Method and system for constructing product upstream and downstream relationship knowledge graph of instrument and meter enterprise

By constructing a knowledge graph of upstream and downstream relationships of enterprise products on a network collaborative service platform in the instrumentation manufacturing industry, and using Stanford CoreNLP and conditional random field models for entity recognition and relationship extraction, the shortcomings of existing platforms in resource linkage and collaborative manufacturing are solved, and efficient information sharing and collaborative manufacturing between enterprises are realized.

CN118349689BActive Publication Date: 2026-11-10BEIJING INFORMATION SCI & TECH UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410564998.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-09
Publication Date
2026-11-10
Estimated Expiration
2044-05-09

AI Technical Summary

Technical Problem

Existing network collaborative service platforms in the instrument and meter manufacturing industry lack in-depth research on resource linkage and collaborative manufacturing among upstream and downstream enterprises, resulting in insufficient informatization levels.

Method used

The Stanford CoreNLP entity recognition toolkit was used for enterprise entity recognition, and the results were corrected by combining conditional random field model and product entity group. Enterprise product relationships were extracted through relation indicator vocabulary and dependency parsing. The Neo4j database was used to construct a knowledge graph of upstream and downstream relationships of intelligent instrumentation enterprises.

Benefits of technology

It improved the accuracy of product identification and relationship extraction, promoted information sharing among enterprises, and accelerated the efficiency of networked collaborative manufacturing among intelligent instrument and meter enterprises.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118349689B_ABST
    Figure CN118349689B_ABST
Patent Text Reader

Abstract

The application relates to a kind of instrument and meter enterprise product upstream and downstream relationship knowledge graph construction method and system, which include: after the input text vector is preprocessed, enterprise entity recognition is carried out;The input text vector is corrected using the product entity group based on the conditional random field model, to realize product entity recognition;According to the results of enterprise entity recognition and the results of product entity recognition, the supply relationship of upstream and downstream enterprises is extracted for enterprise product relationship;The extracted enterprise entity, product entity and supply relationship between enterprise product are integrated to construct intelligent instrument and meter enterprise upstream and downstream relationship knowledge graph, and the knowledge graph is published.The application can improve the relationship extraction precision and improve the product recognition accuracy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of computer integrated manufacturing technology, and in particular to a method and system for constructing a knowledge graph of upstream and downstream relationships of products in an instrument and meter enterprise. Background Technology

[0002] Information exchange and business coordination among upstream and downstream enterprises in the instrumentation industry are crucial for its informatization. However, due to the large quantity and variety of manufacturing resources, their wide geographical distribution, the significant proportion of unstructured data, and the complexity of many enterprises' internal production information systems, establishing a network-based collaborative manufacturing public service platform for the instrumentation industry is of great importance in connecting upstream and downstream enterprises and improving inter-enterprise collaboration efficiency and business continuity. By leveraging regional network technology to provide collaborative services for collaborative manufacturing among intelligent instrumentation enterprises and to offer collaborative support for product development, the platform can enhance the industry's informatization level and powerfully promote the development of the intelligent instrumentation industry.

[0003] Accurate, fast, and practical service platforms are a crucial foundation for improving the efficiency of network-based collaborative manufacturing. Enterprises upload and share manufacturing resources through these platforms, retrieve resources based on specific needs, and interact with each other online, thereby enhancing collaborative manufacturing efficiency. For example, the China Machinery Industry Federation has established a Manufacturing Collaborative Service Platform. This platform integrates core supporting forces and fundamental research results from my country's science and technology fields, relying on cutting-edge Geographic Information System (GIS) technology, collaborative innovation concepts, and intelligent decision support technologies. Its aim is to promote the informatization of manufacturing and comprehensively improve the level of collaborative services among manufacturing enterprises. The platform includes rich market information on the instrumentation manufacturing industry, relevant research status, and technical training. It also has promotional functions, including showcasing enterprise information, production capacity, technical capabilities, product images, qualification certificates, engineering drawings, enterprise case studies, and enterprise websites. Finally, it has constructed a collaborative sharing and collaborative design knowledge resource base, containing 5 million high-quality resources on manufacturing data, manufacturing processes, automation, general technologies, and enterprise management.

[0004] However, current online collaborative service platforms for the instrument and meter manufacturing industry mainly focus on information announcements and information retrieval, but have not yet conducted in-depth research on related aspects such as resource linkage between upstream and downstream enterprises, collaborative manufacturing, and supply and demand matching. Summary of the Invention

[0005] To address the aforementioned problems, the purpose of this invention is to provide a method and system for constructing a knowledge graph of upstream and downstream relationships among products of instrument and meter enterprises, which can improve the accuracy of relationship extraction and increase the accuracy of product identification.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for constructing a knowledge graph of upstream and downstream relationships of instrument and meter enterprises, comprising: preprocessing the input text vector and then performing enterprise entity recognition; correcting the input text vector using a conditional random field model and product entity groups to achieve product entity recognition; extracting enterprise-product relationships based on the results of enterprise entity recognition and product entity recognition, targeting the supply relationships between upstream and downstream enterprises; integrating the extracted enterprise entities, product entities, and supply relationships between enterprise products to construct a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises, and publishing the knowledge graph.

[0007] Furthermore, the Stanford CoreNLP Entity Recognition Toolkit was used for enterprise entity recognition.

[0008] Furthermore, enterprise entity identification includes:

[0009] The input text vector is preprocessed to remove spaces and punctuation characters.

[0010] The preprocessed text is segmented according to its structure to obtain word sequence.

[0011] Identify enterprise entity names based on phrase sequences.

[0012] Furthermore, the input text vector is refined using a conditional random field model and product entity groups, including:

[0013] The text vector is processed by a feature extraction network to extract word features, part-of-speech features, boundary word features, and dictionary features from the input text. These features are then used as input to the CRF model, which outputs a coarse recognition result of the product entity.

[0014] The input text vector is segmented by word segmentation and part-of-speech tagging. After segmentation, product entity groups are obtained through syntactic analysis. The coarse recognition results of product entities are corrected using product entity groups to obtain the product entity recognition results.

[0015] Furthermore, the CRF model is trained in a supervised manner, including:

[0016] Data augmentation is performed by using entities of similar types within the domain dictionary. The jieba word segmentation tool is used to segment the existing text and replace the entities by analogy to generate a new data entry, thereby expanding the data.

[0017] The output is converted to the CRF++ standard format and used as the training set to train the CRF model. The F-value is used as the evaluation criterion for the CRF model. When the F-value is greater than the preset value α, it proves that the CRF model is effective.

[0018] Furthermore, product relationships between upstream and downstream enterprises are extracted, including:

[0019] By traversing the relational lexicon and the enterprise dictionary, it is determined that the overall input text is related to enterprise supply relationships, and the enterprise set and product name set are divided according to the enterprise and product identification results;

[0020] Based on the set of company names, sentences containing company names are filtered out, and after being judged by preset rules, company entity pairs are generated to identify suppliers and obtain customer company and product corpus.

[0021] The product information is determined, and multiple companies or products appearing in the enterprise set and product name set are obtained through syntactic analysis. The correspondence and matching between customer companies and products are then determined to identify the relationship chain between suppliers, customers and products, and the supply relationship between enterprise entities, product entities and enterprise products is obtained.

[0022] Furthermore, the extracted enterprise entities, product entities, and supply relationships between enterprise products are integrated to construct a knowledge graph of upstream and downstream relationships among intelligent instrument and meter enterprises, including:

[0023] The basic data format consists of <enterprise-relationship-product> triples, which are composed of enterprise entities, product entities, and supply relationships between enterprise products.

[0024] The Neo4j graph database is used to visualize and store knowledge nodes and edges, thus completing the construction of a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises.

[0025] A knowledge graph construction system for upstream and downstream product relationships of instrumentation enterprises includes: an enterprise entity recognition module, which preprocesses the input text vector and then performs enterprise entity recognition; a product entity recognition module, which corrects the input text vector using a conditional random field model and product entity groups to achieve product entity recognition; a relationship extraction module, which extracts enterprise-product relationships based on the results of enterprise entity recognition and product entity recognition, targeting the supply relationships between upstream and downstream enterprises; and a construction and publishing module, which integrates the extracted enterprise entities, product entities, and supply relationships between enterprise products to construct a knowledge graph of upstream and downstream relationships of intelligent instrumentation enterprises and publishes the knowledge graph.

[0026] A computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform any of the methods described above.

[0027] A computing device includes: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for performing any of the methods described above.

[0028] The present invention has the following advantages due to the adoption of the above technical solutions:

[0029] 1. This invention targets intelligent instrument manufacturing resource information and uses a fusion method based on CRF model and product entity group correction to complete product entity recognition, thereby improving the product recognition accuracy.

[0030] 2. This invention uses a combination of relation indicator lexicon and dependency parsing to extract enterprise product relationships, which improves the accuracy of relationship extraction, efficiently obtains the upstream and downstream supply relationship chain of intelligent instrument and meter enterprises, and constructs a knowledge graph of the supply relationship of upstream and downstream enterprises of intelligent instrument and meter. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the process for constructing a knowledge graph of upstream and downstream relationships of instrument and meter enterprises in an embodiment of the present invention;

[0032] Figure 2 This is a schematic diagram of the product entity recognition processing flow in an embodiment of the present invention;

[0033] Figure 3 This is a schematic diagram of the syntactic analysis results in product entity recognition in an embodiment of the present invention;

[0034] Figure 4 This is a schematic diagram of the relationship extraction method in an embodiment of the present invention;

[0035] Figure 5 This is a schematic diagram of the knowledge graph of the supply relationship between upstream and downstream enterprises of intelligent instruments and meters in an embodiment of the present invention. Detailed Implementation

[0036] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0038] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0039] The intelligentization, informatization, and networking of network collaborative service platforms have become a development trend. Massive manufacturing resources and knowledge provide data support for the digitalization and informatization of manufacturing. However, their distributed and discrete nature, significant redundancy, and weak correlations limit the efficiency of system knowledge learning and reasoning. Fully extracting the rich semantic information between various manufacturing resources and fully expressing the relationships between multimodal and multi-source data has become crucial for intelligent network collaborative manufacturing. Knowledge graphs, a new generation of intelligent search engine technology, can realize knowledge storage, relationship retrieval, and knowledge expression functions. They perform excellently in multiple downstream tasks. By organically integrating multimodal data relationships, knowledge graphs are constructed to achieve knowledge storage, relationship retrieval, and knowledge expression. They can fully acquire resource relationships and knowledge content within a specific domain. Relying on third-party software platforms, databases and related retrieval and inference engines are established, and rich service functions are implemented through API interfaces.

[0040] To address the lack of research on upstream and downstream enterprise resource linkage, collaborative manufacturing, and supply-demand matching in existing network collaborative service platforms for the instrumentation manufacturing industry, this invention provides a method and system for constructing a knowledge graph of upstream and downstream relationships among instrumentation enterprises. This involves establishing a regional network collaborative manufacturing service platform, creating an inter-enterprise relationship network, and constructing a knowledge graph of upstream and downstream relationships among intelligent instrumentation enterprises. Stanford CoreNLP is used for enterprise entity recognition. Next, a combination of boundary words, product keywords, and CRF models is employed to identify product entities. Then, a method based on a relation indicator lexicon and dependency parsing is proposed to automatically acquire upstream and downstream relationships. Finally, the Neo4j database is used to store the knowledge graph of upstream and downstream relationships among intelligent instrumentation enterprises, and the knowledge graph is published through a third-party service platform to promote information sharing among enterprises and accelerate the efficiency of networked collaborative manufacturing among intelligent instrumentation enterprises.

[0041] In one embodiment of the present invention, a method for constructing a knowledge graph of upstream and downstream relationships of instrument and meter enterprises is provided, applicable to network collaboration and resource sharing application scenarios in the intelligent instrument and meter manufacturing industry. In this embodiment, the method for constructing a knowledge graph of upstream and downstream relationships of instrument and meter enterprises includes four parts: enterprise entity identification, product entity identification, enterprise-product relationship extraction, and construction of the knowledge graph of upstream and downstream relationships of enterprise products. First, the acquired intelligent instrument and meter manufacturing resources are subjected to enterprise entity identification and product entity identification. Then, the upstream and downstream relationships between enterprises and products are extracted. Finally, a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises is constructed and published on the intelligent instrument and meter regional network collaborative manufacturing service platform system to realize cross-enterprise resource sharing of instruments and meters. Specifically, as shown... Figure 1 As shown, the method includes the following steps:

[0042] 1) After preprocessing the input text vector, enterprise entity recognition is performed;

[0043] 2) The input text vector is corrected using a conditional random field model and product entity groups to achieve product entity recognition;

[0044] 3) Based on the results of enterprise entity identification and product entity identification, extract enterprise-product relationships for the supply relationships of upstream and downstream enterprises;

[0045] 4) Integrate the extracted enterprise entities, product entities, and supply relationships between enterprise products to construct a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises, and publish the knowledge graph.

[0046] In this embodiment, the identification model mentioned below includes enterprise entity identification and product entity identification in steps 1) and 2).

[0047] In step 1) above, this embodiment uses the Stanford CoreNLP Entity Recognition Toolkit for enterprise entity recognition.

[0048] In step 1) above, enterprise entity identification includes the following steps:

[0049] 1.1) Preprocess the input text vector to remove spaces, punctuation marks, and other characters.

[0050] 1.2) Segment the preprocessed text according to its structure to obtain word sequence;

[0051] 1.3) Identify enterprise entity names based on phrase sequences.

[0052] In step 2) above, if Figure 2As shown, the input text vector is corrected using a Conditional Random Field (CRF) model and a product entity group, including the following steps:

[0053] 2.1) The text vector is processed by a feature extraction network to extract word features, part-of-speech features, boundary word features and dictionary features from the input text. These features are then used as input to the CRF model, which outputs a coarse recognition result of the product entity.

[0054] In this embodiment, the CRF model is trained in a supervised manner, including the following steps:

[0055] 2.1.1) To address the issue of limited sample size for intelligent instrument manufacturing resources, data augmentation is performed using entities of similar types within the domain dictionary. The jieba word segmentation tool is used to segment existing text and perform analogy replacement on entities to generate new data, thereby expanding the data.

[0056] 2.1.2) Convert the output to the CRF++ standard format, use it as the training set, and train the CRF model. Use the F-value as the evaluation criterion for the CRF model. When the F-value is greater than the preset value α, it proves that the CRF model is effective.

[0057] Preferably, the preset value α ∈ (80%, 100%).

[0058] Specifically, the basic rules for annotation are as follows: the left boundary label for the product name is B_PRODUCT, the right boundary label is E_PRODUCT, the middle character label between the left and right boundaries is I_PRODUCT, and other non-product name words are labeled O. After annotation is complete, the output is converted to the CRF++ standard format, as shown in Table 1.

[0059] Table 1. Examples of CRF++ Corpus Formats

[0060]

[0061] Where n represents a noun, adv represents an adverb, v represents a verb, adj represents an adjective, Y represents yes, and N represents no.

[0062] Precision (P), recall (R), and F-score were used as evaluation criteria for the recognition model, with the precision threshold α set at 85%.

[0063]

[0064]

[0065]

[0066] The experimental results are shown in Table 2.

[0067] Table 2. Experimental Results of Entity Recognition Model Performance

[0068]

[0069] Wherein, T(W / P) represents using word features and part-of-speech features, T(W / P / B / D) represents adding boundary word features and dictionary features on top of that, and T(W / P / B / D / rule post-processing) represents using all four features and rule post-processing simultaneously. Experimental results show that after correction using syntactic analysis rule post-processing, precision and recall were further increased, with a final F-value of 87.3%, which is greater than 85%, proving its effectiveness.

[0070] 2.2) The input text vector is segmented by word segmentation and part-of-speech tagging. After segmentation, product entity groups are obtained through syntactic analysis. The coarse recognition results of product entities are corrected using product entity groups to obtain the product entity recognition results.

[0071] In this embodiment, the Language Technology Platform (LTP) module is used to segment the input text vector into words and parts of speech.

[0072] In this embodiment, product entity groups are obtained through syntactic analysis after segmentation. Specifically, taking the Beijing Intelligent Instrument Manufacturing Industry Resources as an example, product entity groups are obtained through syntactic analysis. Taking the statement "The company is preparing to purchase electronic products such as photoelectric sensors, laser detectors, power modules, lithium batteries, and switching devices" as an example, the result obtained by rule post-processing is: "company, preparing, adopting, photoelectric / PRODUCT and laser / PRODUCT detectors / PRODUCT and power / PRODUCT modules, e.g., electromagnetic / PRODUCT and switching devices, etc." The two identified product entity groups are {photoelectric sensors, laser detectors} and {power modules, lithium batteries}, and the syntactic analysis result is as follows: Figure 3 As shown.

[0073] In step 3) above, if Figure 4 As shown, the extraction of enterprise product relationships based on the supply relationship between upstream and downstream enterprises includes the following steps:

[0074] 3.1) By traversing the relational lexicon and enterprise dictionary, determine whether the overall input text is related to enterprise supply relationships, and divide the enterprise name set and product name set according to the enterprise and product identification results;

[0075] 3.2) Based on the set of enterprise names, filter out statements containing enterprise names, and generate enterprise entity pairs after determining them according to preset rules, identify suppliers, and obtain customer company and product corpora;

[0076] Both the supplier and the customer companies are corporate entities.

[0077] 3.3) Determine product information, obtain multiple companies or products appearing in the set of company names and product names through syntactic analysis, match and correspond customer companies with products, and then determine the relationship chain between suppliers, customers and products to obtain the supply relationship between enterprise entities, product entities and enterprise products.

[0078] Specifically, in this embodiment, the extraction process is illustrated using the text "Vell Technology and Renesas Electronics are cooperating, and Renesas Electronics plans to supply Vell Technology with 100 self-developed image sensors" as an example. By traversing the relational lexicon and enterprise dictionary, text indicator words are filtered. The keyword "cooperation" leads the model to determine that the statement is related to a supply relationship between enterprises. Steps 1) and 2) identify the enterprises and products. Through word segmentation, part-of-speech and dependency parsing, and utilizing nearest grammatical dependencies, the set of enterprise names is obtained as C = {Vell Technology, Renesas Electronics}, and the set of product names is P = {image sensors}. Then, enterprise entity pairs {Vell Technology, Renesas Electronics} are generated, confirming that the supplier is Renesas Electronics, and product corpus is obtained. Finally, product information is determined. Through syntactic analysis, multiple enterprises or products appearing in the statement are obtained, and the correspondence and matching between customers and products are performed. Finally, the supply chain relationship triple <Renesas Electronics, supply, image sensor> and <Vell Technology, purchase, image sensor> are determined.

[0079] In this embodiment, the F-value is used as the evaluation criterion for the extracted model. When the F-value is greater than β, it proves that the model is effective, where β∈(75%, 100%).

[0080] Precision, recall, and F-score were used as performance evaluation indicators for the enterprise product relationship extraction model. A total of 3 groups were conducted. Group 1 did not use statement filtering, Group 2 did not use the nearest syntactic dependent verb, and Group 3 used both of the above methods. The evaluation precision threshold β was set to 80%. The final experimental results are shown in Table 3.

[0081] Table 3. Experimental Results of Product Relationship Extraction Model

[0082]

[0083] The results show that the method of using relation indicator filtering and nearest grammatical dependency to determine product cascades can achieve an F value of 84.1%, which meets the requirement of more than 80%. Compared with the other two methods, this method can significantly improve the effect of product relation extraction.

[0084] In step 4) above, the extracted enterprise entities, product entities, and supply relationships between enterprise products are integrated to construct a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises, including the following steps:

[0085] 4.1) The basic data format consists of a <firm-relationship-product> triple, which is composed of firm entities, product entities, and the supply relationships between firm products;

[0086] 4.2) The Neo4j graph database is used to visualize and store knowledge nodes and edges, completing the construction of a knowledge graph of upstream and downstream relationships for intelligent instrument and meter enterprises, such as... Figure 5 As shown.

[0087] In this embodiment, the established knowledge graph is published on the software service platform system to promote resource sharing and collaborative manufacturing of intelligent instruments and meters in the regional network.

[0088] In one embodiment of the present invention, a knowledge graph construction system for upstream and downstream relationships of products of instrument and meter enterprises is provided, comprising:

[0089] The enterprise entity recognition module preprocesses the input text vector before performing enterprise entity recognition.

[0090] The product entity recognition module corrects the input text vector using a conditional random field model and product entity groups to achieve product entity recognition.

[0091] The relationship extraction module extracts enterprise-product relationships based on the results of enterprise entity identification and product entity identification, targeting the supply relationships between upstream and downstream enterprises.

[0092] The module for building and publishing integrates the extracted enterprise entities, product entities, and supply relationships between enterprise products to construct a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises, and then publishes this knowledge graph.

[0093] In the above embodiments, the Stanford CoreNLP Entity Recognition Toolkit is used for enterprise entity recognition.

[0094] In this embodiment, specifically, enterprise entity identification includes:

[0095] The input text vector is preprocessed to remove spaces and punctuation characters.

[0096] The preprocessed text is segmented according to its structure to obtain word sequence.

[0097] Identify enterprise entity names based on phrase sequences.

[0098] In the above embodiments, the input text vector is corrected using a conditional random field model and a product entity group, including:

[0099] The text vector is processed by a feature extraction network to extract word features, part-of-speech features, boundary word features, and dictionary features from the input text. These features are then used as input to the CRF model, which outputs a coarse recognition result of the product entity.

[0100] The input text vector is segmented by word segmentation and part-of-speech tagging. After segmentation, product entity groups are obtained through syntactic analysis. The coarse recognition results of product entities are corrected using product entity groups to obtain the product entity recognition results.

[0101] In this embodiment, the CRF model is trained in a supervised manner, including:

[0102] Data augmentation is performed by using entities of similar types within the domain dictionary. The jieba word segmentation tool is used to segment the existing text and replace the entities by analogy to generate a new data entry, thereby expanding the data.

[0103] The output is converted to the CRF++ standard format and used as the training set to train the CRF model. The F-value is used as the evaluation criterion for the CRF model. When the F-value is greater than the preset value α, it proves that the CRF model is effective.

[0104] In the above embodiments, the extraction of enterprise product relationships based on the supply relationship between upstream and downstream enterprises includes:

[0105] By traversing the relational lexicon and the enterprise dictionary, it is determined that the overall input text is related to enterprise supply relationships, and the enterprise set and product name set are divided according to the enterprise and product identification results;

[0106] Based on the set of company names, sentences containing company names are filtered out, and after being judged by preset rules, company entity pairs are generated to identify suppliers and obtain customer company and product corpus.

[0107] The product information is determined, and multiple companies or products appearing in the enterprise set and product name set are obtained through syntactic analysis. The correspondence and matching between customer companies and products are then determined to identify the relationship chain between suppliers, customers and products, and the supply relationship between enterprise entities, product entities and enterprise products is obtained.

[0108] In the above embodiments, the extracted enterprise entities, product entities, and supply relationships between enterprise products are integrated to construct a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises, including:

[0109] The basic data format consists of <enterprise-relationship-product> triples, which are composed of enterprise entities, product entities, and supply relationships between enterprise products.

[0110] The Neo4j graph database is used to visualize and store knowledge nodes and edges, thus completing the construction of a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises.

[0111] The system provided in this embodiment is used to execute the above-described method embodiments. For specific processes and details, please refer to the above embodiments, which will not be repeated here.

[0112] In one embodiment of the present invention, a computing device is provided, which can be a terminal and may include: a processor, a communication interface, memory, a display screen, and an input device. The processor, communication interface, and memory communicate with each other via a communication bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs, which are executed by the processor to implement the aforementioned methods; the internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals. Wireless communication can be achieved through Wi-Fi, a management network, NFC (Near Field Communication), or other technologies. The display screen can be a liquid crystal display or an e-ink display. The input device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad mounted on the casing of the computing device, or an external keyboard, touchpad, or mouse. The processor can call logical instructions stored in the memory.

[0113] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0114] In one embodiment of the present invention, a computer program product is provided, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, and when the program instructions are executed by a computer, the computer is able to perform the methods provided in the above-described method embodiments.

[0115] In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided, which stores server instructions that cause a computer to perform the methods provided in the above embodiments.

[0116] The computer-readable storage medium provided in the above embodiments has a similar implementation principle and technical effect to the above method embodiments, and will not be described again here.

[0117] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0118] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0119] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0120] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a knowledge graph of upstream and downstream relationships of products in an instrumentation enterprise, characterized in that, include: After preprocessing the input text vector, enterprise entity recognition is performed. The input text vector is corrected using a conditional random field model and product entity groups to achieve product entity recognition. Based on the results of enterprise entity identification and product entity identification, enterprise-product relationship extraction is performed for the supply relationship between upstream and downstream enterprises. The extracted enterprise entities, product entities, and supply relationships between enterprise products are integrated to construct a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises, and this knowledge graph is published. The input text vector is corrected using a Conditional Random Field (CRF) model and product entity groups. This process includes: extracting word features, part-of-speech features, boundary word features, and dictionary features from the input text vector using a feature extraction network, which are then used as input to the CRF model. The CRF model outputs a coarse recognition result for product entities. The input text vector is segmented by word segmentation and part-of-speech tagging. After segmentation, syntactic analysis is used to obtain product entity groups. These product entity groups are then used to correct the coarse recognition result for product entities, resulting in the final product entity recognition result. The process involves extracting enterprise-product relationships based on the supply relationships between upstream and downstream enterprises. This includes: traversing a relational lexicon and an enterprise dictionary to determine if the overall input text is related to enterprise supply relationships; dividing the input text into enterprise sets and product name sets based on enterprise and product identification results; filtering out statements containing enterprise names based on the enterprise name set, generating enterprise entity pairs after judgment according to preset rules, identifying suppliers, and obtaining customer company and product corpora; determining product information, obtaining multiple enterprises or products appearing in the enterprise set and product name set through syntactic analysis, matching customer companies with products, and thus determining the relationship chain between suppliers, customers, and products, obtaining the supply relationships between enterprise entities, product entities, and enterprise products.

2. The method for constructing a knowledge graph of upstream and downstream relationships of instrument and meter enterprises as described in claim 1, characterized in that, Enterprise entity recognition was performed using the Stanford CoreNLP Entity Recognition Toolkit.

3. The method for constructing a knowledge graph of upstream and downstream relationships of instrument and meter enterprises as described in claim 2, characterized in that, Enterprise entity identification, including: The input text vector is preprocessed to remove spaces and punctuation characters. The preprocessed text is segmented according to its structure to obtain word sequence. Identify enterprise entity names based on phrase sequences.

4. The method for constructing a knowledge graph of upstream and downstream relationships of instrument and meter enterprises as described in claim 1, characterized in that, CRF models are trained in a supervised manner, including: Data augmentation is performed by using entities of similar types within the domain dictionary. The jieba word segmentation tool is used to segment the existing text and replace the entities by analogy to generate a new data entry, thereby expanding the data. The output is converted to the CRF++ standard format and used as the training set to train the CRF model. The F-score is used as the evaluation metric for the CRF model. When the F-score is greater than a preset value, the model is considered successful. α This proves that the CRF model is effective.

5. The method for constructing a knowledge graph of upstream and downstream relationships of instrument and meter enterprises as described in claim 1, characterized in that, The extracted enterprise entities, product entities, and supply relationships between enterprise products are integrated to construct a knowledge graph of upstream and downstream relationships among intelligent instrumentation enterprises, including: The basic data format consists of <enterprise-relationship-product> triples, which are composed of enterprise entities, product entities, and supply relationships between enterprise products. The Neo4j graph database is used to visualize and store knowledge nodes and edges, thus completing the construction of a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises.

6. A knowledge graph construction system for upstream and downstream relationships of instrument and meter enterprises, used to implement the knowledge graph construction method for upstream and downstream relationships of instrument and meter enterprises as described in any one of claims 1 to 5, characterized in that, include: The enterprise entity recognition module preprocesses the input text vector before performing enterprise entity recognition. The product entity recognition module corrects the input text vector using a conditional random field model and product entity groups to achieve product entity recognition. The relationship extraction module extracts enterprise-product relationships based on the results of enterprise entity identification and product entity identification, targeting the supply relationships between upstream and downstream enterprises. The module for building and publishing integrates the extracted enterprise entities, product entities, and supply relationships between enterprise products to construct a knowledge graph of upstream and downstream relationships of intelligent instrument and meter enterprises, and then publishes this knowledge graph.

7. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods described in claims 1 to 5.

8. A computing device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods described in claims 1 to 5.