only record detailed transaction information, but also include rich business data, such as transaction amount, type of purchased goods, transaction time, etc. The existing data processing and analysis platform cannot associate enterprises with upstream and downstream companies based on the electronic invoice information of business transactions and the type of enterprise. At the same time, it cannot assist enterprises in selecting suppliers and distributors based on their own scale, and cannot deeply explore the potential value of data. There are problems of low practicality and functionality; In response to the above, this case proposes an industrial brain data analysis platform based on industrial knowledge graph to solve the above technical problems. Summary of the invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes an industrial brain data analysis platform based on an industrial knowledge graph to solve the above technical problems by improving the detection method and processing method.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions: The industry brain data analysis platform based on the industry knowledge graph includes enterprise data collection module, industry graph construction module, upstream and downstream association matching module, and priority recommendation and judgment module; The enterprise data collection module collects data information of enterprises settled in the platform, including enterprise name, enterprise business license, enterprise location, enterprise invoice, bill of exchange, and check data. Establish enterprise files, record the data information of settled enterprises, and transfer data into the industrial map construction module; The industrial map construction module classifies the bill data of enterprises based on the types of enterprises, extracts different data of different bills for different types of enterprises, generates labels and constructs an industrial map; The upstream and downstream association matching module combines the electronic bill data information of the enterprise itself to determine the enterprise credit rating, conducts upstream and downstream association matching analysis on the enterprise, and generates upstream files and downstream files of different enterprises; The priority recommendation determination module combines the recent electronic invoice information of the enterprises in the upstream and downstream archives to calculate the purchasing power and productivity level of the enterprises, and ranks the enterprises in the upstream and downstream archives based on the monthly order quantity of the enterprises in the archives.

[0005] Furthermore, the industrial map construction module classifies the bill data of enterprises based on the types of enterprises, extracts different data of different bills for different types of enterprises, generates labels and constructs an industrial map, and the specific steps are as follows: Based on the business licenses of the enterprises settled in the platform collected by the enterprise data collection module, the enterprises are divided into different types, including basic raw material industry, manufacturing industry, financial services industry, commercial services industry, and other industries. The scope of electronic bill information collection for different types of enterprises is determined, and the characters in the electronic bills are extracted through OCR technology. Based on the keyword matching in NLP technology combined with the enterprise type, the key fields are extracted respectively, including: For the basic raw materials industry, extract the product name, amount, date, customer name, and bill status from the sales invoice in the company's historical electronic bills; For the manufacturing industry, extract the product name, amount, date, customer name in the sales invoice, the product name, amount, date, customer name, and bill status in the purchase invoice from the company's historical electronic bills; For the financial services industry, extract the amount of bills and checks from the company's historical electronic bills; For commercial services, extract the product name, amount, date, customer name, and bill status from the purchase invoice in the company's historical electronic bills; For other industries, electronic invoice data extraction is not performed; Combine the invoice data extracted from different types of enterprises on the platform to generate enterprise labels and establish an industry map.

[0006] Furthermore, the invoice data extracted from different types of enterprises in the platform are combined to generate enterprise tags to establish an industry map, and the specific steps are: Based on the known product names, a keyword list is created for each category. Through the fuzzy matching method, the Jaccard similarity is used to match the keyword list of data extracted from the electronic invoices of different types of enterprises, where: For the basic raw materials industry, product type labels are divided according to the commodity names in the sales invoice, including metals and ores, non-metallic ores, chemical raw materials, building materials, agricultural raw materials, energy raw materials, rare and precious raw materials, and other basic raw materials. Based on the commodity names in the sales invoices of enterprises, product type labels are recorded in the enterprise files of the basic raw materials industry; For the manufacturing industry, the raw material type labels are classified according to the commodity names in the purchase invoice, including metals and ores, non-metallic ores, chemical raw materials, building materials, agricultural raw materials, energy raw materials, rare and precious raw materials, and other basic raw materials. At the same time, the product type labels are classified according to the commodity names in the sales invoice, including machinery and equipment, electronic and electrical equipment, automobiles and parts, chemical products, metal products, building materials and home furnishings, textiles and clothing, food and packaging, medical devices, and other manufacturing products. Based on the commodity names in the enterprise's purchase invoices and sales invoices, the raw material type labels and product type labels are recorded in the manufacturing enterprise archives; For the commercial service industry, product type labels are divided according to the commodity names in the purchase invoice, including machinery and equipment, electronic and electrical equipment, automobiles and parts, chemical products, metal products, building materials and home furnishings, textiles and clothing, food and packaging, medical devices, and other manufacturing categories, and the product type labels are recorded in the commercial service industry enterprise files.

[0007] Furthermore, the upstream and downstream association matching module determines the credit rating of the enterprise by combining the electronic bill data information of the enterprise itself, performs upstream and downstream association matching analysis on the enterprise, and generates upstream files and downstream files of different enterprises. The specific steps are as follows: For enterprises in the basic raw materials industry, manufacturing industry and commercial services industry, the credit rating of the enterprise is determined based on the status of the bill of exchange; For enterprises in the basic raw materials industry, manufacturing industry, financial services industry and commercial services industry, we combine the enterprise's electronic invoice data information and credit rating, conduct upstream and downstream correlation matching analysis on enterprises within the platform based on enterprise type, and generate upstream and downstream files for different enterprises.

[0008] Furthermore, for enterprises in the basic raw material industry, manufacturing industry and commercial service industry, the enterprise credit rating is determined based on the bill status, and the specific steps are as follows: Based on the electronic invoice data identified in the industrial graph building module, the above types of enterprises are collected During the year and to The bill data status of the year is used to classify the credit rating of the current enterprise. The bill data status includes overdue and normal. The specific steps are as follows: When the enterprise The bill of exchange data status is not overdue during the year, and to If the annual bill of exchange data status is not overdue, the enterprise label will be recorded as excellent credit in the enterprise file; When the enterprise The bill of exchange data status is not overdue during the year, and to If the annual bill of exchange data status is overdue, the enterprise label will be recorded as general credit in the enterprise file; When the enterprise If the bill of exchange data status is overdue during the year, the enterprise label will be recorded as poor credit in the enterprise file.

[0009] Furthermore, for the basic raw material industry, manufacturing industry, financial service industry and commercial service industry enterprises, combined with the enterprise electronic bill data information, enterprise purchasing power, productivity and credit rating, the upstream and downstream correlation matching analysis of enterprises in the platform is performed based on the enterprise type, and the upstream and downstream files of different enterprises are generated for different enterprises. The specific steps are: For the financial services industry, the platform collects the data on the amount of bills and checks within two years of the financial services industry enterprises, obtains the maximum bill amount, the minimum bill amount, the maximum check amount, and the minimum check amount, and calculates the maximum and minimum transaction values ​​of the current enterprise: ; ; Combine the current enterprise's maximum and minimum transaction values ​​to obtain the enterprise customer range : ; Count the purchase invoice amounts of the basic raw materials industry, manufacturing industry, and commercial services industry within 2 years, obtain the maximum purchase amount, the minimum purchase amount, and the enterprise transaction amount range According to the current financial services industry , establish enterprise downstream archives, The enterprise files are summarized into the downstream files of current financial service enterprises; For the manufacturing industry, downstream and upstream files are established in the enterprise files respectively. Based on the raw material type tags in the enterprise files, the enterprise files with product type tags that are the same as the raw material type tags are divided into the upstream files of the current enterprise. Based on the product type tags in the enterprise files, the enterprise files with raw material type tags that are the same as the product type tags are divided into the downstream files of the current enterprise. For the basic raw material industry, a downstream file is established in the enterprise file. Based on the product type label in the enterprise file, the enterprise file with the raw material type label that is the same as the product type label is classified and entered into the downstream file of the current enterprise; For commercial service industry, upstream files are established in enterprise files. Based on the raw material type tags in the enterprise files, enterprise files with product type tags that are the same as the raw material type tags are classified and entered into the upstream files of the current enterprise. Based on the current corporate credit rating, the upstream and downstream files of the enterprise are further screened. The specific steps are as follows: When the current enterprise file label has an excellent credit, all enterprise files in the upstream or downstream files of the current enterprise are retained; When the credit in the current enterprise file label is average, the enterprise label in the upstream or downstream file of the current enterprise is retained as an average or poor credit enterprise; When the credit in the current enterprise file label is poor, the enterprise label with poor credit in the upstream file or downstream file of the current enterprise is retained.

[0010] Furthermore, the priority recommendation determination module combines the recent electronic invoice information of the enterprise in the upstream and downstream archives, calculates the purchasing power and productivity level of the enterprise, and ranks the enterprises in the upstream and downstream archives in combination with the monthly order quantity of the enterprise in the archives. The specific steps are as follows: Based on the amounts in the purchase and sales invoices of the basic raw material industry, manufacturing industry, and commercial service industry within the platform, the purchasing power and productivity of the enterprise are determined. The specific steps are as follows: For the basic raw material industry and manufacturing industry, the sales invoice amount of enterprises with the same product type label in the enterprise archives on the platform within 2 years is counted, the average annual sales volume is calculated, and the enterprise productivity level is divided based on the distribution of the amount. The specific algorithm formula is: Calculate the average annual sales for companies under different product type labels : ; Based on the average annual sales of companies with the same product type label on the platform, obtain the maximum annual sales Minimum annual sales , and divide the interval width : ; in, Represents the number of segments, according to and , calculate the upper and lower limits of each productivity level, and the algorithm formula is: ; ; Based on the average annual sales of different enterprises, the enterprises are divided into different ranges, and the upper and lower limits of the ranges are recorded in the enterprise files; For the manufacturing and commercial services industries, the purchase invoice amount of enterprises with the same product type label in the enterprise archives on the platform within 2 years is counted, the average annual purchase amount is calculated, and the purchasing power level of enterprises is divided based on the distribution of the amount. The specific algorithm formula is as follows: Calculate the average annual purchase amount for enterprises under different product type labels : ; Based on the average annual purchase amount of companies with the same product type label on the platform, obtain the maximum annual purchase amount Minimum annual purchase amount , and divide the interval width : ; in, Represents the number of segments, according to and , calculate the upper and lower limits of each purchasing power level, and the algorithm formula is: ; ; Based on the average annual purchase amount of different enterprises, the enterprises are divided into different ranges, and the upper and lower limits of the ranges are recorded in the enterprise files; Based on the purchasing power and productivity level of enterprises, combined with the product information of enterprise purchase invoices and sales invoices, the upstream or downstream files of enterprises in the basic raw materials industry, manufacturing industry, and commercial service industry are screened twice, and the enterprises in the upstream and downstream files are ranked based on the number of monthly orders of the enterprises in the files.

[0011] Furthermore, based on the purchasing power and productivity level of the enterprise, combined with the product information of the enterprise's purchase invoice and sales invoice, the upstream files or downstream files of the basic raw material industry, manufacturing industry, and commercial service industry are screened twice, and the correlation score is calculated in combination with the enterprise's transaction frequency, and the upstream files or downstream files retained in the enterprise files are arranged in descending order of correlation score. The specific steps are: For upstream archives of manufacturing and commercial service enterprises, extract the commodity names in the current enterprise purchase invoices and the commodity names in the enterprise sales invoices in the upstream archives, and establish a collection of enterprise purchase commodities And the collection of goods sold by each enterprise in the upstream archive ,in Remove duplicate product names from the current company's purchase invoices. Remove duplicate product names from sales invoices of various enterprises in the upstream files of the current enterprise. is the number of enterprises in the upstream archive, when If it is not an empty set, the relevant enterprises are retained in the upstream archive. If it is an empty set, the relevant enterprises will be removed from the upstream archives; For the downstream archives of basic raw material industries and manufacturing enterprises, extract the commodity names in the current enterprise sales invoices and the commodity names in the enterprise purchase invoices in the downstream archives, and establish the enterprise sales commodity collection And the collection of goods purchased by each enterprise in the downstream archives ,in Remove duplicate product names from the current company's sales invoices. Remove duplicate product names from purchase invoices of various companies in the current company's downstream files. is the number of enterprises in the downstream archives, when If it is not an empty set, the relevant enterprises are retained in the downstream archive. If it is an empty set, the relevant enterprises will be removed from the downstream archives; The number of sales invoices of different retained enterprises in the upstream files within 2 years and the number of purchase invoices of different retained enterprises in the downstream files within 2 years are counted respectively, and the average number of monthly orders is calculated. The algorithm formula is: For upstream archives: ; For downstream archives: ; For the enterprises in the upstream files, the different enterprises in the upstream files are classified according to the productivity level, and in each classification, they are arranged in descending order based on the average monthly order size of different enterprises; For the enterprises in the downstream files, different enterprises in the downstream files are classified according to the purchasing power level, and in each classification, they are arranged in descending order based on the average monthly order size of different enterprises.

[0012] Compared with the prior art, the present invention has the following beneficial effects: 1. In the present invention, based on the type of electronic bills of enterprises, different data are extracted by selecting different types of electronic bills for different types of enterprises, so as to label and record the enterprise files within the platform, obtain the industrial map, assist the enterprises within the platform to establish upstream and downstream associations with each other, and extract the transaction behaviors and business models of enterprises; 2. In the present invention, by acquiring the bill information of enterprises within the platform, establishing upstream and downstream files in combination with the types of different enterprises, and screening them according to credit rating and commodity name, it is ensured that the upstream and downstream files meet the actual needs of the current enterprises. At the same time, for the upstream files of the enterprises, by sorting by productivity level and average monthly order number, it can assist the enterprises to give priority to potential suppliers with high productivity and stable order volume. For the downstream files of the enterprises, by sorting by purchasing power level and average monthly order number, it can assist the enterprises to discover high-quality customers; 3. In the present invention, by automatically identifying and matching upstream and downstream enterprises for settled enterprises based on the electronic invoice data of enterprises within the platform, manual intervention is reduced, the upstream and downstream relationships between enterprises are identified, and the supply chain network of enterprises is helped to be built. At the same time, enterprises can optimize their own supply chain structure based on the upstream and downstream relationships provided by the platform, select more suitable suppliers or customers, reduce costs, and enhance the practicality and functionality of the platform; 4. In the present invention, by targeted analysis of different types of enterprises, the bill data of basic raw materials industry, manufacturing industry, financial services industry, commercial services industry and other industries are analyzed in a targeted manner to mine information that better meets the actual needs of enterprises, and the matching analysis of upstream and downstream enterprises is carried out in combination with the bill scale, purchased goods and credit status of the enterprise itself, so as to objectively and systematically evaluate potential suppliers and distributors for the enterprises entering the platform; The entire industrial brain data analysis platform based on the industrial knowledge graph can realize enterprise classification, electronic invoice data extraction, and upstream and downstream enterprise linkage, making it convenient for companies on the platform to discover potential suppliers and distributors, thereby enhancing the practicality and functionality of the platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a block diagram of the industrial brain data analysis platform based on the industrial knowledge graph of the present invention. DETAILED DESCRIPTION

[0014] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0015] Example 1: Figure 1 As shown in the figure, the industrial brain data analysis platform based on the industrial knowledge graph includes an enterprise data collection module, an industrial graph construction module, an upstream and downstream association matching module, and a priority recommendation judgment module; The enterprise data collection module collects data information of enterprises on the platform, including enterprise name, business license, enterprise location, enterprise invoice, bill of exchange, and check data. Establish enterprise files, record the data information of settled enterprises, and transfer data into the industrial map construction module; The industrial graph construction module classifies the bill data of enterprises based on their types, extracts different data of different bills for different types of enterprises, generates labels and constructs industrial graphs. The specific steps are as follows: Based on the business licenses of the enterprises settled in the platform collected by the enterprise data collection module, the enterprises are divided into different types, including basic raw material industry, manufacturing industry, financial services industry, commercial services industry, and other industries. The scope of electronic bill information collection for different types of enterprises is determined, and the characters in the electronic bills are extracted through OCR technology. Based on the keyword matching in NLP technology combined with the enterprise type, the key fields are extracted respectively, including: For the basic raw materials industry, extract the product name, amount, date, customer name, and bill status from the sales invoice in the company's historical electronic bills; For the manufacturing industry, extract the product name, amount, date, customer name in the sales invoice, the product name, amount, date, customer name, and bill status in the purchase invoice from the company's historical electronic bills; For the financial services industry, extract the amount of bills and checks from the company's historical electronic bills; For commercial services, extract the product name, amount, date, customer name, and bill status from the purchase invoice in the company's historical electronic bills; For other industries, electronic invoice data extraction is not performed; It should be noted that by using Tesseract OCR to convert bill images or PDF files into text, different types of keywords and regular expressions are defined according to the type of enterprise, and keywords and regular expressions are used to match and extract key data from the text extracted by OCR. The basic raw material industry is an enterprise that mainly engages in the mining, processing and sales of raw materials, such as mining companies, agricultural companies, forestry companies, etc. The main daily business direction is to provide raw materials to downstream enterprises. The manufacturing industry is an enterprise that produces products by processing raw materials, such as machinery manufacturing, electronic equipment, clothing production, etc. The main daily business direction is to provide products to downstream enterprises and obtain raw materials from upstream enterprises. The financial services industry is an enterprise that provides financial services, such as banks, insurance, securities, etc. The main daily business direction is to provide checks, bills, etc. The commercial services industry is an enterprise that provides commercial services, such as retail, wholesale, logistics, etc. The main business direction is to obtain products from upstream enterprises for sale. Other industries such as software development, Internet services, education, medical care, etc., because their business models are quite different from those of the manufacturing industry, basic raw material industry or financial services industry, usually involve service contracts, technology development contracts, software license agreements, etc., involving less electronic bill data.

[0016] Combine the bill data extracted from different types of enterprises on the platform to generate enterprise tags and establish an industry map. The specific steps are as follows: Based on the known product names, a keyword list is created for each category. The fuzzy matching method is used to match the keyword list of data extracted from electronic invoices of different types of enterprises using Jaccard similarity, where: For the basic raw materials industry, product type labels are divided according to the commodity names in the sales invoice, including metals and ores, non-metallic ores, chemical raw materials, building materials, agricultural raw materials, energy raw materials, rare and precious raw materials, and other basic raw materials. Based on the commodity names in the sales invoices of enterprises, product type labels are recorded in the enterprise files of the basic raw materials industry; For the manufacturing industry, the raw material type labels are classified according to the commodity names in the purchase invoice, including metals and ores, non-metallic ores, chemical raw materials, building materials, agricultural raw materials, energy raw materials, rare and precious raw materials, and other basic raw materials. At the same time, the product type labels are classified according to the commodity names in the sales invoice, including machinery and equipment, electronic and electrical equipment, automobiles and parts, chemical products, metal products, building materials and home furnishings, textiles and clothing, food and packaging, medical devices, and other manufacturing products. Based on the commodity names in the enterprise's purchase invoices and sales invoices, the raw material type labels and product type labels are recorded in the manufacturing enterprise archives; For the commercial service industry, product type labels are divided according to the commodity names in the purchase invoice, including machinery and equipment, electronic and electrical equipment, automobiles and parts, chemical products, metal products, building materials and home furnishings, textiles and clothing, food and packaging, medical devices, and other manufacturing categories, and the product type labels are recorded in the commercial service industry enterprise files.

[0017] It should be noted that metals and ores include but are not limited to iron ore, copper ore, and aluminum ore; non-metallic ores include but are not limited to limestone and quartz sand; chemical raw materials include but are not limited to petrochemical raw materials, organic chemical raw materials, and inorganic chemical raw materials; building materials include but are not limited to cement and sand and gravel; energy raw materials include but are not limited to coal and oil; rare and precious raw materials include but are not limited to rare earth elements and diamonds; other basic raw materials include but are not limited to water resources and salts; Mechanical equipment includes but is not limited to machine tools and compressors; electronic and electrical equipment includes but is not limited to refrigerators, washing machines, air conditioners, and televisions; automobiles and parts include but are not limited to passenger cars, commercial vehicles, engines, and gearboxes; chemical products include but are not limited to plastic products and rubber products; metal products include but are not limited to steel pipes, steel plates, and steel bars; building materials and home furnishings include but are not limited to doors and windows, floors, and tiles; textiles and clothing include but are not limited to cotton yarn, chemical fibers, and wool spinning; food and packaging include but are not limited to canned food, frozen food, and beverages; medical devices include but are not limited to CT machines, medical device accessories, and medical gauze; other manufacturing categories include but are not limited to office supplies, sports equipment, toys, and handicrafts.

[0018] Embodiment 2: The upstream and downstream association matching module combines the electronic bill data information of the enterprise itself to determine the enterprise credit rating, conducts upstream and downstream association matching analysis on the enterprise, and generates upstream files and downstream files of different enterprises. The specific steps are as follows: For enterprises in the basic raw materials industry, manufacturing industry and commercial service industry, the credit rating of the enterprise is determined based on the status of the bill of exchange. The specific steps are as follows: Based on the electronic invoice data identified in the industrial graph building module, the above types of enterprises are collected During the year and to The bill data status of the year is used to classify the credit rating of the current enterprise. The bill data status includes overdue and normal. The specific steps are as follows: When the enterprise The bill of exchange data status is not overdue during the year, and to If the annual bill of exchange data status is not overdue, the enterprise label will be recorded as excellent credit in the enterprise file; When the enterprise The bill of exchange data status is not overdue during the year, and to If the annual bill of exchange data status is overdue, the enterprise label will be recorded as general credit in the enterprise file; When the enterprise If the bill of exchange data status is overdue during the year, the enterprise label will be recorded as poor credit in the enterprise file.

[0019] It should be noted that The value of is usually 1, that is, the credit rating of enterprises within the platform is divided according to the bill status of the relevant enterprise within 1 year and the bill status within 2-5 years; For enterprises in the basic raw material industry, manufacturing industry, financial services industry and commercial service industry, combined with the enterprise electronic bill data information and credit rating, the upstream and downstream correlation matching analysis of enterprises on the platform is carried out based on the enterprise type, and the upstream and downstream files of different enterprises are generated for different enterprises. The specific steps are as follows: For the financial services industry, the platform collects the data on the amount of bills and checks within two years of the financial services industry enterprises, obtains the maximum bill amount, the minimum bill amount, the maximum check amount, and the minimum check amount, and calculates the maximum and minimum transaction values ​​of the current enterprise: ; ; Combine the current enterprise's maximum and minimum transaction values ​​to obtain the enterprise customer range : ; Count the purchase invoice amounts of the basic raw materials industry, manufacturing industry, and commercial services industry within 2 years, obtain the maximum purchase amount, the minimum purchase amount, and the enterprise transaction amount range According to the current financial services industry , establish enterprise downstream archives, The enterprise files are summarized into the downstream files of current financial service enterprises; It should be noted that The purchasing amounts of related enterprises in the basic raw materials industry, manufacturing industry, and commercial service industry within the platform are within the range of the business handling amounts of current financial services companies within two years, and they are potential downstream customers.

[0020] For the manufacturing industry, downstream and upstream files are established in the enterprise files respectively. Based on the raw material type tags in the enterprise files, the enterprise files with product type tags that are the same as the raw material type tags are divided into the upstream files of the current enterprise. Based on the product type tags in the enterprise files, the enterprise files with raw material type tags that are the same as the product type tags are divided into the downstream files of the current enterprise. For the basic raw material industry, a downstream file is established in the enterprise file. Based on the product type label in the enterprise file, the enterprise file with the raw material type label that is the same as the product type label is classified and entered into the downstream file of the current enterprise; For commercial service industry, upstream files are established in enterprise files. Based on the raw material type tags in the enterprise files, enterprise files with product type tags that are the same as the raw material type tags are classified and entered into the upstream files of the current enterprise. Based on the current corporate credit rating, the upstream and downstream files of the enterprise are further screened. The specific steps are as follows: When the current enterprise file label has an excellent credit, all enterprise files in the upstream or downstream files of the current enterprise are retained; When the credit in the current enterprise file label is average, the enterprise label in the upstream or downstream file of the current enterprise is retained as an average or poor credit enterprise; When the credit in the current enterprise file label is poor, the enterprise label with poor credit in the upstream file or downstream file of the current enterprise is retained.

[0021] It should be noted that by screening the upstream or downstream files of the current enterprise based on the enterprise's creditworthiness, enterprises with higher creditworthiness can be more selective in upstream files or downstream files, which helps to positively promote enterprises to comply with agreements and improve their creditworthiness.

[0022] Embodiment 3: The priority recommendation determination module combines the recent electronic bill information of the enterprise in the upstream and downstream archives to calculate the purchasing power and productivity level of the enterprise, and combines the monthly order quantity of the enterprise in the archive to sort the enterprises in the upstream and downstream archives. The specific steps are as follows: Based on the amounts in the purchase and sales invoices of the basic raw material industry, manufacturing industry, and commercial service industry within the platform, the purchasing power and productivity of the enterprise are determined. The specific steps are as follows: For the basic raw material industry and manufacturing industry, the sales invoice amount of enterprises with the same product type label in the enterprise archives on the platform within 2 years is counted, the average annual sales volume is calculated, and the enterprise productivity level is divided based on the distribution of the amount. The specific algorithm formula is: Calculate the average annual sales for companies under different product type labels : ; Based on the average annual sales of companies with the same product type label on the platform, obtain the maximum annual sales Minimum annual sales , and divide the interval width : ; in, Represents the number of segments, according to and , calculate the upper and lower limits of each productivity level, and the algorithm formula is: ; ; Based on the average annual sales of different enterprises, the enterprises are divided into different ranges, and the upper and lower limits of the ranges are recorded in the enterprise files; For the manufacturing and commercial services industries, the purchase invoice amount of enterprises with the same product type label in the enterprise archives on the platform within 2 years is counted, the average annual purchase amount is calculated, and the purchasing power level of enterprises is divided based on the distribution of the amount. The specific algorithm formula is as follows: Calculate the average annual purchase amount for enterprises under different product type labels : ; Based on the average annual purchase amount of companies with the same product type label on the platform, obtain the maximum annual purchase amount Minimum annual purchase amount , and divide the interval width : ; in, Represents the number of segments, according to and , calculate the upper and lower limits of each purchasing power level, and the algorithm formula is: ; ; Based on the average annual purchase amount of different enterprises, the enterprises are divided into different ranges, and the upper and lower limits of the ranges are recorded in the enterprise files; It should be noted that The value is usually 1. The number of segments can refer to the standards in different industries and can be set by consulting experts in related industries. The productivity and purchasing power of different types of enterprises are divided into intervals and marked with upper and lower limits to further optimize the upstream and downstream files of enterprises.

[0023] Based on the purchasing power and productivity level of enterprises, combined with the product information of enterprise purchase invoices and sales invoices, secondary screening is carried out on the upstream or downstream files of enterprises in the basic raw material industry, manufacturing industry, and commercial service industry. Combined with the monthly order quantity of enterprises in the files, the enterprises in the upstream and downstream files are sorted. The specific steps are as follows: For upstream archives of manufacturing and commercial service enterprises, extract the commodity names in the current enterprise purchase invoices and the commodity names in the enterprise sales invoices in the upstream archives, and establish a collection of enterprise purchase commodities And the collection of goods sold by each enterprise in the upstream archive ,in Remove duplicate product names from the current company's purchase invoices. Remove duplicate product names from sales invoices of various enterprises in the upstream files of the current enterprise. is the number of enterprises in the upstream archive, when If it is not an empty set, the relevant enterprises are retained in the upstream archive. If it is an empty set, the relevant enterprises will be removed from the upstream archives; For the downstream archives of basic raw material industries and manufacturing enterprises, extract the commodity names in the current enterprise sales invoices and the commodity names in the enterprise purchase invoices in the downstream archives, and establish the enterprise sales commodity collection And the collection of goods purchased by each enterprise in the downstream archives ,in Remove duplicate product names from the current company's sales invoices. Remove duplicate product names from purchase invoices of various companies in the current company's downstream files. is the number of enterprises in the downstream archives, when If it is not an empty set, the relevant enterprises are retained in the downstream archive. If it is an empty set, the relevant enterprises will be removed from the downstream archives; It should be noted that by establishing a collection of the enterprise's products and the products in the upstream or downstream archives, it can be ensured that the enterprises in the upstream or downstream archives have business overlaps with the current enterprise.

[0024] The number of sales invoices of different retained enterprises in the upstream files within 2 years and the number of purchase invoices of different retained enterprises in the downstream files within 2 years are counted respectively, and the average number of monthly orders is calculated. The algorithm formula is: For upstream archives: ; For downstream archives: ; For the enterprises in the upstream files, the different enterprises in the upstream files are classified according to the productivity level, and in each classification, they are arranged in descending order based on the average monthly order size of different enterprises; For the enterprises in the downstream files, different enterprises in the downstream files are classified according to the purchasing power level, and in each classification, they are arranged in descending order based on the average monthly order size of different enterprises.

[0025] It should be noted that for the upstream files of the enterprise, by sorting by productivity level and average monthly number of orders, we can give priority to suppliers with high productivity and stable order volume, thereby improving the reliability and efficiency of the supply chain. For the downstream files, by sorting by purchasing power level and average monthly number of orders, we can better understand the purchasing power and needs of customers, and conduct precision marketing and customer classification.

[0026] In the embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation; the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the method of this embodiment.

[0027] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. The industry brain data analysis platform based on the industry knowledge graph is characterized by: It includes enterprise data collection module, industry map construction module, upstream and downstream association matching module, and priority recommendation and judgment module; The enterprise data collection module collects data information of enterprises settled in the platform, including enterprise name, enterprise business license, enterprise location, enterprise invoice, bill of exchange, and check data. Establish enterprise files, record the data information of settled enterprises, and transfer data into the industrial map construction module; The industrial map construction module classifies the bill data of enterprises based on the types of enterprises, extracts different data of different bills for different types of enterprises, generates labels and constructs an industrial map; The upstream and downstream association matching module combines the electronic bill data information of the enterprise itself to determine the enterprise credit rating, conducts upstream and downstream association matching analysis on the enterprise, and generates upstream files and downstream files of different enterprises; The priority recommendation determination module combines the recent electronic invoice information of the enterprises in the upstream and downstream archives to calculate the purchasing power and productivity level of the enterprises, and ranks the enterprises in the upstream and downstream archives based on the monthly order quantity of the enterprises in the archives.

2. The industrial brain data analysis platform based on industrial knowledge graph according to claim 1 is characterized by: The industrial map construction module classifies the bill data of enterprises based on the types of enterprises, extracts different data of different bills for different types of enterprises, generates labels and constructs industrial maps. The specific steps are as follows: Based on the business licenses of the enterprises settled in the platform collected by the enterprise data collection module, the enterprises are divided into different types, including basic raw material industry, manufacturing industry, financial services industry, commercial services industry, and other industries. The scope of electronic bill information collection for different types of enterprises is determined, and the characters in the electronic bills are extracted through OCR technology. Based on the keyword matching in NLP technology combined with the enterprise type, the key fields are extracted respectively, including: For the basic raw materials industry, extract the product name, amount, date, customer name, and bill status from the sales invoice in the company's historical electronic bills; For the manufacturing industry, extract the product name, amount, date, customer name in the sales invoice, the product name, amount, date, customer name, and bill status in the purchase invoice from the company's historical electronic bills; For the financial services industry, extract the amount of bills and checks from the company's historical electronic bills; For commercial services, extract the product name, amount, date, customer name, and bill status from the purchase invoice in the company's historical electronic bills; For other industries, electronic invoice data extraction is not performed; Combine the invoice data extracted from different types of enterprises on the platform to generate enterprise labels and establish an industry map.

3. The industrial brain data analysis platform based on industrial knowledge graph according to claim 2 is characterized by: The specific steps of combining the bill data extracted from different types of enterprises in the platform to generate enterprise tags and establish an industry map are as follows: Based on the known product names, a keyword list is created for each category. The fuzzy matching method is used to match the keyword list of data extracted from electronic invoices of different types of enterprises using Jaccard similarity, where: For the basic raw materials industry, product type labels are divided according to the commodity names in the sales invoice, including metals and ores, non-metallic ores, chemical raw materials, building materials, agricultural raw materials, energy raw materials, rare and precious raw materials, and other basic raw materials. Based on the commodity names in the sales invoices of enterprises, product type labels are recorded in the enterprise files of the basic raw materials industry; For the manufacturing industry, the raw material type labels are classified according to the commodity names in the purchase invoice, including metals and ores, non-metallic ores, chemical raw materials, building materials, agricultural raw materials, energy raw materials, rare and precious raw materials, and other basic raw materials. At the same time, the product type labels are classified according to the commodity names in the sales invoice, including machinery and equipment, electronic and electrical equipment, automobiles and parts, chemical products, metal products, building materials and home furnishings, textiles and clothing, food and packaging, medical devices, and other manufacturing products. Based on the commodity names in the enterprise's purchase invoices and sales invoices, the raw material type labels and product type labels are recorded in the manufacturing enterprise archives; For the commercial service industry, product type labels are divided according to the commodity names in the purchase invoice, including machinery and equipment, electronic and electrical equipment, automobiles and parts, chemical products, metal products, building materials and home furnishings, textiles and clothing, food and packaging, medical devices, and other manufacturing categories, and the product type labels are recorded in the commercial service industry enterprise files.

4. The industrial brain data analysis platform based on industrial knowledge graph according to claim 3 is characterized by: The upstream and downstream association matching module combines the electronic bill data information of the enterprise itself to determine the enterprise credit rating, conducts upstream and downstream association matching analysis on the enterprise, and generates upstream files and downstream files of different enterprises. The specific steps are as follows: For enterprises in the basic raw materials industry, manufacturing industry and commercial services industry, the credit rating of the enterprise is determined based on the status of the bill of exchange; For enterprises in the basic raw materials industry, manufacturing industry, financial services industry and commercial services industry, we combine the enterprise's electronic invoice data information and credit rating, conduct upstream and downstream correlation matching analysis on enterprises within the platform based on enterprise type, and generate upstream and downstream files for different enterprises.

5. The industrial brain data analysis platform based on industrial knowledge graph according to claim 4 is characterized by: For enterprises in the basic raw material industry, manufacturing industry and commercial service industry, the specific steps for determining the enterprise credit rating based on the bill status are as follows: Based on the electronic invoice data identified in the industrial graph building module, the above types of enterprises are collected During the year and to The bill data status of the year is used to classify the credit rating of the current enterprise. The bill data status includes overdue and normal. The specific steps are as follows: When the enterprise The bill data status is not overdue during the year, and to If the annual bill of exchange data status is not overdue, the enterprise label will be recorded as excellent credit in the enterprise file; When the enterprise The bill data status is not overdue during the year, and to If the annual bill of exchange data status is overdue, the enterprise label will be recorded as general credit in the enterprise file; When the enterprise If the bill of exchange data status is overdue during the year, the enterprise label will be recorded as poor credit in the enterprise file.

6. The industrial brain data analysis platform based on industrial knowledge graph according to claim 5 is characterized by: For the basic raw material industry, manufacturing industry, financial service industry and commercial service industry enterprises, combined with the enterprise electronic bill data information, enterprise purchasing power, productivity and credit rating, the upstream and downstream correlation matching analysis of enterprises in the platform is carried out based on the enterprise type, and the upstream and downstream files of different enterprises are generated for different enterprises. The specific steps are as follows: For the financial services industry, the platform collects the data on the amount of bills and checks within two years of the financial services industry enterprises, obtains the maximum bill amount, the minimum bill amount, the maximum check amount, and the minimum check amount, and calculates the maximum and minimum transaction values ​​of the current enterprise: ； ； Combine the current enterprise's maximum and minimum transaction values ​​to obtain the enterprise customer range : ； Count the purchase invoice amounts of the basic raw materials industry, manufacturing industry, and commercial services industry within 2 years, obtain the maximum purchase amount, the minimum purchase amount, and the enterprise transaction amount range According to the current financial services industry , establish enterprise downstream archives, The enterprise files are summarized into the downstream files of current financial service enterprises; For the manufacturing industry, downstream and upstream files are established in the enterprise files respectively. Based on the raw material type tags in the enterprise files, the enterprise files with product type tags that are the same as the raw material type tags are divided into the upstream files of the current enterprise. Based on the product type tags in the enterprise files, the enterprise files with raw material type tags that are the same as the product type tags are divided into the downstream files of the current enterprise. For the basic raw material industry, a downstream file is established in the enterprise file. Based on the product type label in the enterprise file, the enterprise file with the raw material type label that is the same as the product type label is classified and entered into the downstream file of the current enterprise; For commercial service industry, upstream files are established in enterprise files. Based on the raw material type tags in the enterprise files, enterprise files with product type tags that are the same as the raw material type tags are classified and entered into the upstream files of the current enterprise. Based on the current corporate credit rating, the upstream and downstream files of the enterprise are further screened. The specific steps are as follows: When the current enterprise file label has an excellent credit, all enterprise files in the upstream or downstream files of the current enterprise are retained; When the credit in the current enterprise file label is average, the enterprise label in the upstream or downstream file of the current enterprise is retained as an average or poor credit enterprise; When the credit in the current enterprise file label is poor, the enterprise label with poor credit in the upstream file or downstream file of the current enterprise is retained.

7. The industrial brain data analysis platform based on industrial knowledge graph according to claim 1 is characterized by: The priority recommendation determination module combines the recent electronic bill information of enterprises in the upstream and downstream archives, calculates the purchasing power and productivity level of the enterprises, and sorts the enterprises in the upstream and downstream archives based on the monthly order quantity of the enterprises in the archives. The specific steps are as follows: Based on the amounts in the purchase and sales invoices of the basic raw material industry, manufacturing industry, and commercial service industry within the platform, the purchasing power and productivity of the enterprise are determined. The specific steps are as follows: For the basic raw material industry and manufacturing industry, the sales invoice amount of enterprises with the same product type label in the enterprise archives on the platform within 2 years is counted, the average annual sales volume is calculated, and the enterprise productivity level is divided based on the distribution of the amount. The specific algorithm formula is: Calculate the average annual sales for companies under different product type labels : ； Based on the average annual sales of companies with the same product type label on the platform, obtain the maximum annual sales Minimum annual sales , and divide the interval width : ； in, Represents the number of segments, according to and , calculate the upper and lower limits of each productivity level, and the algorithm formula is: ； ； Based on the average annual sales of different enterprises, the enterprises are divided into different ranges, and the upper and lower limits of the ranges are recorded in the enterprise files; For the manufacturing and commercial services industries, the purchase invoice amount of enterprises with the same product type label in the enterprise archives on the platform within 2 years is counted, the average annual purchase amount is calculated, and the purchasing power level of enterprises is divided based on the distribution of the amount. The specific algorithm formula is as follows: Calculate the average annual purchase amount for enterprises under different product type labels : ； Based on the average annual purchase amount of companies with the same product type label on the platform, obtain the maximum annual purchase amount Minimum annual purchase amount , and divide the interval width : ； in, Represents the number of segments, according to and , calculate the upper and lower limits of each purchasing power level, and the algorithm formula is: ； ； Based on the average annual purchase amount of different enterprises, the enterprises are divided into different ranges, and the upper and lower limits of the ranges are recorded in the enterprise files; Based on the purchasing power and productivity level of enterprises, combined with the product information of enterprise purchase invoices and sales invoices, the upstream or downstream files of enterprises in the basic raw materials industry, manufacturing industry, and commercial service industry are screened twice, and the enterprises in the upstream and downstream files are ranked based on the number of monthly orders of the enterprises in the files.

8. The industrial brain data analysis platform based on industrial knowledge graph according to claim 7 is characterized by: Based on the purchasing power and productivity level of enterprises, combined with the product information of enterprise purchase invoices and sales invoices, the upstream files or downstream files of enterprises in the basic raw material industry, manufacturing industry, and commercial service industry are screened twice, and the correlation scores are calculated in combination with the transaction frequency of the enterprises, and the upstream files or downstream files retained enterprises in the enterprise files are arranged in descending order of correlation scores. The specific steps are: For upstream archives of manufacturing and commercial service enterprises, extract the commodity names in the current enterprise purchase invoices and the commodity names in the enterprise sales invoices in the upstream archives, and establish a collection of enterprise purchase commodities And the collection of goods sold by each enterprise in the upstream archive ,in Remove duplicate product names from the current company's purchase invoices. Remove duplicate product names from sales invoices of various enterprises in the upstream files of the current enterprise. is the number of enterprises in the upstream archives, when If it is not an empty set, the relevant enterprises are retained in the upstream archive. If it is an empty set, the relevant enterprises will be removed from the upstream archives; For the downstream archives of basic raw material industries and manufacturing enterprises, extract the commodity names in the current enterprise sales invoices and the commodity names in the enterprise purchase invoices in the downstream archives, and establish the enterprise sales commodity collection And the collection of goods purchased by each enterprise in the downstream archives ,in Remove duplicate product names from the current company's sales invoices. Remove duplicate product names from the purchase invoices of each enterprise in the downstream files of the current enterprise. is the number of enterprises in the downstream archives, when If it is not an empty set, the relevant enterprises are retained in the downstream archive. If it is an empty set, the relevant enterprises will be removed from the downstream archives; The number of sales invoices of different retained enterprises in the upstream files within 2 years and the number of purchase invoices of different retained enterprises in the downstream files within 2 years are counted respectively, and the average number of monthly orders is calculated. The algorithm formula is: For upstream archives: ; For downstream archives: ; For the enterprises in the upstream files, the different enterprises in the upstream files are classified according to the productivity level, and in each classification, they are arranged in descending order based on the average monthly order size of different enterprises; For the enterprises in the downstream files, different enterprises in the downstream files are classified according to their purchasing power levels, and in each classification, they are arranged in descending order based on the average monthly order size of different enterprises.

Images