Method and device for carrying out industrial chain classification on enterprise and related equipment

Abstract

The invention provides a method and device for carrying out industrial chain classification on enterprises and related equipment, and the method comprises the steps: obtaining the enterprise operation information of a to-be-classified enterprise, and taking the obtained enterprise operation information as to-be-analyzed data; performing word segmentation processing on the to-be-analyzed data by using a BERT pre-training model, and determining commodity code information corresponding to the enterprise according to a word segmentation processing result; determining a first semantic similarity between industry chain node information covered on the constructed commodity industry chain and the commodity code information and a second semantic similarity between the commodity industry chain node information and the keyword information by using the constructed industry chain similarity identification model; and according to the first semantic similarity and the second semantic similarity, determining an industry chain classification and grading result of the enterprise. The efficiency of classification and grading is effectively improved, and the accuracy of mapping enterprises to an industrial chain for classification and grading is improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device and related equipment for classifying an enterprise's industrial chain. Background Art

[0002] With the upgrading and transformation of the industry, industrial chain analysis has become an increasingly important reference data in the field of industrial planning. Currently, the most common industrial chain analysis methods on the market are based on public data, or manually determined by investigating each company, or batch analysis based on keywords, or NLP analysis using intellectual property data. These traditional industrial chain analysis methods either use outdated data from the industry analysis source, resulting in a certain period of time that the industrial chain analysis is out of touch with reality, and require a large amount of manpower for research and analysis, which takes a long time to analyze. In addition, the research data is not comprehensive and accurate enough, resulting in large errors in the analysis results, and cannot accurately reflect the current industrial chain. For this reason, how to provide a method with accurate analysis results, low analysis cost, and the ability to classify industrial chain analysis for a large number of companies has become a technical problem that needs to be solved in the industry. Summary of the invention

[0003] In view of this, the embodiments of the present application provide a method, device and related equipment for classifying the industrial chain of an enterprise to at least or partially solve the above-mentioned problems.

[0004] In a first aspect, an embodiment of the present application provides a method for classifying an enterprise's industrial chain, including:

[0005] Acquire business operation information of the enterprise to be classified, and use the business operation information as data to be analyzed;

[0006] Using the BERT pre-trained model to perform word segmentation processing on the data to be analyzed, and determining the product code information and keyword information corresponding to the enterprise according to the result of the word segmentation processing;

[0007] Using the constructed industrial chain similarity recognition model, determine the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information;

[0008] The industrial chain classification and grading result of the enterprise is determined according to the first semantic similarity and the second semantic similarity.

[0009] Optionally, in an embodiment of the present application, before using the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered by the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information, the method further includes:

[0010] Based on the NLP similarity analysis algorithm, a similarity recognition model for the commodity industry chain is constructed.

[0011] Optionally, in an embodiment of the present application, before using the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered by the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information, the method further includes:

[0012] Based on the preset commodity classification coding standards, determine the most detailed coding information suitable for the data to be analyzed;

[0013] One or more commodity industry chains are constructed according to the most detailed coding information adapted to the data to be analyzed.

[0014] Optionally, in an embodiment of the present application, the using of the BERT pre-trained model to perform word segmentation processing on the data to be analyzed, and determining the commodity coding information and keyword information corresponding to the enterprise according to the result of the word segmentation processing, includes:

[0015] Using the BERT pre-trained model, the WordPiece algorithm is used to perform word segmentation on the text lines of the data to be analyzed;

[0016] The output layer of the BERT pre-trained model is used to convert the result of the word segmentation processing into probability distribution information corresponding to different commodity code categories covered in the commodity code table;

[0017] The commodity coding information corresponding to the enterprise is determined according to the probability distribution information.

[0018] Optionally, in an embodiment of the present application, determining the most detailed level coding information suitable for the data to be analyzed based on a preset commodity classification coding standard includes:

[0019] Determine commodity coding entries based on preset commodity classification coding standards;

[0020] Extracting the most detailed coding information in the commodity coding entry;

[0021] The most detailed level coding information is screened to obtain the most detailed level coding information suitable for the current data to be analyzed.

[0022] Optionally, in an embodiment of the present application, before using the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered by the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information, the method further includes:

[0023] Perform keyword association on the final industrial chain nodes on the one or more commodity industrial chains constructed to obtain upstream and downstream industries associated with the final industrial chain nodes;

[0024] The commodity industry chain is expanded according to the upstream and downstream industries.

[0025] Optionally, in an embodiment of the present application, the method further comprises: configuring a pre-screening strategy based on the industry information, registered capital information, annual sales information of commodities, ranking information of association between commodities and the commodity industry chain, and proportion information of commodities in the annual sales of the enterprise;

[0026] Correspondingly, determining the industrial chain classification and grading result of the enterprise based on the first semantic similarity and the second semantic similarity includes: combining the pre-screening strategy, the first semantic similarity and the first semantic similarity to determine the industrial chain classification and grading information of the enterprise.

[0027] Optionally, in an embodiment of the present application, before performing word segmentation processing on the data to be analyzed using the BERT pre-trained model and determining the commodity coding information corresponding to the enterprise according to the result of the word segmentation processing, the method further includes: extracting entity information from the data to be identified and analyzed using a pre-trained entity recognition model;

[0028] The extracted entity information is used as the data to be analyzed, and participates in the processing process of the BERT pre-training model to perform word segmentation processing on the data to be analyzed, and the result of the word segmentation processing is obtained.

[0029] In a second aspect, based on the method for classifying an enterprise's industrial chain described in the first aspect of the present application, an embodiment of the present application further provides a device for classifying an enterprise's industrial chain, including:

[0030] An acquisition module, used to acquire business operation information of the enterprise to be classified, and use the business operation information as data to be analyzed;

[0031] A prediction module, used to perform word segmentation processing on the data to be analyzed using the BERT pre-training model, and determine the commodity coding information corresponding to the enterprise according to the result of the word segmentation processing;

[0032] A matching module, used to determine the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information by using the constructed industrial chain similarity recognition model;

[0033] A determination module is used to determine the industrial chain classification and grading result of the enterprise according to the first semantic similarity and the second semantic similarity.

[0034] In a third aspect, an embodiment of the present application further provides a computer storage medium having computer executable instructions stored thereon, and when the computer executable instructions are executed, any one of the methods for classifying an enterprise's industrial chain as described in the first aspect of the embodiment of the present application is performed.

[0035] In a fourth aspect, an embodiment of the present application further provides an electronic device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other through the communication bus;

[0036] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute any one of the methods for classifying the industrial chain of enterprises as described in the first aspect of the embodiment of the present application.

[0037] The present application provides a method, device and related equipment for classifying enterprises in the industrial chain, by obtaining the business information of the enterprise to be classified, and using the obtained business information as the data to be analyzed; using the BERT pre-training model to perform word segmentation on the data to be analyzed, and determining the commodity coding information corresponding to the enterprise according to the result of the word segmentation; using the constructed industrial chain similarity recognition model, determining the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information; according to the first semantic similarity and the second semantic similarity, determining the industrial chain classification and grading result of the enterprise. The method for classifying enterprises in the industrial chain disclosed in the present application uses the BERT pre-training model and its word segmenter to extract the enterprise's name, name and other business information correspondingly, so as to perform similarity analysis in multiple dimensions, and comprehensively determines the results of the classification and grading of the enterprise according to the results of the multi-dimensional similarity analysis, which effectively improves the efficiency of classification and grading, and improves the accuracy of mapping enterprises to the industrial chain for classification and grading. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present application 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 recorded in the embodiments of the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0039] Figure 1 A schematic diagram of a workflow of a method for classifying an enterprise's industrial chain provided in an embodiment of the present application;

[0040] Figure 2 A schematic diagram of the structure of a device for classifying the industrial chain of enterprises provided in an embodiment of the present application.

[0041] Figure 3 A schematic structural diagram of an electronic device is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0042] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the embodiments of the present application should fall within the scope of protection of the embodiments of the present application.

[0043] It should be understood that the various steps described in the method implementation of the present application can be performed in different orders and / or performed in parallel. In addition, the method implementation may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.

[0044] Embodiment 1

[0045] This application embodiment provides a method for classifying an enterprise's industrial chain, such as Figure 1 As shown, Figure 1 A workflow diagram of a method for classifying an enterprise's industrial chain provided in an embodiment of the present application includes:

[0046] Step S101, obtain the business information of the enterprise to be classified, and use the business information as the data to be analyzed. In the embodiment of the present application, the business information may specifically include, for example, enterprise name information, main commodity information or service information, etc., which is not limited in the embodiment of the present application. Specifically, this information can be obtained from the industrial and commercial registration information and the tax invoice information of the enterprise, and the obtained results are accurate and reliable, and the amount of data is relatively large and rich. At this stage, the embodiment of the present application is limited to obtaining the business information of the enterprise to be classified from the industrial and commercial registration information and the tax invoice information of the enterprise, so as to more accurately determine the industrial chain category and proportion level of the enterprise to be classified in the industrial chain.

[0047] Step S102, use the BERT pre-training model to perform word segmentation processing on the data to be analyzed, and determine the product coding information and keyword information corresponding to the enterprise according to the result of the word segmentation processing. BERT (Bidirectional Encoder Representations from Transformers) is a natural language processing (NLP) model based on deep learning. It adopts the Transformer architecture and is pre-trained on a large-scale corpus. One of the characteristics of BERT is its bidirectional processing capability, which can simultaneously take into account the context of all words in the input text sentence, not just the part before or after the word. This bidirectionality makes BERT perform well in natural language processing tasks, such as text classification, question answering, and named entity recognition. In this stage of the embodiment of the present application, word segmentation processing is performed through the BERT pre-training model, and the input enterprise information is converted into a corresponding vector representation. Based on the vector representation, the product coding information and keyword information of the corresponding data to be analyzed can be obtained more efficiently and accurately. The process is easy to implement.

[0048] Specifically, in an optional implementation method of the embodiment of the present application, the BERT pre-training model is used to perform word segmentation processing on the data to be analyzed, and the commodity coding information and keyword information corresponding to the enterprise are determined according to the result of the word segmentation processing, including: using the BERT pre-training model, using the WordPiece algorithm to perform word segmentation processing on the text line of the data to be analyzed; using the output layer of the BERT pre-training model to convert the result of the word segmentation processing into probability distribution information corresponding to different commodity coding categories covered in the commodity coding table; according to the probability distribution information, determining the commodity coding information corresponding to the enterprise. At this stage, the embodiment of the present application specifically limits the implementation process of determining the commodity coding information corresponding to the enterprise, and limits the BERT pre-training model to use the WordPiece algorithm to perform word segmentation processing on the text line of the data to be analyzed. The core idea of ​​the WordPiece algorithm is to split words into multiple prefix symbols (such as ## in BERT) minimum units, and then merge the minimum units into subword levels through subword merging rules. Then, the merge is performed through the merge rule, so as to construct a vocabulary list through cyclic iteration, so that the BERT pre-trained model can process unregistered words (OVV) to improve the completeness and accuracy of the word segmentation processing performed by the BERT pre-trained model. Furthermore, in an implementation method of an embodiment of the present application, the Encoder layer of the BERT pre-trained model is configured as a multi-layer TransformerEncoder stack, each layer containing a self-attention mechanism and a feedforward neural network, which can accurately capture the contextual dependencies in the input data text to be analyzed. Thereby better improving the accuracy of the word segmentation results. And its output layer is configured as a fully connected layer, which can convert the result output of the word segmentation processing into the probability distribution of the corresponding commodity code classification information, so as to determine the commodity coding information of the corresponding enterprise more comprehensively and accurately according to the probability distribution.

[0049] Step S103, using the constructed industrial chain similarity recognition model, determine the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information. In an embodiment of the present application, a commodity industrial chain covering a variety of commodity coding information and enterprise keywords is pre-constructed, and each node of the commodity industrial chain is based on the commodity coding information, and is also accompanied by corresponding commodity or service keywords, as well as commodity or service category information. At this stage, the embodiment of the present application uses the similarity matching of keyword information and industrial chain information, and the similarity matching of commodity coding information in these two dimensions as the basis for mapping the enterprise to be classified to the commodity industrial chain, which can relatively completely and accurately determine the accuracy of the final industrial chain classification and grading of the enterprise.

[0050] Step S104: Determine the industrial chain classification result of the enterprise based on the first semantic similarity and the second semantic similarity. The embodiment of the present application classifies and grades the industrial chain of the enterprises to be classified based on the two similarities, which can effectively ensure the accuracy of the classification and grading results.

[0051] Specifically, in an optional implementation of the embodiment of the present application, the determination of the industrial chain classification and grading result of the enterprise according to the first semantic similarity and the second semantic similarity includes: setting a first similarity threshold corresponding to the first semantic similarity, setting a second similarity threshold corresponding to the second semantic similarity, and mapping the enterprises whose first semantic similarity is greater than or equal to the first similarity threshold and whose second semantic similarity is greater than or equal to the second similarity threshold to the corresponding industrial chain node. The classification and grading result of the enterprise is determined according to the mapping result.

[0052] Furthermore, in an optional implementation method of an embodiment of the present application, the first similarity is the similarity of the commodity coding information, and the second similarity can be the similarity between the enterprise name and the commodity name corresponding to the node of the commodity industry chain, the similarity between the commodity name or service name operated by the enterprise and the commodity name corresponding to the node of the commodity industry chain, or the similarity between the enterprise's business scope information and the commodity information corresponding to the node of the commodity industry chain, thereby ensuring the accuracy and comprehensiveness of the similarity calculation results.

[0053] Optionally, in one embodiment of the present application, before using the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information, the method also includes: constructing the commodity industrial chain similarity recognition model based on the NLP similarity analysis algorithm. Among them, the NLP similarity analysis algorithm may include one or more text similarity algorithms in the field of natural language processing such as cosine similarity algorithm, Jaccard similarity, Word2Vec and word vector, BM25 algorithm or similarity calculation based on deep learning. At this stage, the embodiment of the present application limits the use of one or more text similarity algorithms to construct a model that can adapt to the data processing requirements of the processed data of different text types, and improve the accuracy and scene adaptability of the similarity calculation of the processed data.

[0054] Optionally, in an embodiment of the present application, before the method uses the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered on the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information, the method further includes: determining the most detailed coding information adapted to the data to be analyzed based on the preset commodity classification coding standard; and constructing one or more commodity industrial chains according to the most detailed coding information adapted to the data to be analyzed. The present application here exemplarily lists an example to illustrate its process, wherein the preset commodity classification coding standard is set as the "Tax Classification Code 51.0" standard, which includes 4666 coding entries, of which 3927 are the most detailed coding. According to the relationship between these most detailed coding information and the commodities covered in the data to be analyzed, one or more commodity industrial chains are pre-constructed for classifying and grading the enterprises to be classified corresponding to the data to be analyzed, and the process is easy to implement, and the classification results are accurate and reliable. This construction of the commodity industrial chain based on the most detailed coding information makes the industrial chain information reflected by the final result of the grading of the commodity industrial chain more detailed and accurate.

[0055] Specifically, in an optional implementation of the embodiment of the present application, based on the preset commodity classification coding standard, determining the most detailed coding information suitable for the data to be analyzed includes: determining the commodity coding entry based on the preset commodity classification coding standard; extracting the most detailed coding information in the commodity coding entry; screening and processing the most detailed coding information to obtain the most detailed coding information suitable for the current data to be analyzed. The present application here exemplifies an example to illustrate its process. In the embodiment of the present application, the obtained enterprise information cannot correspond to all the most detailed coding information. If in an actual application scenario of the embodiment of the present application, a large amount of tax invoice data is obtained through the tax administration department, and these large amounts of tax invoice data are used as the original sample data (data to be analyzed) of the enterprise to be classified, the commodity coding types covered are 5516, and these commodity coding types are reduced to 4061 after the commodity is substantially the same or the service is substantially the same. Then, 1455 unique commodity coding labels corresponding to some of the data to be classified are screened out. Further removing some of the rough commodity coding information of major categories reduced more than 480,000 data to be classified and their corresponding 378 major categories of rough commodity coding labels, making the number of commodity coding information items associated with the corresponding classification information of the enterprise currently to be processed 3,683. At this time, only one or more commodity industry chains covering 3,683 commodity coding information need to be built to meet the current actual scene work needs, reducing the amount of data that needs to be analyzed. The software and hardware resources consumed by the construction of the commodity industry chain are reduced, the utilization rate of the constructed commodity industry chain is improved, and the amount of useless information or information with extremely low utilization rate contained in the constructed commodity industry chain is reduced.

[0056] Optionally, in one implementation of the embodiment of the present application, before using the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information, the method also includes: performing keyword association on the final industrial chain nodes on the constructed one or more commodity industrial chains to obtain the upstream and downstream industries associated with the final industrial chain nodes; and expanding the commodity industrial chain according to the upstream and downstream industries. In this stage of the implementation of the present application, the length of the commodity industrial chain is further mined and expanded by keyword association, so that the constructed commodity industrial chain is more complete, thereby ensuring a more detailed classification and grading of the enterprises to be analyzed. Improve the accuracy of classification and grading.

[0057] Optionally, in an embodiment of the present application, the method further includes: configuring a pre-screening strategy based on the industry information, registered capital information, annual sales information of commodities, information on the association and ranking of commodities with the commodity industry chain, and information on the proportion of commodities in the annual sales of the enterprise; correspondingly, determining the classification and grading result of the enterprise's industry chain according to the first semantic similarity and the second semantic similarity includes: combining the pre-screening strategy, the first semantic similarity and the first semantic similarity to determine the classification and grading information of the enterprise's industry chain. At this stage of the present application embodiment, by setting a pre-screening strategy based on the above five dimensions, enterprises with relatively important industrial chain node relationships in the determined commodity industry chain can be mapped to the commodity industry chain for classification and grading, or classified and graded according to the main commodity items of the enterprise to be classified, without mapping enterprises that are related to the nodes of the commodity industry chain but have a small proportion and do not have much impact on the commodity industry chain to the commodity industry chain, thereby interfering with the application process of the user's industrial analysis based on the results of enterprise classification and grading and the mapping of enterprises to the commodity industry chain.

[0058] Optionally, in one embodiment of the present application, before using the BERT pre-trained model to perform word segmentation processing on the data to be analyzed, and determining the commodity coding information corresponding to the enterprise according to the result of the word segmentation processing, the method also includes: using the pre-trained entity recognition model to extract entity information from the data to be identified and analyzed; using the extracted entity information as the data to be analyzed, participating in the processing process of the BERT pre-trained model performing word segmentation processing on the data to be analyzed, and obtaining the result of the word segmentation processing. In this stage of the embodiment of the present application, through the extraction of entity information, such as after the entity recognition of the business information of the enterprise "AB City CD Chemical Co., Ltd.", the entity information of the business trunk information is extracted to obtain the entity "chemical", and after the entity recognition of the business information of the enterprise "EFG Super Spicy Chicken Leg Burger Without Chili Sauce", the commodity trunk information is extracted to obtain the entity information of "chicken leg burger", so as to perform data purification on the data to be analyzed, and use the purified data to support the calculation of the similarity model, which can effectively improve the accuracy and efficiency of similarity calculation.

[0059] The present application provides a method for classifying enterprises in the industrial chain, by obtaining the business information of the enterprise to be classified, and using the obtained business information as the data to be analyzed; using the BERT pre-training model to perform word segmentation on the data to be analyzed, and determining the commodity coding information corresponding to the enterprise according to the result of the word segmentation; using the constructed industrial chain similarity recognition model, determining the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information; according to the first semantic similarity and the second semantic similarity, determining the industrial chain classification and grading result of the enterprise. The method for classifying enterprises in the industrial chain disclosed in the present application uses the BERT pre-training model and its word segmenter to extract the enterprise's name, name and other business information accordingly, so as to perform similarity analysis in multiple dimensions, and comprehensively determine the results of the classification and grading of the enterprise according to the results of the multi-dimensional similarity analysis, which effectively improves the efficiency of classification and grading, and improves the accuracy of mapping enterprises to the industrial chain for classification and grading.

[0060] Embodiment 2

[0061] Based on the method for classifying the industrial chain of enterprises provided in the first embodiment of the present application, the embodiment of the present application also provides a device for classifying the industrial chain of enterprises, such as Figure 2 As shown, Figure 2 A schematic diagram of a device 20 for classifying an enterprise's industrial chain provided in an embodiment of the present application, wherein the device 20 for classifying an enterprise's industrial chain includes:

[0062] An acquisition module 201 is used to acquire business operation information of the enterprise to be classified and use the business operation information as data to be analyzed;

[0063] Prediction module 202, used to perform word segmentation processing on the data to be analyzed using the BERT pre-training model, and determine the commodity code information corresponding to the enterprise according to the result of the word segmentation processing;

[0064] The matching module 203 is used to determine the first semantic similarity between the industrial chain node information covered by the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information by using the constructed industrial chain similarity recognition model;

[0065] The determination module 204 is used to determine the industrial chain classification and grading result of the enterprise according to the first semantic similarity and the second semantic similarity.

[0066] Optionally, in one embodiment of the present application, the device 20 also includes a first construction module (not shown in the accompanying drawings), which is used to construct the commodity industry chain similarity recognition model based on the NLP similarity analysis algorithm before using the constructed industry chain similarity recognition model to determine the first semantic similarity between the industry chain node information covered in the constructed commodity industry chain and the commodity coding information, and the second semantic similarity between the industry chain node information and the keyword information.

[0067] Optionally, in one embodiment of the present application, the device 20 also includes a second construction module (not shown in the accompanying drawings), which is used to: determine the most detailed level coding information suitable for the data to be analyzed based on a preset commodity classification coding standard; and construct one or more of the commodity industry chains based on the most detailed level coding information suitable for the data to be analyzed, before determining the first semantic similarity between the industry chain node information covered in the constructed commodity industry chain and the commodity coding information, and the second semantic similarity between the industry chain node information and the keyword information using the constructed industry chain similarity recognition model.

[0068] Optionally, in one embodiment of the present application, the prediction module 202 is also used to: use the BERT pre-trained model to perform word segmentation on the text lines of the data to be analyzed using the WordPiece algorithm; use the output layer of the BERT pre-trained model to convert the result of the word segmentation processing into probability distribution information corresponding to different commodity coding categories covered in the commodity coding table; and determine the commodity coding information corresponding to the enterprise based on the probability distribution information.

[0069] Optionally, in one embodiment of the present application, the prediction module 202 is also used to: determine the commodity coding entry based on a preset commodity classification coding standard; extract the most detailed coding information in the commodity coding entry; and filter and process the most detailed coding information to obtain the most detailed coding information that is suitable for the current data to be analyzed.

[0070] Optionally, in one embodiment of the present application, the device 20 also includes an expansion module (not shown in the accompanying drawings), which is used to perform keyword association on the last-level industrial chain nodes on the constructed one or more commodity industrial chains before using the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information, so as to obtain upstream and downstream industries associated with the last-level industrial chain nodes; and expand the commodity industrial chain according to the upstream and downstream industries.

[0071] Optionally, in one embodiment of the present application, the device 20 also includes a screening module (not shown in the accompanying drawings), which is used to configure a pre-screening strategy based on the industry information of the enterprise, registered capital information, annual sales information of goods, sorting information of the association between goods and the commodity industry chain, and the proportion of goods in the annual sales of the enterprise; correspondingly, the determination module 204 is also used to determine the industrial chain classification and grading information of the enterprise in combination with the pre-screening strategy, the first semantic similarity and the first semantic similarity.

[0072] Optionally, in one embodiment of the present application, the device 20 also includes an entity extraction module (not shown in the drawings), which is used to use a pre-trained entity recognition model to extract entity information from the data to be identified and analyzed before using the BERT pre-trained model to perform word segmentation on the data to be analyzed, and determining the product coding information corresponding to the enterprise according to the result of the word segmentation processing; the extracted entity information is used as the data to be analyzed, and participates in the processing process of the BERT pre-trained model performing word segmentation on the data to be analyzed, to obtain the result of the word segmentation processing.

[0073] The present application provides a device for classifying enterprises in the industrial chain, wherein an acquisition module is set to acquire the business information of the enterprise to be classified, and the acquired business information is used as the data to be analyzed; a prediction module is set to perform word segmentation processing on the data to be analyzed using the BERT pre-training model, and the commodity coding information corresponding to the enterprise is determined according to the result of the word segmentation processing; an identification module is set to use the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information; a determination module is set to determine the industrial chain classification and grading result of the enterprise according to the first semantic similarity and the second semantic similarity. The device for classifying enterprises in the industrial chain disclosed in the present application uses the BERT pre-training model and its word segmenter to extract the enterprise's name, name and other business information correspondingly, so as to perform similarity analysis in multiple dimensions, and comprehensively determines the results of classifying and grading the enterprise based on the results of the multi-dimensional similarity analysis, which effectively improves the efficiency of classification and grading, and improves the accuracy of mapping the enterprise to the industrial chain for classification and grading. And the structure of the device is simple and easy to implement.

[0074] Embodiment 3

[0075] An embodiment of the present application also provides a storage medium on which a computer program is stored. When the program is executed by a processor, any one of the methods for classifying the industrial chain of an enterprise as described in the first embodiment of the present application is implemented.

[0076] Embodiment 4:

[0077] The present application also provides an electronic device, such as Figure 3 As shown, Figure 3 A schematic diagram of the structure of an electronic device 30 is provided for an embodiment of the present application. The electronic device 30 includes:

[0078] One or more processors 301, communication interface 302, memory 303 and communication bus 304, the processor 301, memory 303 and communication interface 302 communicate with each other via the communication bus 304;

[0079] Memory 303, used to store one or more programs;

[0080] When the one or more programs are executed by the one or more processors 301, the one or more processors 301 implement any one of the methods for classifying the industrial chain of an enterprise as described in Example 1 of the present application.

[0081] So far, the application has described specific embodiments of the subject matter. In some cases, the actions recorded in the claims can be performed in different orders and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing can be advantageous.

[0082] In the 1990s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the method flow). However, with the development of technology, many improvements to the method flow today can be regarded as direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system layer on a PLD by programming it themselves, without having to ask a chip manufacturer to design and produce a dedicated integrated circuit chip. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0083] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.

[0084] The system layers, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0085] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0086] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0087] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, system layers or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt 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.) that contain computer-usable program code.

[0088] The present application may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific transactions or implement specific abstract data types. The present application may also be practiced in distributed computing environments where transactions are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0089] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system layer embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0090] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for classifying enterprises by industry chain, characterized in that: include: Acquire business operation information of the enterprise to be classified, and use the business operation information as data to be analyzed; Using the BERT pre-trained model to perform word segmentation processing on the data to be analyzed, and determining the product code information and keyword information corresponding to the enterprise according to the result of the word segmentation processing; Using the constructed industrial chain similarity recognition model, determine the first semantic similarity between the industrial chain node information covered by the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information; The industrial chain classification and grading result of the enterprise is determined according to the first semantic similarity and the second semantic similarity.

2. The method for classifying the industrial chain of enterprises according to claim 1, characterized in that: Before using the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered by the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information, the method further includes: Based on the NLP similarity analysis algorithm, a similarity recognition model for the commodity industry chain is constructed.

3. The method for classifying enterprises in the industrial chain according to claim 1 is characterized in that: Before determining the first semantic similarity between the industrial chain node information covered by the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information by using the constructed industrial chain similarity recognition model, the method further includes: Based on the preset commodity classification coding standards, determine the most detailed coding information suitable for the data to be analyzed; One or more commodity industry chains are constructed according to the most detailed coding information adapted to the data to be analyzed.

4. The method for classifying the industrial chain of enterprises according to claim 1 is characterized in that: The using of the BERT pre-training model to perform word segmentation processing on the data to be analyzed, and determining the commodity code information and keyword information corresponding to the enterprise according to the result of the word segmentation processing, includes: Using the BERT pre-trained model, the WordPiece algorithm is used to perform word segmentation on the text lines of the data to be analyzed; The output layer of the BERT pre-trained model is used to convert the result of the word segmentation processing into probability distribution information corresponding to different commodity code categories covered in the commodity code table; The commodity coding information corresponding to the enterprise is determined according to the probability distribution information.

5. The method for classifying enterprises in the industrial chain according to claim 3 is characterized in that: The step of determining the most detailed coding information suitable for the data to be analyzed based on the preset commodity classification coding standard includes: Determine commodity coding entries based on preset commodity classification coding standards; Extracting the most detailed coding information in the commodity coding entry; The most detailed level coding information is screened to obtain the most detailed level coding information suitable for the current data to be analyzed.

6. The method for classifying the industrial chain of enterprises according to claim 1, characterized in that: Before using the constructed industrial chain similarity recognition model to determine the first semantic similarity between the industrial chain node information covered by the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information, the method further includes: Perform keyword association on the final industrial chain nodes on the one or more commodity industrial chains constructed to obtain upstream and downstream industries associated with the final industrial chain nodes; The commodity industry chain is expanded according to the upstream and downstream industries.

7. The method for classifying the industrial chain of enterprises according to claim 1 is characterized in that: The method further includes: configuring a pre-screening strategy based on the industry information, registered capital information, annual sales information of commodities, ranking information of association between commodities and the commodity industry chain, and the proportion of commodities in the annual sales of the enterprise; Correspondingly, determining the industrial chain classification and grading result of the enterprise based on the first semantic similarity and the second semantic similarity includes: combining the pre-screening strategy, the first semantic similarity and the first semantic similarity to determine the industrial chain classification and grading information of the enterprise.

8. The method for classifying the industrial chain of enterprises according to claim 1, characterized in that: Before performing word segmentation processing on the data to be analyzed using the BERT pre-trained model and determining the commodity coding information corresponding to the enterprise according to the result of the word segmentation processing, the method further includes: extracting entity information from the data to be identified and analyzed using the pre-trained entity recognition model; The extracted entity information is used as the data to be analyzed, and participates in the processing process of the BERT pre-training model to perform word segmentation processing on the data to be analyzed, and the result of the word segmentation processing is obtained.

9. A device for classifying enterprises in the industrial chain, characterized in that: include: An acquisition module, used to acquire business operation information of the enterprise to be classified, and use the business operation information as data to be analyzed; A prediction module, used to perform word segmentation processing on the data to be analyzed using the BERT pre-training model, and determine the commodity coding information corresponding to the enterprise according to the result of the word segmentation processing; A matching module, used to determine the first semantic similarity between the industrial chain node information covered in the constructed commodity industrial chain and the commodity coding information, and the second semantic similarity between the industrial chain node information and the keyword information by using the constructed industrial chain similarity recognition model; A determination module is used to determine the industrial chain classification and grading result of the enterprise according to the first semantic similarity and the second semantic similarity.

10. A computer storage medium, characterized in that: The computer storage medium stores computer executable instructions, and when the computer executable instructions are executed, the method for classifying the industrial chain of enterprises as described in any one of claims 1 to 8 is executed.

Images