A classification method and system for industrial Internet industry chain based on word segmentation technology

Abstract

The present invention provides an industrial Internet industry chain classification method and system based on word segmentation technology, and the method includes: firstly obtaining historical industrial Internet core industry chain data and extracting characteristic information therefrom; taking several sub-categories of the core industries of the industrial Internet as sample sets, processing the characteristic information by using the grey correlation analysis method, and obtaining correlation factors; taking the correlation factors as the input of the KNN classification model, optimizing the KNN classification model based on the naive Bayes algorithm, and obtaining the industrial Internet industry chain classification model; obtaining the current industrial Internet core industry chain data as the data to be classified, inputting the correlation factors of the data to be classified and several sub-categories of industries into the industrial Internet industry chain classification model, and obtaining the sub-categories of industries to which the current industrial Internet core industry chain data belongs. The present invention accurately and efficiently processes data similarity when classifying the industrial Internet industry chain, and improves the classification accuracy and classification efficiency.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an industrial Internet industry chain classification method and system based on word segmentation technology. Background Art

[0002] At present, a new round of scientific and technological revolution and industrial transformation is booming. As a key supporting force for digital transformation, the Industrial Internet is constantly subverting traditional manufacturing models, production organization methods and industrial forms on a global scale, promoting the accelerated transformation and upgrading of traditional industries and the accelerated development and growth of new industries. The essence and core of the Industrial Internet is to closely connect and integrate equipment, production lines, factories, suppliers, products and customers through the Industrial Internet platform, which can help the manufacturing industry to extend the industrial chain and form Internet interoperability across equipment, systems, factories and regions, thereby improving efficiency and promoting the intelligence of the entire manufacturing service system. It is also conducive to promoting the integrated development of the manufacturing industry, realizing the leapfrog development between the manufacturing industry and the service industry, and enabling the efficient sharing of various elements and resources of the industrial economy.

[0003] The industrial Internet industry chain can be divided into six major parts: equipment layer, network layer, platform layer, software layer, application layer and security system. Among them, the equipment layer refers to intelligent production equipment, production status intelligent terminals, embedded software and industrial data centers; the network layer refers to the communication inside and outside the factory; the platform layer includes industrial cloud platforms such as collaborative R&D, collaborative manufacturing, information trading and data integration; the software layer includes software such as R&D design, information management and production control, which are the core links to help enterprises realize digital value; the application layer includes vertical industry applications, process applications and applications based on data analysis; the security system is permeated in the above layers and is an important support and guarantee for the industry. At present, the research on the above six major parts of the industrial Internet industry chain is relatively common. However, this type of research is only based on the analysis and extraction of the industrial chain, and is simply classified. There are still defects in the similarity processing and data fusion of data in terms of the business scope of enterprises in the industrial Internet industry chain. Summary of the invention

[0004] In view of this, the present application proposes an industrial Internet industry chain classification method and system based on word segmentation technology, which is used to solve the problem that the existing technology cannot accurately and efficiently process data similarity and data fusion when classifying the industrial Internet industry chain, resulting in inaccurate classification results.

[0005] The technical solution of the present invention is achieved in this way:

[0006] The present invention proposes an industrial Internet industry chain classification method based on word segmentation technology, the method comprising:

[0007] S1, obtain the historical industrial Internet core industry chain data, analyze the historical industrial Internet core industry data chain, and obtain the enterprise information related to each historical industrial Internet core industry chain data;

[0008] S2, segmenting the industrial Internet enterprise information, removing stop words and illegal format characters, and extracting feature information from it;

[0009] S3, the core industry of the industrial Internet has several sub-industries. The sub-industries are used as sample sets, and the characteristic information is processed by using the grey correlation analysis method to obtain the correlation factor between the characteristic information and the sample set;

[0010] S4, taking the correlation factor as the input of the KNN classification model, optimizing the KNN classification model based on the naive Bayes algorithm, and obtaining the industrial Internet industry chain classification model;

[0011] S5, obtain the current industrial Internet core industrial chain data as the data to be classified, obtain the correlation factors between the data to be classified and several small industries, input the correlation factors between the data to be classified and several small industries into the industrial Internet industrial chain classification model, and obtain the small industry to which the current industrial Internet core industrial chain data belongs.

[0012] On the basis of the above technical solution, preferably, step S1 specifically includes:

[0013] Obtain historical industrial Internet core industry chain data, perform word segmentation on the historical industrial Internet core industry data chain, extract feature information of the historical industrial Internet industry chain data, and establish the first word segmentation database;

[0014] Obtain the basic classification of the industrial Internet industry, perform word segmentation on the basic classification of the industrial Internet industry, extract feature information of the basic classification of the industrial Internet industry, and establish a second word segmentation database;

[0015] Based on the first word library and the second word library, enterprise information related to each historical industrial Internet core industrial chain data is obtained.

[0016] In the above technical solution, preferably, step S2 specifically includes:

[0017] The improved TF-IDF algorithm is used to extract feature information, and the processed industrial Internet enterprise information is used as the corpus. The category set of the corpus is C = {C1, C2, ..., C m}, m is the number of categories;

[0018] The improved TF-IDF calculation formula is:

[0019]

[0020] Among them, T k represents the feature item, k represents the number of feature words, W(T ik ) indicates T k In class C i The weight in tf(T ik ) indicates T k In class C i The frequency of occurrence in the idf(T k ) indicates T k In class C i The importance of D ac Denotes the inter-class dispersion, D ic represents the intra-class dispersion, n represents the number of categories, i=1,2,…,m.

[0021] In the above technical solution, preferably, step S3 specifically includes:

[0022] All sub-categories of industries in the sample set are taken as all associated factors related to the characteristic information;

[0023] Taking the sample set as the reference sequence and all the correlation factors as the comparison sequence, dimensionless processing is performed to obtain the grey correlation coefficients of the two sequences.

[0024] Calculating the correlation between the characteristic information and all correlation factors respectively, and sorting the correlation;

[0025] Set the correlation threshold, and use the correlation factor with a correlation greater than the correlation threshold as the final correlation factor input into the KNN classification model.

[0026] In the above technical solution, preferably, the sample set is used as a reference sequence, all the correlation factors are used as a comparison sequence, dimensionless processing is performed, and the grey correlation coefficients of the two sequences are obtained; the correlation degree between the feature information and all the correlation factors is calculated respectively, and the correlation degree ranking is performed, which specifically includes:

[0027] The reference sequence and the comparison sequence are dimensionless, and the calculation formula is:

[0028]

[0029] in, represents the mean of the i-th sequence, x i (k) represents the sequence before dimensionless transformation, x i ′(k) is the dimensionless sequence, k is the kth parameter of the sequence, i = 1, 2, ..., m, m is the number of comparison sequences, k = 1, 2, ..., n, n is the number of parameters;

[0030] Get the grey correlation coefficient of two sequences, and the calculation formula is:

[0031]

[0032] Among them, ξ i (k) is the correlation coefficient between the kth parameter of the ith comparison sequence and the kth parameter of the reference sequence, ρ is the resolution coefficient, and y(k) is the reference sequence;

[0033] The correlation between the characteristic information and all correlation factors is calculated respectively, and the calculation formula is:

[0034]

[0035] Among them, r i is the correlation degree of the ith comparison sequence.

[0036] On the basis of the above technical solution, preferably, step S4 specifically includes:

[0037] S401, dividing the correlation factors finally input into the KNN classification model into a test data set and a training data set;

[0038] S402, based on the training data set, an empirical selection method is used to select a K value, and a naive Bayes algorithm is used to select an optimal K value;

[0039] S403, calculating the distance between the test data set and each data in the training data set, and selecting K neighboring data from the training data set in the order of the distance from the test data set from small to large;

[0040] S404, taking the category with the largest number of occurrences among the K nearest neighbor data as the category of the input data.

[0041] Based on the above technical solution, preferably, in step S402, the empirical selection method is used to select the K value, which specifically includes:

[0042] The empirical selection method selects K values ​​that meet the following conditions:

[0043]

[0044] Where N represents the total number of training data sets, and n represents the feature dimension of the training data sets.

[0045] Based on the above technical solution, preferably, in step S402, selecting the optimal K value by using the naive Bayes algorithm specifically includes:

[0046] According to the naive Bayes algorithm, the selected known points are learned. Let X be the selected K value, Y be the category set, Y = {y1, y2, ..., yp}, learn through the naive Bayes formula, and its calculation formula is:

[0047]

[0048] Among them, P(Y|X) is the posterior probability of Y under condition X, P(X|Y) is the posterior probability of X under condition Y, P(X) is the prior probability of X, and P(Y) is the prior probability of Y;

[0049] The different K values ​​obtained by the selection method that satisfies the empirical formula are sequentially substituted into the Bayesian formula for learning, the posterior probabilities of the different K values ​​obtained are compared, and the K value with the highest posterior probability is selected.

[0050] Based on the above technical solution, preferably, in step S403, calculating the distance between each data in the test data set and the training data set specifically includes:

[0051] The Euclidean distance is used to calculate the data a in the test data set = (a1, a2, ..., a m ) and the data in the training data set b=(b1,b2,…,b m ), the calculation formula is:

[0052]

[0053] Among them, i=1,2,…,m, and m represents the feature dimension.

[0054] The present invention also proposes an industrial Internet industry chain classification system based on word segmentation technology, characterized in that the system includes:

[0055] The data acquisition module is used to acquire historical industrial Internet core industry chain data, analyze the historical industrial Internet core industry data chain, and obtain enterprise information related to each historical industrial Internet core industry chain data;

[0056] The data processing module is used to segment the information of industrial Internet enterprises, remove stop words and illegal format characters, and extract feature information from them;

[0057] A correlation calculation module is used to obtain several sub-industries of the core industries of the industrial Internet, take the several sub-industries as sample sets, use the grey correlation analysis method to process the characteristic information, and obtain the correlation factor between the characteristic information and the sample set;

[0058] The model building module is used to use the correlation factor as the input feature vector of the KNN classification model, optimize the KNN classification model based on the naive Bayes algorithm, and obtain the industrial Internet industry chain classification model;

[0059] The classification module is used to obtain the current industrial Internet core industrial chain data as the data to be classified, obtain the correlation factors between the data to be classified and several small industries, input the correlation factors between the data to be classified and several small industries into the industrial Internet industrial chain classification model, and obtain the small industry to which the current industrial Internet core industrial chain data belongs.

[0060] Compared with the prior art, the industrial Internet industry chain classification method and system based on word segmentation technology of the present invention has the following beneficial effects:

[0061] (1) Based on the historical Internet core industry chain data and the basic classification of the industrial Internet industry, the enterprise information related to the historical Internet core industry chain data is obtained; the characteristic information of the enterprise information is extracted, and the correlation factor of the final input KNN classification model is determined by the grey correlation analysis method; the industrial Internet industry chain classification model is generated based on the correlation factor training for classification; when classifying the Internet core industry chain data, the business scope of the relevant enterprises and the correlation factors between each enterprise and the sub-industry are combined to make the classification results more accurate.

[0062] (2) The grey correlation analysis method is used to calculate the correlation between the characteristic information of enterprise information and each sub-category of industry, and the correlation factor is obtained. The correlation factor is then used as the input of the KNN classification model, which accurately and efficiently handles the problems of data similarity and data fusion.

[0063] (3) The improved TF-IDF algorithm is used to extract feature information from enterprise information, which solves the problem of insufficient distribution of feature items between and within classes that is ignored by the traditional TF-IDF algorithm and improves the accuracy of feature information extraction.

[0064] (4) The K value of the KNN classification model is determined by using the empirical selection method and the naive Bayes algorithm, which reduces the computational complexity of the traditional KNN algorithm and improves the classification accuracy and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0066] Figure 1 A schematic diagram of a process flow of an industrial Internet industry chain classification method based on word segmentation technology of the present invention;

[0067] Figure 2A schematic diagram of a process for optimizing a KNN classification model in an industrial Internet industry chain classification method based on word segmentation technology of the present invention;

[0068] Figure 3 This is a structural schematic diagram of an industrial Internet industry chain classification system based on word segmentation technology in the present invention. DETAILED DESCRIPTION

[0069] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] The embodiment of the present invention provides an industrial Internet industry chain classification method based on word segmentation technology, and the specific steps are as follows: Figure 1 As shown, including but not limited to the following steps:

[0071] S1, obtain historical industrial Internet core industry chain data, analyze the historical industrial Internet core industry data chain, and obtain enterprise information related to each historical industrial Internet core industry chain data.

[0072] Specifically, historical industrial Internet core industry chain data is obtained, the historical industrial Internet core industry data chain is segmented, characteristic information of the historical industrial Internet industry chain data is extracted, and a first segmentation database is established;

[0073] Obtain the basic classification of the industrial Internet industry, perform word segmentation on the basic classification of the industrial Internet industry, extract feature information of the basic classification of the industrial Internet industry, and establish a second word segmentation database;

[0074] Based on the first word library and the second word library, enterprise information related to each historical industrial Internet core industrial chain data is obtained.

[0075] In this embodiment, the core industries of the Industrial Internet can be divided into five core industries, including digital equipment, interconnected automation, Internet network, Internet security, Internet platform and industrial software; the industrial Internet industry chain includes upstream, midstream and downstream. The upstream of the industrial Internet industry chain mainly supports basic industrial machine hardware equipment for data collection, storage, analysis and development; the industrial Internet platform in the midstream is oriented to the digitalization, networking and intelligent needs of the manufacturing industry, and builds a service system; the downstream is a variety of application scenarios. At present, the industrial Internet is widely used in vertical industries and fields such as construction, energy, transportation, medical care, smart cities, petrochemicals, steel, electronic information, home appliances, clothing, machinery, automobiles, equipment, aerospace, etc. In addition, the industrial Internet industry chain can be divided into six major parts: equipment layer, network layer, platform layer, software layer, application layer and security system. Based on the above classification and characteristics, the historical industrial Internet core industry data chain is segmented, the characteristic information of the historical industrial Internet industry chain data is extracted, the basic classification of the industrial Internet industry is segmented, and the characteristic information of the basic classification of the industrial Internet industry is extracted.

[0076] S2, segment the industrial Internet enterprise information, remove stop words and illegal format characters, and extract feature information from it.

[0077] Industrial Internet companies are divided into industrial Internet industry product and service types: including cloud platform manufacturers, Apps manufacturers, edge computing manufacturers, etc.; divided according to the nature of industrial Internet industry manufacturers, such as industrial equipment manufacturers, automation integration manufacturers, industrial software manufacturers, etc.; divided according to industrial Internet industry security system manufacturers, including firewall technology manufacturers, information encryption technology manufacturers, identity authentication technology manufacturers, security protocol manufacturers, intrusion detection system manufacturers, etc.

[0078] Furthermore, the improved TF-IDF algorithm is used to extract feature information, and the processed industrial Internet enterprise information is used as the corpus. The category set of the corpus is C = {C1, C2, ..., C m}, m is the number of categories;

[0079] The improved TF-IDF calculation formula is:

[0080]

[0081] Among them, T k represents the feature item, k represents the number of feature words, W(T ik ) indicates T k In class C i The weight in tf(T ik ) indicates T k In class C i The frequency of occurrence in the idf(Tk ) indicates T k In class C i The importance of D ac Denotes the inter-class dispersion, D ic represents the intra-class dispersion, n represents the number of categories, i=1,2,…,m.

[0082] The TF-IDF algorithm is the most frequently used feature weight algorithm. The traditional TF-IDF algorithm lacks information on the distribution of feature items between classes, within classes, and incomplete classifications, and is prone to extracting feature information locally, thereby ignoring feature words in other key positions. Therefore, the discreteness of the distribution of feature items between classes and within classes is introduced to improve the TF-IDF algorithm and improve the accuracy of feature information extraction.

[0083] S3, the core industry of the industrial Internet has several small industries. Taking the several small industries as sample sets, the grey correlation analysis method is used to process the characteristic information to obtain the correlation factor between the characteristic information and the sample set.

[0084] The core industries of the Industrial Internet have several sub-industries. For example, the energy Internet industry includes new energy power generation equipment manufacturers, energy storage product manufacturers, access equipment manufacturers, power software system research and development companies, etc.; energy equipment manufacturing includes energy hubs, energy switches, energy routers, energy network operators and virtual operators, information energy system integrators, information energy fusion application developers, etc.

[0085] Specifically, all sub-categories of industries in the sample set are taken as all associated factors related to the characteristic information;

[0086] Taking the sample set as the reference sequence and all the correlation factors as the comparison sequence, dimensionless processing is performed to obtain the grey correlation coefficients of the two sequences.

[0087] Calculating the correlation between the characteristic information and all correlation factors respectively, and sorting the correlation;

[0088] Set the correlation threshold, and use the correlation factor with a correlation greater than the correlation threshold as the final correlation factor input into the KNN classification model.

[0089] Taking the sample set as the reference sequence and all the correlation factors as the comparison sequence, dimensionless processing is performed to obtain the grey correlation coefficients of the two sequences; respectively calculating the correlation degree between the feature information and all the correlation factors, and sorting the correlation degree specifically includes:

[0090] The reference sequence and the comparison sequence are dimensionless, and the calculation formula is:

[0091]

[0092] in, represents the mean of the i-th sequence, x i (k) represents the sequence before dimensionless transformation, x i ′(k) is the dimensionless sequence, k is the kth parameter of the sequence, i = 1, 2, ..., m, m is the number of comparison sequences, k = 1, 2, ..., n, n is the number of parameters;

[0093] Get the grey correlation coefficient of two sequences, and the calculation formula is:

[0094]

[0095] Among them, ξ i (k) is the correlation coefficient between the kth parameter of the ith comparison sequence and the kth parameter of the reference sequence, ρ is the resolution coefficient, and y(k) is the reference sequence; ρ is the resolution coefficient, and its value range is generally [0,1]. In the embodiment of the present invention, it can be set to 0.5.

[0096] The correlation between the characteristic information and all correlation factors is calculated respectively, and the calculation formula is:

[0097]

[0098] Among them, r i is the correlation degree of the ith comparison sequence.

[0099] This embodiment processes the feature information based on the grey correlation analysis method to obtain a relatively small number of correlation factors related to several small industries. There is no need to input all the feature information into the KNN classification model for training. This improves the classification accuracy and greatly reduces the computational complexity of training the KNN classification model.

[0100] S4, taking the correlation factor as the input of the KNN classification model, optimizing the KNN classification model based on the Naive Bayes algorithm, and obtaining the industrial Internet industry chain classification model.

[0101] Further, the specific process of step S4 is as follows Figure 2 As shown, including but not limited to the following steps:

[0102] S401, dividing the correlation factors finally input into the KNN classification model into a test data set and a training data set.

[0103] S402, based on the training data set, an empirical selection method is used to select a K value, and a naive Bayes algorithm is used to select an optimal K value.

[0104] Specifically, the empirical selection method is used to select the K value, including:

[0105] The empirical selection method selects K values ​​that meet the following conditions:

[0106]

[0107] Where N represents the total number of training data sets, and n represents the feature dimension of the training data sets.

[0108] Selecting the optimal K value through the Naive Bayes algorithm specifically includes:

[0109] According to the naive Bayes algorithm, the selected known points are learned. Let X be the selected K value, Y be the category set, Y = {y1, y2, ..., y p}, learn through the naive Bayes formula, and its calculation formula is:

[0110]

[0111] Among them, P(Y|X) is the posterior probability of Y under condition X, P(X|Y) is the posterior probability of X under condition Y, P(X) is the prior probability of X, and P(Y) is the prior probability of Y;

[0112] The different K values ​​obtained by the selection method that satisfies the empirical formula are sequentially substituted into the Bayesian formula for learning, the posterior probabilities of the different K values ​​obtained are compared, and the K value with the highest posterior probability is selected.

[0113] It should be understood that the calculation methods of the prior probability and the posterior probability in the naive Bayes algorithm in this embodiment are all existing technologies and will not be described in detail here.

[0114] S403, calculating the distance between the test data set and each data in the training data set, and selecting K neighboring data from the training data set in the order of the distance from the test data set from small to large.

[0115] Specifically, calculating the distance between each data in the test data set and the training data set includes:

[0116] The Euclidean distance is used to calculate the data a in the test data set = (a1, a2, ..., a m ) and the data in the training data set b=(b1,b2,…,b m ), the calculation formula is:

[0117]

[0118] Among them, i=1,2,…,m, and m represents the feature dimension.

[0119] S404, taking the category with the largest number of occurrences among the K nearest neighbor data as the category of the input data.

[0120] This embodiment improves the KNN algorithm based on the empirical selection method and the naive Bayes algorithm. The range of the K value is first determined by the empirical selection method, and then the posterior probabilities of different K values ​​are obtained by the naive Bayes algorithm, and the K value with the highest posterior probability is selected. The K value obtained by this method has a higher accuracy rate, which improves the classification accuracy of the KNN classification model.

[0121] S5, obtain the current industrial Internet core industrial chain data as the data to be classified, obtain the correlation factors between the data to be classified and several small industries, input the correlation factors between the data to be classified and several small industries into the industrial Internet industrial chain classification model, and obtain the small industry to which the current industrial Internet core industrial chain data belongs.

[0122] In this embodiment, the processing method of the data to be classified is the same as the processing method of the historical industrial Internet core industrial chain data in steps S1 to S3. The correlation factor is obtained using the grey correlation analysis method in S3, so it will not be repeated here.

[0123] The embodiment of the present invention also provides an industrial Internet industry chain classification system based on word segmentation technology, such as Figure 3 As shown, it mainly includes a data acquisition module 1, a data processing module 2, a correlation calculation module 3, a model building module 4 and a classification module 5, wherein:

[0124] Data acquisition module 1 is used to acquire historical industrial Internet core industry chain data, analyze the historical industrial Internet core industry data chain, and obtain enterprise information related to each historical industrial Internet core industry chain data;

[0125] Data processing module 2, used to perform word segmentation, remove stop words and remove illegal format characters on the industrial Internet enterprise information, and extract feature information therefrom;

[0126] The correlation calculation module 3 is used to obtain several sub-industries of the core industries of the industrial Internet, take the several sub-industries as sample sets, use the grey correlation analysis method to process the characteristic information, and obtain the correlation factor between the characteristic information and the sample set;

[0127] Model building module 4 is used to use the correlation factor as the input feature vector of the KNN classification model, optimize the KNN classification model based on the naive Bayes algorithm, and obtain the industrial Internet industry chain classification model;

[0128] Classification module 5 is used to obtain the current industrial Internet core industrial chain data as the data to be classified, obtain the correlation factors between the data to be classified and several small industries, input the correlation factors between the data to be classified and several small industries into the industrial Internet industrial chain classification model, and obtain the small industry to which the current industrial Internet core industrial chain data belongs.

[0129] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A classification method for industrial Internet industry chain based on word segmentation technology, characterized in that: The method comprises: S1, obtain the historical industrial Internet core industry chain data, analyze the historical industrial Internet core industry data chain, and obtain the enterprise information related to each historical industrial Internet core industry chain data; S2, segmenting the industrial Internet enterprise information, removing stop words and illegal format characters, and extracting feature information from it; S3, the core industry of the industrial Internet has several sub-industries. The sub-industries are used as sample sets, and the characteristic information is processed by using the grey correlation analysis method to obtain the correlation factor between the characteristic information and the sample set; S4, taking the correlation factor as the input of the KNN classification model, optimizing the KNN classification model based on the naive Bayes algorithm, and obtaining the industrial Internet industry chain classification model; S5, obtaining the current industrial Internet core industrial chain data as the data to be classified, obtaining the correlation factors between the data to be classified and several sub-categories of industries, inputting the correlation factors between the data to be classified and several sub-categories of industries into the industrial Internet industrial chain classification model, and obtaining the sub-categories of industries to which the current industrial Internet core industrial chain data belongs; Step S1 specifically includes: Obtain historical industrial Internet core industry chain data, perform word segmentation on the historical industrial Internet core industry data chain, extract feature information of the historical industrial Internet industry chain data, and establish the first word segmentation database; Obtain the basic classification of the industrial Internet industry, perform word segmentation on the basic classification of the industrial Internet industry, extract feature information of the basic classification of the industrial Internet industry, and establish a second word segmentation database; Based on the first word library and the second word library, obtain the enterprise information related to each historical industrial Internet core industrial chain data; Step S2 specifically includes: The improved TF-IDF algorithm is used to extract feature information, and the processed industrial Internet enterprise information is used as the corpus. The category set of the corpus is C = {C1, C2, … ,C m }, m is the number of categories; The improved TF-IDF calculation formula is: Among them, T k represents the feature item, k represents the number of feature words, W(T ik ) indicates T k In class C i The weight in tf(T ik ) indicates T k In class C i The frequency of occurrence in the idf(T k ) indicates T k In class C i The importance of D ac represents the inter-class dispersion, D ic represents the intra-class dispersion, n represents the number of categories, i = 1, 2, … ,m; Step S3 specifically includes: All sub-categories of industries in the sample set are taken as all associated factors related to the characteristic information; Taking the sample set as the reference sequence and all the correlation factors as the comparison sequence, dimensionless processing is performed to obtain the grey correlation coefficients of the two sequences. Calculating the correlation between the characteristic information and all correlation factors respectively, and sorting the correlation; Set the correlation threshold, and use the correlation factor with a correlation greater than the correlation threshold as the final correlation factor input into the KNN classification model.

2. The industrial Internet industry chain classification method based on word segmentation technology as claimed in claim 1 is characterized in that: Taking the sample set as the reference sequence and all the correlation factors as the comparison sequence, dimensionless processing is performed to obtain the grey correlation coefficients of the two sequences. Calculating the correlation between the characteristic information and all correlation factors respectively and sorting the correlations specifically includes: The reference sequence and the comparison sequence are dimensionless, and the calculation formula is: in, represents the mean of the i-th sequence, x i (k) represents the sequence before dimensionless transformation, x i ′(k) is the dimensionless sequence, k is the kth parameter of the sequence, i=1,2, … ,m, m is the number of comparison sequences, k=1,2, … ,n, n is the number of parameters; Get the grey correlation coefficient of two sequences, and the calculation formula is: Among them, ξ i (k) is the correlation coefficient between the kth parameter of the ith comparison sequence and the kth parameter of the reference sequence, ρ is the resolution coefficient, and y(k) is the reference sequence; The correlation between the characteristic information and all correlation factors is calculated respectively, and the calculation formula is: Among them, r i is the correlation degree of the ith comparison sequence.

3. The industrial Internet industry chain classification method based on word segmentation technology as claimed in claim 1 is characterized in that: Step S4 specifically includes: S401, dividing the correlation factors finally input into the KNN classification model into a test data set and a training data set; S402, based on the training data set, an empirical selection method is used to select a K value, and a naive Bayes algorithm is used to select an optimal K value; S403, calculating the distance between the test data set and each data in the training data set, and selecting K neighboring data from the training data set in the order of the distance from the test data set from small to large; S404, taking the category with the largest number of occurrences among the K nearest neighbor data as the category of the input data.

4. The industrial Internet industry chain classification method based on word segmentation technology as claimed in claim 3 is characterized in that: In step S402, the empirical selection method is used to select the K value, specifically including: The empirical selection method selects K values ​​that meet the following conditions: Where N represents the total number of training data sets, and n represents the feature dimension of the training data sets.

5. The Red Leaf Internet Industry Chain classification method based on word segmentation technology as claimed in claim 4 is characterized in that: In step S402, selecting the optimal K value by using the naive Bayes algorithm specifically includes: According to the naive Bayes algorithm, the selected known points are learned. Let X be the selected K value, Y be the category set, Y = {y1, y2, … ,y p }, learn through the naive Bayes formula, and its calculation formula is: Among them, P(Y|X) is the posterior probability of Y under condition X, P(X|Y) is the posterior probability of X under condition Y, P(X) is the prior probability of X, and P(Y) is the prior probability of Y; The different K values ​​obtained by the selection method that satisfies the empirical formula are sequentially substituted into the Bayesian formula for learning, the posterior probabilities of the different K values ​​obtained are compared, and the K value with the highest posterior probability is selected.

6. The industrial Internet industry chain classification method based on word segmentation technology as claimed in claim 3 is characterized in that: In step S403, calculating the distance between each data in the test data set and the training data set specifically includes: The Euclidean distance is used to calculate the data a in the test data set = (a1, a2, ..., a m ) and the data in the training data set b=(b1,b2,…,b m ), the calculation formula is: Among them, i=1,2,…,m, and m represents the feature dimension.

7. An industrial Internet industry chain classification system based on word segmentation technology, characterized in that: Used to implement the industrial Internet industry chain classification method based on word segmentation technology as described in any one of claims 1 to 6, the system includes: The data acquisition module is used to acquire historical industrial Internet core industry chain data, analyze the historical industrial Internet core industry data chain, and obtain enterprise information related to each historical industrial Internet core industry chain data; The data processing module is used to segment the information of industrial Internet enterprises, remove stop words and illegal format characters, and extract feature information from them; A correlation calculation module is used to obtain several sub-industries of the core industries of the industrial Internet, take the several sub-industries as sample sets, use the grey correlation analysis method to process the characteristic information, and obtain the correlation factor between the characteristic information and the sample set; The model building module is used to use the correlation factor as the input of the KNN classification model, optimize the KNN classification model based on the naive Bayes algorithm, and obtain the industrial Internet industry chain classification model; The classification module is used to obtain the current industrial Internet core industrial chain data as the data to be classified, obtain the correlation factors between the data to be classified and several small industries, input the correlation factors between the data to be classified and several small industries into the industrial Internet industrial chain classification model, and obtain the small industry to which the current industrial Internet core industrial chain data belongs.

Images