Method and system for setting enterprise and industrial chain atlas relation label

Abstract

The invention discloses a method and system for setting enterprise and industrial chain atlas relation labels, and the method comprises the steps: setting industrial chain atlas nodes based on collected industrial chain data; obtaining multi-dimensional enterprise dimension data of a target unit, and processing the enterprise dimension data; calculating a similarity score between the processed enterprise dimension data of each dimension and an industrial chain atlas node; determining a total similarity score of the enterprise dimension data based on the similarity score of the enterprise dimension data of each dimension; and setting a label of the target unit on the industrial chain atlas node based on the total similarity score of the enterprise dimension data.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise industrial chain extraction, and more specifically, to a method and system for setting relationship labels between enterprises and industrial chain graphs based on enterprise dimension data. Background Art

[0002] The industrial chain is an important form of modern economy. With the development of the globalized economy and the rapid advancement of information technology, cooperation and competition among enterprises are becoming increasingly fierce. Against this background, accurately matching the position of enterprises in the industrial chain has become the key to improving the efficiency of the industrial chain, optimizing resource allocation, and enhancing the core competitiveness of enterprises.

[0003] In recent years, the development of big data technology has provided new solutions for enterprises to match the industrial chain map. Through data mining, the business scope, product services, supply chain relationships and other information of the enterprise can be automatically analyzed, so as to locate the role and status of the enterprise in the industrial chain map.

[0004] However, the traditional industrial chain analysis method has the following disadvantages:

[0005] (1) Traditional industrial chain analysis methods mostly rely on expert experience and manual research, which is not only time-consuming and labor-intensive, but also difficult to adapt to rapidly changing market demands. Although big data technology provides new ideas, existing methods still face challenges when processing large-scale, multi-dimensional and dynamically changing industrial chain data. For example, how to accurately identify the complex supply and demand relationships between enterprises, how to update the industrial chain map in real time to reflect market changes, and how to comprehensively consider the multiple attributes and environmental factors of enterprises have not been fully resolved.

[0006] (2) Current technology is also insufficient in identifying the positioning of enterprises in cross-industry or emerging industries. With the integration and innovation of industries, traditional industry classification standards may no longer be applicable, which requires the industrial chain matching method to have higher flexibility and adaptability.

[0007] Therefore, a technology is needed to set relationship labels between enterprises and industrial chain maps based on enterprise dimension data. Summary of the invention

[0008] The technical solution of the present invention provides a method and system for setting relationship labels between enterprises and industrial chain maps, so as to solve the problem of how to set relationship labels between enterprises and industrial chain maps based on enterprise dimension data.

[0009] In order to solve the above problem, the present invention provides a method for setting a label of a relationship between an enterprise and an industrial chain graph, the method comprising:

[0010] Based on the collected industrial chain data, establish industrial chain map nodes; obtain multi-dimensional enterprise dimension data of the target unit, and process the enterprise dimension data;

[0011] Calculate the similarity scores between the processed enterprise dimension data of each dimension and the nodes of the industrial chain graph;

[0012] Determine a total similarity score of the enterprise dimension data based on the similarity scores of the enterprise dimension data of each dimension;

[0013] The label of the target unit on the industrial chain map node is set based on the total similarity score of the enterprise dimension data.

[0014] Preferably, the establishment of industrial chain map nodes based on the collected industrial chain data includes:

[0015] Establish industrial chain map nodes according to the nature of economic activities;

[0016] Establish industrial chain map nodes according to the granularity of industrial chain nodes;

[0017] Establish industry chain map nodes based on the upstream and downstream relationships of products;

[0018] Establish industrial chain map nodes according to the boundaries of the upstream and downstream relationships of products.

[0019] Preferably, the enterprise dimension data includes: enterprise information, enterprise patents, enterprise software copyrights, enterprise bidding, industry standards, enterprise recruitment, enterprise trademarks and enterprise business scope information.

[0020] Preferably, the processing of the enterprise dimension data includes:

[0021] Segment the enterprise dimension data to obtain information items of each enterprise dimension data;

[0022] Clean the information items of each enterprise dimension data after word segmentation;

[0023] A dimension word segmentation array vector is established based on the cleaned information items of each enterprise dimension data.

[0024] Preferably, the calculating of the similarity scores between the processed enterprise dimension data of each dimension and the industrial chain graph nodes includes:

[0025] Let the node name array vector be Assume the dimension of word segmentation array vector is Calculate the cosine similarity of each vector value in the name array vector V1 and the dimension word array vector V2 respectively, and use the calculated cosine similarity value as the initial similarity initial score Q of each dimension;

[0026] Based on the preset maximum number of scoring items b and the actual number of scoring items a of each enterprise dimension data, the similarity score of each dimension is determined:

[0027]

[0028] Among them, X represents the similarity score of enterprise dimension data, and w represents the weight of enterprise dimension data.

[0029] Preferably, the calculating of the similarity scores between the processed enterprise dimension data of each dimension and the industrial chain graph nodes also includes setting scoring conditions:

[0030]

[0031] Among them, Y means that the similarity score scoring conditions are met, C means the company name, and the company name is scored when it contains any word in the node's name array; D means that the business scope, patents, software, industry standards, recruitment positions, bidding, and trademarks are scored when (max (similarity score between the dimension word array data and the node name array data) ≥ m), where m is the preset value.

[0032] Preferably, the step of setting the label of the target unit on the industrial chain graph node based on the total similarity score of the enterprise dimension data includes:

[0033] When the total similarity score of the enterprise dimension data is greater than a preset threshold n, a label of the target unit on the industrial chain map node is set.

[0034] Preferably, the method further includes verifying the label set by the target unit on the industrial chain graph node:

[0035] Build a deep neural network model, and establish a training data set and a validation set including industry chain graph nodes, enterprise dimension data and labels;

[0036] Training the deep neural network model based on the training data set and adjusting the parameters of the deep neural network model; verifying the deep neural network model through the verification set;

[0037] When the deep neural network model passes the verification of the verification set, determining the deep neural network model as the optimal deep neural network model;

[0038] The labels are verified by the optimal deep neural network model.

[0039] Based on another aspect of the present invention, the present invention provides a system for setting labels for the relationship between an enterprise and an industrial chain graph, the system comprising:

[0040] An acquisition unit is used to establish an industrial chain map node based on the collected industrial chain data; acquire multi-dimensional enterprise dimension data of the target unit, and process the enterprise dimension data;

[0041] A calculation unit, used to calculate the similarity scores between the processed enterprise dimension data of each dimension and the industrial chain map nodes; used to determine the total similarity score of the enterprise dimension data based on the similarity scores of the enterprise dimension data of each dimension;

[0042] A result unit is used to set the label of the target unit on the industrial chain map node based on the total similarity score of the enterprise dimension data.

[0043] Preferably, the acquisition unit is used to establish an industrial chain map node based on the collected industrial chain data, and is also used to:

[0044] Establish industrial chain map nodes according to the nature of economic activities;

[0045] Establish industrial chain map nodes according to the granularity of industrial chain nodes;

[0046] Establish industry chain map nodes based on the upstream and downstream relationships of products;

[0047] Establish industrial chain map nodes according to the boundaries of the upstream and downstream relationships of products.

[0048] Preferably, the enterprise dimension data includes: enterprise information, enterprise patents, enterprise software copyrights, enterprise bidding, industry standards, enterprise recruitment, enterprise trademarks and enterprise business scope information.

[0049] Preferably, the acquisition unit is used to process the enterprise dimension data and is also used to:

[0050] Segment the enterprise dimension data to obtain information items of each enterprise dimension data;

[0051] Clean the information items of each enterprise dimension data after word segmentation;

[0052] A dimension word segmentation array vector is established based on the cleaned information items of each enterprise dimension data.

[0053] Preferably, the calculation unit is used to calculate the similarity score between the processed enterprise dimension data of each dimension and the industrial chain graph node, and is also used to:

[0054] Let the node name array vector be Assume the dimension of word segmentation array vector is Calculate the cosine similarity of each vector value in the name array vector V1 and the dimension word array vector V2 respectively, and use the calculated cosine similarity value as the initial similarity initial score Q of each dimension;

[0055] Based on the preset maximum number of scoring items b and the actual number of scoring items a of each enterprise dimension data, the similarity score of each dimension is determined:

[0056]

[0057] Among them, X represents the similarity score of enterprise dimension data, and w represents the weight of enterprise dimension data.

[0058] Preferably, the calculation unit is used to calculate the similarity score between the processed enterprise dimension data of each dimension and the industrial chain graph node, and also includes setting the scoring conditions:

[0059]

[0060] Among them, Y means that the similarity score scoring conditions are met, C means the company name, and the company name is scored when it contains any word in the node's name array; D means that the business scope, patents, software, industry standards, recruitment positions, bidding, and trademarks are scored when (max (similarity score between the dimension word array data and the node name array data) ≥ m), where m is the preset value.

[0061] Preferably, the result unit is used to set the label of the target unit on the industrial chain graph node based on the total similarity score of the enterprise dimension data, and is also used to:

[0062] When the total similarity score of the enterprise dimension data is greater than a preset threshold n, a label of the target unit on the industrial chain map node is set.

[0063] Preferably, the result unit is further used to verify the label set by the target unit on the industrial chain graph node:

[0064] Build a deep neural network model, and establish a training data set and a validation set including industry chain graph nodes, enterprise dimension data and labels;

[0065] Training the deep neural network model based on the training data set and adjusting the parameters of the deep neural network model; verifying the deep neural network model through the verification set;

[0066] When the deep neural network model passes the verification of the verification set, determining the deep neural network model as the optimal deep neural network model;

[0067] The labels are verified by the optimal deep neural network model.

[0068] Based on another aspect of the present invention, the present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, and the computer program is used to execute a method for setting relationship labels between enterprises and industrial chain maps.

[0069] Based on another aspect of the present invention, the present invention provides an electronic device, characterized in that the electronic device comprises: a processor and a memory; wherein,

[0070] The memory is a memory for storing instructions executable by the processor;

[0071] The processor is used to read the executable instructions from the memory and execute the instructions to implement a method for setting relationship labels between enterprises and industrial chain maps.

[0072] The technical solution of the present invention provides a method and system for setting labels for the relationship between enterprises and industrial chain maps, wherein the method includes: setting up industrial chain map nodes based on collected industrial chain data; obtaining multi-dimensional enterprise dimension data of the target unit, and processing the enterprise dimension data; calculating the similarity score between the processed enterprise dimension data of each dimension and the industrial chain map node; determining the total similarity score of the enterprise dimension data based on the similarity of the enterprise dimension data of each dimension; setting the label of the target unit on the industrial chain map node based on the total similarity score of the enterprise dimension data. The technical solution of the present invention is based on the method and system for extracting labels for the relationship between enterprises and industrial chain maps based on enterprise dimension data, and through the enterprise high-dimensional data mining technology, analyzes consumer data feedback to understand customer preferences, and is applicable to enterprise information, enterprise business scope, enterprise patents, enterprise software copyrights and enterprise bidding, thereby improving product quality and enhancing market competitiveness. The industrial chain matching method provided by the present invention has higher flexibility and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] A more complete understanding of exemplary embodiments of the present invention may be obtained by referring to the following drawings:

[0074] Figure 1 A flowchart of a method for setting labels for the relationship between an enterprise and an industrial chain graph according to a preferred embodiment of the present invention;

[0075] Figure 2 Schematic diagram of the architecture of an extraction system according to a preferred embodiment of the present invention.

[0076] Figure 3 A schematic diagram of the architecture of an information collection platform according to a preferred embodiment of the present invention;

[0077] Figure 4A schematic diagram of the architecture of a data preprocessing platform according to a preferred embodiment of the present invention;

[0078] Figure 5 A diagram showing a module for setting up a node according to a preferred embodiment of the present invention;

[0079] Figure 6 A schematic diagram of the architecture of a data processing platform according to a preferred embodiment of the present invention;

[0080] Figure 7 A flowchart of a data processing platform according to a preferred embodiment of the present invention;

[0081] Figure 8 A schematic diagram of an architecture for establishing a platform for enterprise dimensions according to a preferred embodiment of the present invention;

[0082] Fig. 9 A schematic diagram of the architecture of a label platform according to a preferred embodiment of the present invention;

[0083] Fig.10 A flowchart of a method for setting labels for the relationship between an enterprise and an industrial chain graph according to a preferred embodiment of the present invention;

[0084] Fig.11 A flowchart of determining the relationship between an enterprise and an industrial chain map according to a preferred embodiment of the present invention;

[0085] Fig.12 A schematic diagram of the architecture of a model verification platform according to a preferred embodiment of the present invention; and

[0086] Fig.13 This is a system structure diagram for setting labels for the relationship between enterprises and industrial chain maps according to a preferred implementation of the present invention. DETAILED DESCRIPTION

[0087] Now, exemplary embodiments of the present invention are described with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely and to fully convey the scope of the present invention to those skilled in the art. The terms used in the exemplary embodiments shown in the accompanying drawings are not intended to limit the present invention. In the accompanying drawings, the same units / elements are marked with the same reference numerals.

[0088] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0089] Figure 1 The present invention is a flowchart of a method for setting labels for the relationship between an enterprise and an industrial chain map according to a preferred embodiment of the present invention.

[0090] The present invention provides a method for extracting the relationship labels between enterprises and industrial chain maps based on enterprise dimension data, so as to solve the problem that traditional industrial chain analysis methods mostly rely on expert experience and manual research. Traditional technology is not only time-consuming and labor-intensive, but also difficult to adapt to the rapidly changing market demand. Although big data technology provides new ideas, existing methods still face challenges when processing large-scale, multi-dimensional and dynamically changing industrial chain data. For example, how to accurately identify the complex supply and demand relationship between enterprises, how to update the industrial chain map in real time to reflect market changes, how to comprehensively consider the multiple attributes and environmental factors of enterprises, etc., have not been fully solved; current technology also has shortcomings in identifying the positioning of enterprises in cross-industry or emerging industries. With the integration and innovation of industries, traditional industry classification standards are no longer applicable, which requires industrial chain matching methods to have higher flexibility and adaptability.

[0091] The technical solution of the present invention provides a method for setting a label of a relationship between an enterprise and an industrial chain graph, the method comprising:

[0092] Step 101: Based on the collected industrial chain data, establish an industrial chain map node; obtain multi-dimensional enterprise dimension data of the target unit, and process the enterprise dimension data;

[0093] Preferably, based on the collected industrial chain data, an industrial chain map node is established, including:

[0094] Establish industrial chain map nodes according to the nature of economic activities;

[0095] Establish industrial chain map nodes according to the granularity of industrial chain nodes;

[0096] Establish industry chain map nodes based on the upstream and downstream relationships of products;

[0097] Establish industrial chain map nodes according to the boundaries of the upstream and downstream relationships of products.

[0098] Preferably, the enterprise dimension data includes: enterprise information, enterprise patents, enterprise software copyrights, enterprise bidding, industry standards, enterprise recruitment, enterprise trademarks and enterprise business scope information.

[0099] Preferably, the enterprise dimension data is processed, including:

[0100] Segment the enterprise dimension data to obtain information items of each enterprise dimension data;

[0101] Clean the information items of each enterprise dimension data after word segmentation;

[0102] A dimension word segmentation array vector is established based on the cleaned information items of each enterprise dimension data.

[0103] The method for extracting enterprise and industrial chain graph relationship labels based on enterprise dimension data provided by the present invention is based on a software architecture including an extraction system, and the extraction system includes an information collection platform, a data preprocessing platform, a data processing platform, an enterprise dimension establishment platform, a model verification platform and a labeling platform;

[0104] The information collection platform collects information on enterprises, enterprise patents, enterprise software copyrights, enterprise bidding and enterprise business scope;

[0105] The data preprocessing platform preprocesses enterprise data and industrial chain data respectively;

[0106] The data processing platform cleans, organizes and formats the collected enterprise dimension data;

[0107] Step 102: Calculate the similarity scores between the processed enterprise dimension data of each dimension and the industrial chain graph nodes;

[0108] Preferably, calculating the similarity scores between the processed enterprise dimension data of each dimension and the industrial chain graph nodes includes:

[0109] Let the node name array vector be Assume the dimension of word segmentation array vector is Calculate the cosine similarity of each vector value in the name array vector V1 and the dimension word array vector V2 respectively, and use the calculated cosine similarity value as the initial similarity initial score Q of each dimension;

[0110] Based on the preset maximum number of scoring items b and the actual number of scoring items a of each enterprise dimension data, the similarity score of each dimension is determined:

[0111]

[0112] Among them, X represents the similarity score of enterprise dimension data, and w represents the weight of enterprise dimension data.

[0113] Preferably, calculating the similarity scores between the processed enterprise dimension data of each dimension and the industrial chain graph nodes also includes setting scoring conditions:

[0114]

[0115] Among them, Y means that the similarity score conditions are met, C means the company name, and the score is calculated when the company name contains any word in the node's name array; D means that the score is calculated when (max (similarity score between dimension word array data and node name array data) ≥ m) in business scope, patents and software copyrights, industry standards, recruitment positions, bidding, and trademarks, where m is the preset value.

[0116] Step 103: determining a total similarity score of the enterprise dimension data based on the similarity scores of the enterprise dimension data of each dimension;

[0117] The present invention defines the weight of enterprise dimension data in scoring by establishing a platform.

[0118] Step 104: Set the label of the target unit on the industrial chain map node based on the total similarity score of the enterprise dimension data.

[0119] Preferably, setting the label of the target unit on the industrial chain graph node based on the total similarity score of the enterprise dimension data includes:

[0120] When the total similarity score of the enterprise dimension data is greater than the preset threshold n, the label of the target unit on the industrial chain map node is set.

[0121] The present invention marks the node labels of the enterprise's industrial chain graph through a label platform.

[0122] Preferably, the method further includes verifying the label set by the target unit on the industrial chain graph node:

[0123] Build a deep neural network model, and establish a training data set and a validation set including industry chain graph nodes, enterprise dimension data and labels;

[0124] Train the deep neural network model based on the training data set and adjust the parameters of the deep neural network model; verify the deep neural network model through the verification set;

[0125] When the deep neural network model passes the verification of the verification set, the deep neural network model is determined as the optimal deep neural network model;

[0126] The labels are verified by the optimal deep neural network model.

[0127] The present invention constructs a deep neural network model for the data collected by the information collection platform through the model verification platform, inputs the features of different feature sources into the model, and uses the training data to train the network and adjust the parameters, and fuses the learned features through splicing, weighting or other methods. The fused features are passed to the subsequent levels or classifiers, and the performance of the model is evaluated through the validation set or cross-validation, and the labeling content of the label platform is checked and verified after the feature fusion method is tuned according to the needs.

[0128] The information collection platform of the present invention comprises an enterprise information collection module, an enterprise patent collection module, an enterprise software copyright collection module, an industry standard collection module, an enterprise recruitment collection module, an enterprise bidding collection module and an enterprise business scope collection module. The enterprise information collection module collects the main information of the enterprise, the enterprise patent collection module collects the relevant patents of the enterprise, the enterprise software copyright collection module collects the relevant software copyrights of the enterprise, the enterprise bidding collection module collects the enterprise bidding information, the enterprise business scope collection module collects the enterprise business scope information, the industry standard collection module collects information on the industry standards of the enterprise, and the enterprise recruitment collection module collects the enterprise recruitment information.

[0129] The data preprocessing platform of the present invention includes an enterprise data preparation module, an industrial chain data preparation module and a node establishment module. The enterprise data preparation module prepares the enterprise data to be collected and collects network public data. The industrial chain data preparation module prepares the industrial chain data to be collected and collects network public data. The node establishment module establishes nodes for the collected data to facilitate the installation of node sequence for collection. The node establishment module includes an economic activity division unit, a hierarchical division unit, a relationship division unit and a boundary division unit. The economic activity division unit divides nodes according to the nature of the same economic activity. The hierarchical division unit is included in the corresponding hierarchical level in the graph system. The hierarchical distribution should match the granularity of the industrial chain nodes. The relationship division unit covers the upstream and downstream relationships between all products, including production raw materials, production accessories, auxiliary materials, production equipment, auxiliary equipment, processing technology, technical services and upstream and downstream correlations between product businesses. The boundary division unit should clearly define the nodes along the upstream and downstream correlations of the industrial chain graph.

[0130] The data processing platform of the present invention comprises a data classification module, a data cleaning module, a data sorting module, a data formatting module and a data storage module. The data classification module classifies the data collected by the information collection platform, and uses jieba, an open source Chinese word segmenter python library, to perform word segmentation processing on the formatted enterprise dimension data such as business scope, patents, software copyrights, industry standards, recruitment, bidding, and trademarks. The data cleaning module discloses the Nice classification table of data, cleans the enterprise dimension data such as patents, trademarks, bidding, industry standards, and software copyrights, extracts product words, removes stop words, place names, and meaningless words of time. The data sorting module merges and cleans and then vectorizes and sorts the words to obtain a vector word library. The data formatting module sorts the collected data into a unified format for easy processing. The data storage module stores the obtained data in a centralized manner.

[0131] The enterprise dimension establishment platform of the present invention includes a similarity calculation module, a weight ratio calculation module for each dimension of the enterprise, a dimension score calculation module and a scoring condition module. The similarity calculation module calculates the similarity scores between each dimension of the enterprise and the nodes of the industrial chain map. Gensim is used to load the vector vocabulary. The node name array vector is set as Assume the dimension of word segmentation array vector is Calculate the cosine similarity of each vector value in V1 and V2[0]-V2[n] respectively, and obtain the highest similarity value similarity[s1, ···, sn], and the weight proportion calculation module of each dimension score of the enterprise:

[0132] 1 = 0.2 (company name) + 0.1 (business scope) + 0.1 (professional) + 0.1 (software copyright) + 0.1 (industry standard) + 0.1 (recruitment position) + 0.1 (trademark) + 0.2 (bidding),

[0133] Dimension score calculation module:

[0134]

[0135] In the above formula, X represents the dimension score, a represents the number of dimension data scores, b represents the highest score, Q represents the total score, and w represents the dimension weight;

[0136] Scoring condition module:

[0137]

[0138] In the above formula, Y indicates that the scoring conditions are met, C indicates the company name, and the company name contains any word in the node array; D indicates business scope, patents and software copyrights, industry standards, recruitment positions, bidding, and trademarks (max (similarity score between dimension segmentation and node name array) ≥ m).

[0139] As a preferred technical solution of the present invention, the label platform includes a correlation score threshold comparison module and a node signing module. The correlation score threshold comparison module calculates the similarity score of each dimension with the industrial chain map node, and adds up the scores to obtain a total score recorded as sum. When sum≥n, n is recorded as the correlation score threshold between the enterprise and the industrial map node. The node signing module:

[0140] Satisfy the relationship between the enterprise and the industry graph node = ﹛sum (scores of each dimension) ≥ threshold n.

[0141] The model verification platform of the present invention includes a model training module, a model evaluation module and a verification and proofreading module. The model training module selects a data set collected by an information collection platform to train a model, adjusts hyperparameters, and uses a separate verification set to verify the performance of the model. After the model training is completed, the model evaluation module evaluates the model to measure its performance, which includes using a test set that has not been used in the training or verification process to test the model. The pre-defined evaluation indicators are used to measure the accuracy, precision, recall rate and other relevant performance indicators of the model. The verification and proofreading module verifies and proofreads the results of the industry chain map node labels marked by the labeling platform for the enterprise.

[0142] The method of extracting the relationship labels between enterprises and industrial chain graphs based on enterprise dimension data of the present invention comprises the following steps:

[0143] Step 1: Get the node names of the industrial chain graph: The data preprocessing platform uses gensim, an open source Python library for topic modeling, document indexing, and large-scale corpus similarity retrieval, and the vector word library to obtain the top 3 similar words. The similar words are added to the node name to get the phrase of node[n1,n2,n3,n4];

[0144] Step 2: Obtain text list: The information collection platform obtains enterprise name text, enterprise business scope text, patent title text list, software title text list, industry standard name text list, recruitment position text list, bidding title text list, and trademark name text list;

[0145] Step 3: Data processing: The data processing platform uses jieba, an open source Chinese word segmenter Python library, to segment the formatted enterprise dimension data such as business scope, patents, software copyrights, industry standards, recruitment, bidding, and trademarks;

[0146] Step 4: Similarity calculation: Calculate the similarity scores between each dimension of the enterprise and the nodes of the industrial chain graph;

[0147] Step 5: Data counting: The enterprise dimension sets up a platform to clean and count the scores of each dimension;

[0148] Step 6: Node labeling: The labeling platform calculates the total score of each dimension and labels the enterprise and industrial chain map nodes;

[0149] Step 7: Result verification: The model verification platform verifies the labeling results of the labeling platform.

[0150] The present invention is based on the systematic application of enterprise-dimensional data extraction of relationship labels between enterprises and industrial chain maps, and analyzes consumer data feedback through high-dimensional data mining technology to understand customer preferences. It is also applicable to enterprise information, business scope, patents, software copyrights and bidding, thereby improving product quality and enhancing market competitiveness.

[0151] The present invention achieves an effective balance between automation and matching accuracy through advanced algorithm design; the present invention's dynamic data processing mechanism that is updated and adjusted in real time ensures the timeliness and accuracy of the industrial chain map; the present invention uses big data technology to automatically analyze the relationship between enterprises and industrial chain maps, and determines whether there is an association relationship by calculating the similarity between enterprise dimension data and industrial chain map nodes.

[0152] Fig.13 This is a system structure diagram for setting labels for the relationship between enterprises and industrial chain maps according to a preferred implementation of the present invention.

[0153] like Fig.13 As shown, the present invention provides a system for setting labels for the relationship between an enterprise and an industrial chain graph, the system comprising:

[0154] The acquisition unit 1301 is used to establish an industrial chain map node based on the collected industrial chain data; acquire multi-dimensional enterprise dimension data of the target unit, and process the enterprise dimension data;

[0155] Preferably, the acquisition unit 1301 is used to establish an industrial chain map node based on the collected industrial chain data, and is also used to:

[0156] Establish industrial chain map nodes according to the nature of economic activities;

[0157] Establish industrial chain map nodes according to the granularity of industrial chain nodes;

[0158] Establish industry chain map nodes based on the upstream and downstream relationships of products;

[0159] Establish industrial chain map nodes according to the boundaries of the upstream and downstream relationships of products.

[0160] Preferably, the enterprise dimension data includes: enterprise information, enterprise patents, enterprise software copyrights, enterprise bidding, industry standards, enterprise recruitment, enterprise trademarks and enterprise business scope information.

[0161] Preferably, the acquisition unit 1301 is used to process the enterprise dimension data and is also used to:

[0162] Segment the enterprise dimension data to obtain information items of each enterprise dimension data;

[0163] Clean the information items of each enterprise dimension data after word segmentation;

[0164] A dimension word segmentation array vector is established based on the cleaned information items of each enterprise dimension data.

[0165] The calculation unit 1302 is used to calculate the similarity scores between the processed enterprise dimension data of each dimension and the industrial chain graph node; and to determine the total similarity score of the enterprise dimension data based on the similarity scores of the enterprise dimension data of each dimension;

[0166] Preferably, the calculation unit 1302 is used to calculate the similarity score between the processed enterprise dimension data of each dimension and the industrial chain graph node, and is also used to:

[0167] Let the node name array vector be Assume the dimension of word segmentation array vector is Calculate the cosine similarity of each vector value in the name array vector V1 and the dimension word array vector V2 respectively, and use the calculated cosine similarity value as the initial similarity initial score Q of each dimension;

[0168] Based on the preset maximum number of scoring items b and the actual number of scoring items a of each enterprise dimension data, the similarity score of each dimension is determined:

[0169]

[0170] Among them, X represents the similarity score of enterprise dimension data, and w represents the weight of enterprise dimension data.

[0171] Preferably, the calculation unit 1302 is used to calculate the similarity score between the processed enterprise dimension data of each dimension and the industrial chain graph node, and also includes setting the scoring conditions:

[0172]

[0173] Among them, Y means that the similarity score conditions are met, C means the company name, and the score is calculated when the company name contains any word in the node's name array; D means that the score is calculated when (max (similarity score between dimension word array data and node name array data) ≥ m) in business scope, patents and software copyrights, industry standards, recruitment positions, bidding, and trademarks, where m is the preset value.

[0174] The result unit 1303 is used to set the label of the target unit on the industrial chain map node based on the total similarity score of the enterprise dimension data.

[0175] Preferably, the result unit 1303 is used to set the label of the target unit on the industrial chain graph node based on the total similarity score of the enterprise dimension data, and is also used to:

[0176] When the total similarity score of the enterprise dimension data is greater than the preset threshold n, the label of the target unit on the industrial chain map node is set.

[0177] Preferably, the result unit 1303 is also used to verify the label set by the target unit on the industrial chain graph node:

[0178] Build a deep neural network model, and establish a training data set and a validation set including industry chain graph nodes, enterprise dimension data and labels;

[0179] Train the deep neural network model based on the training data set and adjust the parameters of the deep neural network model; verify the deep neural network model through the verification set;

[0180] When the deep neural network model passes the verification of the verification set, the deep neural network model is determined as the optimal deep neural network model;

[0181] The labels are verified by the optimal deep neural network model.

[0182] The present invention provides a computer-readable storage medium, which stores a computer program, and the computer program is used to execute a method for setting relationship labels between enterprises and industrial chain maps.

[0183] The present invention provides an electronic device, the electronic device comprising: a processor and a memory; wherein:

[0184] Memory, a memory for storing processor executable instructions;

[0185] A processor is used to read executable instructions from a memory and execute the instructions to implement a method for setting relationship labels between enterprises and industrial chain maps.

[0186] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0187] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0188] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0189] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0190] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0191] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0192] The invention has been described above with reference to a few embodiments. However, it is readily apparent to a person skilled in the art that other embodiments than the ones disclosed above are equally within the scope of the invention, as defined by the appended patent claims.

[0193] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. All references to "a / said / the [means, components, etc.]" are to be openly interpreted as at least one instance of the means, components, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not necessarily have to be performed in the exact order disclosed, unless explicitly stated otherwise.

Claims

1. A method for setting a label of a relationship between an enterprise and an industrial chain graph, the method comprising: Based on the collected industrial chain data, establish industrial chain map nodes; Acquire multi-dimensional enterprise dimension data of the target unit, and process the enterprise dimension data; Calculate the similarity scores between the processed enterprise dimension data of each dimension and the nodes of the industrial chain graph; Determine a total similarity score of the enterprise dimension data based on the similarity scores of the enterprise dimension data of each dimension; The label of the target unit on the industrial chain map node is set based on the total similarity score of the enterprise dimension data.

2. According to the method of claim 1, the step of establishing an industrial chain map node based on the collected industrial chain data comprises: Establish industrial chain map nodes according to the nature of economic activities; Establish industrial chain map nodes according to the granularity of industrial chain nodes; Establish industry chain map nodes based on the upstream and downstream relationships of products; Establish industrial chain map nodes according to the boundaries of the upstream and downstream relationships of products.

3. The method according to claim 1, wherein the enterprise dimension data comprises: Company information, company patents, company software copyrights, company bidding, industry standards, company recruitment, company trademarks and company business scope information.

4. The method according to claim 3, wherein the processing of the enterprise dimension data comprises: Segment the enterprise dimension data to obtain information items of each enterprise dimension data; Clean the information items of each enterprise dimension data after word segmentation; A dimension word segmentation array vector is established based on the cleaned information items of each enterprise dimension data.

5. According to the method of claim 1, the step of calculating the similarity scores between the processed enterprise dimension data of each dimension and the industrial chain graph nodes comprises: Let the node name array vector be Assume the dimension of word segmentation array vector is Calculate the cosine similarity of each vector value in the name array vector V1 and the dimension word array vector V2 respectively, and use the calculated cosine similarity value as the initial similarity initial score Q of each dimension; Based on the preset maximum number of scoring items b and the actual number of scoring items a of each enterprise dimension data, the similarity score of each dimension is determined: Among them, X represents the similarity score of enterprise dimension data, and w represents the weight of enterprise dimension data.

6. According to the method of claim 5, the step of calculating the similarity scores between the processed enterprise dimension data of each dimension and the industrial chain graph nodes further comprises setting scoring conditions: in, Y indicates that the similarity score conditions are met, C indicates the company name, and the company name is scored when it contains any word in the node's name array; D indicates that the business scope, patents, software, industry standards, recruitment positions, bidding, and trademarks are scored when (max (similarity score between the dimension word array data and the node's name array data) ≥ m), where m is a preset value.

7. According to the method of claim 1, the step of setting the label of the target unit on the industrial chain map node based on the total similarity score of the enterprise dimension data comprises: When the total similarity score of the enterprise dimension data is greater than a preset threshold n, a label of the target unit on the industrial chain map node is set.

8. The method according to claim 1 further comprises verifying the label set by the target unit on the industrial chain map node: Build a deep neural network model, and establish a training data set and a validation set including industry chain graph nodes, enterprise dimension data and labels; Training the deep neural network model based on the training data set and adjusting the parameters of the deep neural network model; verifying the deep neural network model through the verification set; When the deep neural network model passes the verification of the verification set, determining the deep neural network model as the optimal deep neural network model; The labels are verified by the optimal deep neural network model.

9. A system for setting labels for the relationship between an enterprise and an industrial chain graph, the system comprising: An acquisition unit is used to establish industrial chain map nodes based on the collected industrial chain data; Acquire multi-dimensional enterprise dimension data of the target unit, and process the enterprise dimension data; A calculation unit, used to calculate the similarity scores between the processed enterprise dimension data of each dimension and the nodes of the industrial chain graph; Used to determine the total similarity score of the enterprise dimension data based on the similarity scores of the enterprise dimension data of each dimension; A result unit is used to set the label of the target unit on the industrial chain map node based on the total similarity score of the enterprise dimension data.

10. The system according to claim 9, wherein the acquisition unit is used to establish industrial chain map nodes based on the collected industrial chain data, and is also used to: Establish industrial chain map nodes according to the nature of economic activities; Establish industrial chain map nodes according to the granularity of industrial chain nodes; Establish industry chain map nodes based on the upstream and downstream relationships of products; Establish industrial chain map nodes according to the boundaries of the upstream and downstream relationships of products.

11. The system according to claim 9, wherein the enterprise dimension data comprises: Company information, company patents, company software copyrights, company bidding, industry standards, company recruitment, company trademarks and company business scope information.

12. The system according to claim 11, wherein the acquisition unit is used to process the enterprise dimension data and further to: Segment the enterprise dimension data to obtain information items of each enterprise dimension data; Clean the information items of each enterprise dimension data after word segmentation; A dimension word segmentation array vector is established based on the cleaned information items of each enterprise dimension data.

13. The system according to claim 9, wherein the calculation unit is used to calculate the similarity score between the processed enterprise dimension data of each dimension and the industrial chain graph node, and is also used to: Let the node name array vector be Assume the dimension of word segmentation array vector is Calculate the cosine similarity of each vector value in the name array vector V1 and the dimension word array vector V2 respectively, and use the calculated cosine similarity value as the initial similarity initial score Q of each dimension; Based on the preset maximum number of scoring items b and the actual number of scoring items a of each enterprise dimension data, the similarity score of each dimension is determined: in, X represents the similarity score of enterprise dimension data, and w represents the weight of enterprise dimension data.

14. According to the system of claim 13, the calculation unit is used to calculate the similarity score between the processed enterprise dimension data of each dimension and the industrial chain map node, and is also used to set the scoring conditions: in, Y indicates that the similarity score conditions are met, C indicates the company name, and the company name is scored when it contains any word in the node's name array; D indicates that the business scope, patents, software, industry standards, recruitment positions, bidding, and trademarks are scored when (max (similarity score between the dimension word array data and the node's name array data) ≥ m), where m is a preset value.

15. The system according to claim 9, wherein the result unit is used to set the label of the target unit on the industrial chain map node based on the total similarity score of the enterprise dimension data, and is also used to: When the total similarity score of the enterprise dimension data is greater than a preset threshold n, a label of the target unit on the industrial chain map node is set.

16. According to the system of claim 9, the result unit is further used to verify the label set by the target unit on the industrial chain map node: Build a deep neural network model, and establish a training data set and a validation set including industry chain graph nodes, enterprise dimension data and labels; Training the deep neural network model based on the training data set and adjusting the parameters of the deep neural network model; verifying the deep neural network model through the verification set; When the deep neural network model passes the verification of the verification set, determining the deep neural network model as the optimal deep neural network model; The labels are verified by the optimal deep neural network model.

17. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is used to execute the method according to any one of claims 1 to 8.

18. An electronic device, characterized in that: The electronic device comprises: a processor and a memory; wherein, The memory is a memory for storing instructions executable by the processor; The processor is configured to read the executable instructions from the memory and execute the instructions to implement the method according to any one of claims 1 to 8.

Images