Strategic emerging industry classification method and device, storage medium and electronic equipment

Abstract

Embodiments of the present application provide a strategic emerging industry classification method and device, a computer readable storage medium and an electronic device, related information of an enterprise to be classified is obtained, a classification result of strategic emerging industry classification of the enterprise is obtained according to the related information, wherein obtaining the classification result comprises: inputting the related information into a trained recognition model, and obtaining the classification result based on the trained recognition model; and / or classifying the related information based on a classification rule to obtain the classification result; the classification rule is determined according to a classification system of strategic emerging industry classification; in this way, automatic classification of strategic emerging industry of the enterprise is realized according to the related information of the enterprise to be classified, which is fast, accurate and greatly saves the labor cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information classification, in particular to a strategic emerging industry classification method and device, a computer readable storage medium and an electronic device. BACKGROUND

[0002] Strategic emerging industry is an industry based on major technological breakthroughs and major development needs, which has a significant leading and driving effect on the overall and long-term development of the economy and society, is knowledge and technology intensive, has less material resource consumption, great growth potential, and good comprehensive benefits, and includes nine fields such as new generation information technology industry, high-end equipment manufacturing industry, new material industry, biological industry, new energy vehicle industry, new energy industry, energy saving and environmental protection industry, digital creative industry, and related service industry. The Strategic Emerging Industry Classification (2018) provides a standard and basis for the classification of strategic emerging industries.

[0003] At present, experts generally manually label the strategic emerging industry category of an enterprise. On the one hand, manual labeling is highly dependent on experts, and experts need to have sufficient domain knowledge. On the other hand, labeling is difficult, time-consuming, labor-intensive, and costly. SUMMARY

[0004] Therefore, the embodiments of the present application provide a strategic emerging industry classification method and device, a computer readable storage medium and an electronic device to solve at least one problem in the background art.

[0005] In a first aspect, an embodiment of the present application provides a strategic emerging industry classification method, which comprises:

[0006] Obtaining related information of an enterprise to be classified;

[0007] Obtaining a classification result of strategic emerging industry classification of the enterprise according to the related information; wherein, obtaining the classification result comprises:

[0008] Inputting the related information into a trained recognition model, and obtaining the classification result based on the trained recognition model; and / or,

[0009] Classifying the related information based on a classification rule to obtain the classification result; the classification rule is determined according to a classification system of strategic emerging industry classification.

[0010] In combination with the first aspect of the present application, in an optional implementation manner, obtaining the classification result specifically comprises:

[0011] inputting the related information into a trained recognition model, and obtaining a first classification result based on the trained recognition model; the first classification result comprises a classification category of strategic emerging industry classification corresponding to the enterprise or first information, and the first information indicates that no classification category of strategic emerging industry classification corresponding to the enterprise is obtained;

[0012] corresponding to a case where the first classification result comprises the first information, classifying the related information based on a classification rule to obtain a second classification result.

[0013] With reference to the first aspect of the present application, in an optional implementation, the classifying the related information based on the classification rule to obtain the classification result comprises:

[0014] determining, based on a semantic similarity model, a classification category of strategic emerging industry classification that meets a preset condition in terms of similarity to the related information, and obtaining the classification result according to the classification category; and / or,

[0015] determining at least one frequent item set that meets a classification requirement according to a classification system of strategic emerging industry classification, and classifying the related information based on the determined frequent item set to obtain the classification result.

[0016] With reference to the first aspect of the present application, in an optional implementation, the classifying the related information based on the classification rule to obtain the classification result specifically comprises:

[0017] determining, based on a semantic similarity model, a classification category of strategic emerging industry classification that meets a preset condition in terms of similarity to the related information, and obtaining a third classification result according to the classification category; the third classification result comprises a classification category of strategic emerging industry classification corresponding to the enterprise or third information, and the third information indicates that no classification category of strategic emerging industry classification corresponding to the enterprise is obtained; corresponding to a case where the third classification result comprises a classification category of strategic emerging industry classification corresponding to the enterprise, the third classification result is determined as the second classification result.

[0018] corresponding to a case where the third classification result comprises the third information, determining at least one frequent item set that meets a classification requirement according to a classification system of strategic emerging industry classification, and classifying the related information based on the determined frequent item set to obtain a fourth classification result; the fourth classification result comprises a classification category of strategic emerging industry classification corresponding to the enterprise or fourth information, and the fourth information indicates that no classification category of strategic emerging industry classification corresponding to the enterprise is obtained; corresponding to a case where the fourth classification result comprises a classification category of strategic emerging industry classification corresponding to the enterprise, the fourth classification result is determined as the second classification result.

[0019] In an optional implementation of the first aspect of the present application, the determining of the classification category of the strategic emerging industry classification that meets the preset condition in terms of the similarity to the relevant information based on the semantic similarity model comprises:

[0020] determining a first classification category of the strategic emerging industry classification that meets the preset condition in terms of the similarity to the relevant information based on a first semantic similarity model;

[0021] determining a second classification category of the strategic emerging industry classification that meets the preset condition in terms of the similarity to the relevant information based on a second semantic similarity model;

[0022] determining a third classification result according to the first classification category and the second classification category; the third classification result comprises a classification category of the strategic emerging industry classification corresponding to the enterprise or third information, the third information indicating that no classification category of the strategic emerging industry classification corresponding to the enterprise is obtained;

[0023] wherein, in a case where there is an intersection corresponding to the first classification category and the second classification category, the third classification result comprises the classification category of the strategic emerging industry classification corresponding to the enterprise and the classification category of the strategic emerging industry classification is determined according to the intersection; in a case where there is no intersection corresponding to the first classification category and the second classification category, the third classification result comprises the third information;

[0024] wherein, in a case where the third classification result comprises the classification category of the strategic emerging industry classification corresponding to the enterprise, the third classification result is determined as a second classification result.

[0025] In an optional implementation of the first aspect of the present application, the obtaining of the classification result further comprises:

[0026] classifying the relevant information based on a preset rule to obtain a fifth classification result; the fifth classification result comprises a classification category of the strategic emerging industry classification corresponding to the enterprise or fifth information, the fifth information indicating that no classification category of the strategic emerging industry classification corresponding to the enterprise is obtained;

[0027] wherein, in a case where the fifth classification result comprises the classification category of the strategic emerging industry classification corresponding to the enterprise, the fifth classification result is output as the determined classification result.

[0028] With reference to the first aspect of the present application, in an optional implementation, the second classification result comprises a classification category of a strategic emerging industry classification corresponding to the enterprise or second information, the second information indicating that the enterprise does not obtain a classification category of a strategic emerging industry classification;

[0029] corresponding to the fifth classification result comprises the fifth information, and the second classification result comprises a classification category of a strategic emerging industry classification corresponding to the enterprise, the second classification result is output as the determined classification result;

[0030] corresponding to the fifth classification result comprises the fifth information, the second classification result comprises the second information, and the first classification result comprises a classification category of a strategic emerging industry classification corresponding to the enterprise, the first classification result is output as the determined classification result;

[0031] corresponding to the fifth classification result comprises the fifth information, the second classification result comprises the second information, and the first classification result comprises the first information, the classification result is that the enterprise does not belong to a strategic emerging industry.

[0032] With reference to the first aspect of the present application, in an optional implementation, the relevant information comprises at least one of the following: an enterprise name, a business scope, and a brief introduction.

[0033] With reference to the first aspect of the present application, in an optional implementation, the relevant information comprises an enterprise name, a business scope, and a brief introduction; the trained identification model comprises a trained Bert model; the trained Bert model comprises three different Segment Embeddings corresponding to the enterprise name, the business scope, and the brief introduction, respectively.

[0034] With reference to the first aspect of the present application, in an optional implementation, the inputting of the relevant information into the trained identification model and the obtaining of the classification result based on the trained identification model comprise:

[0035] inputting the relevant information into the trained identification model, performing input embedding to obtain vector representations of each text block of the relevant information;

[0036] performing processing on the vector representations of each text block using an average pooling layer to obtain a processed vector; and performing classification based on the processed vector to obtain the classification result.

[0037] In an optional implementation of the first aspect of the present application, the trained recognition model comprises a trained Bert model; and the pre-training process of the trained Bert model comprises domain self-adaptive pre-training and / or task self-adaptive pre-training.

[0038] In an optional implementation of the first aspect of the present application, the pre-training process of the trained Bert model comprises task self-adaptive pre-training; and the task self-adaptive pre-training is based on label information of national economic industry classification.

[0039] In the second aspect, an embodiment of the present application provides a strategic emerging industry classification device, comprising:

[0040] an acquisition module configured to acquire relevant information of an enterprise to be classified;

[0041] a classification module configured to obtain a classification result of strategic emerging industry classification of the enterprise according to the relevant information; wherein the obtaining of the classification result comprises:

[0042] inputting the relevant information into a trained recognition model, and obtaining the classification result based on the trained recognition model; and / or,

[0043] classifying the relevant information based on a classification rule to obtain the classification result; the classification rule is determined according to a classification system of strategic emerging industry classification.

[0044] In the third aspect, an embodiment of the present application provides a computer readable storage medium, which stores instructions, when the instructions are executed by a processor of an electronic device, the electronic device can execute the strategic emerging industry classification method provided in any one of the above embodiments.

[0045] In the fourth aspect, an embodiment of the present application provides an electronic device, comprising:

[0046] a processor;

[0047] a memory for storing computer executable instructions;

[0048] the processor is configured to execute the computer executable instructions to implement the strategic emerging industry classification method in any one of the above embodiments.

[0049] The strategic emerging industry classification method and device, the computer readable storage medium and the electronic device provided by the embodiments of the present application obtain the related information of an enterprise to be classified; obtain the classification result of the strategic emerging industry classification of the enterprise according to the related information; wherein, obtaining the classification result comprises: inputting the related information into a trained recognition model, and obtaining the classification result based on the trained recognition model; and / or classifying the related information based on a classification rule to obtain the classification result; the classification rule is determined according to the classification system of the strategic emerging industry classification; in this way, the automatic classification of the strategic emerging industry of the enterprise is realized according to the related information of the enterprise to be classified, which is fast, accurate and greatly saves the labor cost.

[0050] Additional aspects and advantages of the present application will be made apparent by the following description and the specific examples. BRIEF DESCRIPTION OF DRAWINGS

[0051] The accompanying drawings, which are included to provide a further understanding of the present application and constitute a part of this application, illustrate embodiments of the present application and together with the description serve to explain the present application. In the drawings:

[0052] Figure 1 The application scenario diagram of the strategic emerging industry classification method provided by an embodiment of the present application is shown in the figure;

[0053] Figure 2 The flowchart of the strategic emerging industry classification method provided by an embodiment of the present application is shown in the figure;

[0054] Figure 3 The web page containing the related information of the enterprise to be classified provided by an embodiment of the present application is shown in the figure;

[0055] Figure 4a The diagram of the code structure of the categories except the new material industry in the strategic emerging industry classification is shown in the figure;

[0056] Figure 4b The diagram of the code structure of the new material industry in the strategic emerging industry classification is shown in the figure;

[0057] Figure 5 The flowchart of the strategic emerging industry classification method provided by a specific example is shown in the figure;

[0058] Figure 6 The framework structure diagram of the Bert model provided by a specific example is shown in the figure;

[0059] Figure 7 The structure diagram of the strategic emerging industry classification device provided by an embodiment of the present application is shown in the figure;

[0060] Figure 8 A structural schematic diagram of an electronic device provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0061] In order to make the technical solutions and advantages of the present application more obvious and understandable, the technical solutions in the embodiments of the present application are described clearly and completely below by way of listing specific embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0062] The strategic emerging industry classification method provided by the embodiments of the present application can be specifically applied in an electronic device, which can be a terminal or a server or the like.

[0063] It can be understood that the strategic emerging industry classification method provided by the embodiments of the present application can be executed on a terminal device, or executed on a server, or executed by the terminal device and the server together. The above examples should not be understood as a limitation of the present application.

[0064] Figure 1 A schematic diagram of an application scenario of the strategic emerging industry classification method provided by an embodiment of the present application is shown. Taking the terminal device and the server executing the strategic emerging industry classification method together as an example, Figure 1 The shown scenario includes a server 10 and a terminal device 20 in communication connection with the server 10. Exemplarily, in actual application process, the terminal device 20 can receive an instruction input by a user, and send the received instruction to the server 10; the instruction for example includes an instruction for indicating obtaining a classification result of strategic emerging industry classification of a certain enterprise. The server 10 is used to acquire relevant information of an enterprise to be classified; according to the relevant information, obtain the classification result of strategic emerging industry classification of the enterprise; and output the classification result to the terminal device 20; the terminal device 20 can perform corresponding display based on the received classification result, so as to present to the user.

[0065] The terminal device 20 can include a mobile phone, a smart television, a tablet computer, a notebook computer, a personal computer (PC, Personal Computer), a wearable device, or a vehicle-mounted computer, etc. The terminal device 20 can further be provided with a client, which can be an application program client or a browser client, etc.

[0066] Next, the strategic emerging industry classification method provided by an embodiment of the present application is briefly introduced.

[0067] Figure 2As shown in the figure, the strategic emerging industry classification method provided by the embodiment of the present application includes the following steps:

[0068] In step 210, the relevant information of the enterprise to be classified is obtained.

[0069] Here, step 210 can be performed by an electronic device. Specifically, the electronic device can obtain all the required relevant information on the network according to some or several aspects of information of the enterprise; the electronic device can also obtain the relevant information of the enterprise to be classified input by the user.

[0070] The relevant information of the enterprise to be classified can include at least one of the following: enterprise name, business scope, and introduction. In a specific application, the relevant information includes, for example, the enterprise name, business scope, and introduction. It should be understood that the present application is not limited thereto, and the relevant information can include any one or several aspects of information that can be used as the basis for strategic emerging industry classification; for example, the relevant information can include the main products of the enterprise, etc.

[0071] Figure 3 The web page provided by an embodiment of the present application contains the relevant information of the enterprise to be classified; as shown in the figure, for ABC Automobile Co., Ltd., multiple aspects of information related to it can be obtained on the network, such as its legal representative, unified social credit code, telephone, official website, email, address, etc. Considering the need to determine the classification of strategic emerging industries, as an optional implementation, the following relevant information is selected: enterprise name, business scope, and introduction. It should be noted that the relevant information is not limited to the content in the "enterprise name", "business scope", or "introduction" field, i.e., it is not limited to Figure 3 the information disclosed on the web page shown in the figure; the specific content is the relevant information reflecting the name of the enterprise, the business scope of the enterprise, and the introduction of the enterprise, i.e., the "company name" displayed on a certain web page should obviously be recognized as the relevant information of the enterprise name according to the embodiment of the present application.

[0072] In step 220, the classification result of the strategic emerging industry classification of the enterprise is obtained according to the relevant information.

[0073] Among them, obtaining the classification result includes: inputting the relevant information into a trained recognition model, and obtaining the classification result based on the trained recognition model; and / or classifying the relevant information based on the classification rule to obtain the classification result; the classification rule is determined according to the classification system of the strategic emerging industry classification.

[0074] Step 220 can also be performed by an electronic device. It can be understood that, whether the classification result is obtained based on the trained recognition model, or the classification result is obtained based on classification rules, or the classification result is obtained based on both, is an automatic classification method based on computer technology. According to the related information of the enterprise to be classified, the embodiments of the present application realize the automatic classification of whether the enterprise belongs to the strategic emerging industry and which category of the strategic emerging industry, the classification speed is fast, the accuracy is high, and the human cost is greatly saved.

[0075] The step of obtaining the classification result by using the trained recognition model alone, or the step of obtaining the classification result by classifying the related information based on the classification rules alone, can obtain the classification category of a part of enterprises corresponding to the strategic emerging industry classification; of course, there is a possibility of missing recognition by using a single step. Using the trained recognition model and the classification rules to obtain the classification result can improve the coverage rate of recognition.

[0076] The Strategic Emerging Industry Classification (2018) is compiled based on the Strategic Emerging Industry Key Products and Services Guidance Catalogue (2016) and other relevant national documents, and is based on the current National Economic Industry Classification (GB / T 4754-2017). It reclassifies relevant activities that meet the characteristics of "strategic emerging industry" and provides standards and basis for the classification of strategic emerging industry. This classification is an independent classification system, which uses linear classification method, hierarchical and variable incremental coding method. The main code of this classification is divided into one, two and three layers, and the new material industry uses variable incremental coding and increases to four layers. All coding layers are separated by "." and each layer is coded by Arabic numerals. Figure 4a and Figure 4b The category code structure of the strategic emerging industry classification except for the new material industry and the code structure of the new material industry in the strategic emerging industry classification are shown respectively. For categories other than the new material industry, if the second layer is not further divided, the third layer code is supplemented with a "0". If the third layer of the new material industry is not further divided, the fourth layer code is supplemented with a "0". There are 9 categories in the first layer, 40 categories in the second layer, 189 categories in the third layer, and 166 categories in the fourth layer of this classification. If the second last layer is not divided, the first count is supplemented with 0.

[0077] The classification result obtained in the embodiments of the present application can be to determine that the enterprise belongs to a certain category in the first layer, or to determine that the enterprise belongs to a certain category in the second layer, or to determine that the enterprise belongs to a certain category in the third layer / fourth layer; in other words, the present application does not specifically limit to which layer the enterprise is classified; for example, the classification result includes a certain specific category in the second layer of the strategic emerging industry classification. The certain category in a certain layer can be presented in the form of the above-mentioned code, of course, it can also be presented in the form of the name of the strategic emerging industry classification.

[0078] It can be understood that if the classification result contains a specific category, it means that the enterprise belongs to the strategic emerging industry; on the contrary, if the classification result does not contain any specific category, it means that the enterprise does not belong to the strategic emerging industry.

[0079] As an optional implementation, obtaining the classification result specifically includes:

[0080] inputting the relevant information into the trained recognition model, and obtaining a first classification result based on the trained recognition model; the first classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or first information, and the first information represents that no classification category of the strategic emerging industry classification corresponding to the enterprise is obtained;

[0081] corresponding to the case that the first classification result includes the first information, classifying the relevant information based on the classification rule to obtain a second classification result.

[0082] It can be understood that the trained recognition model is used for classification first, and when no classification category of the corresponding strategic emerging industry classification is obtained, the classification rule is used for classification. When the classification category of the corresponding strategic emerging industry classification is obtained, the step of using the classification rule for classification is no longer executed, and the first classification result is stored. In this way, the trained recognition model can cover relatively more samples, and considering that the samples that can be covered by the classification rule can be relatively small, therefore, the embodiments are beneficial to improve the efficiency of classification.

[0083] The first classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or the first information. As mentioned above, if the classification category of the strategic emerging industry classification corresponding to the enterprise is included in the first classification result, it indicates that the enterprise belongs to the strategic emerging industry; at this time, the first classification result also includes information opposite to the first information, i.e., information representing obtaining the classification category of the strategic emerging industry classification corresponding to the enterprise and specific category information. On the contrary, if the first information is included in the first classification result, it can be understood that no specific category is included in the first classification result. At this time, it can be preliminarily judged that the enterprise does not belong to the strategic emerging industry. Of course, in the embodiment in which the classification result is obtained based on the classification rule, if the classification category of the strategic emerging industry classification corresponding to the enterprise is obtained based on the classification rule, the preliminary judgment is corrected.

[0084] It should be noted that the classification category of the strategic emerging industry classification corresponding to the enterprise or the first information appears in the first classification result in an alternative manner, and the classification category of the strategic emerging industry classification corresponding to the enterprise and the first information cannot appear in the first classification result at the same time, nor can the classification category of the strategic emerging industry classification corresponding to the enterprise and the first information not appear in the first classification result.

[0085] Similarly, the second classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or the second information, and the second information represents that no classification category of the strategic emerging industry classification corresponding to the enterprise is obtained. The relationship between the classification category of the strategic emerging industry classification corresponding to the enterprise and the second information in the second classification result is similar to the corresponding relationship in the first classification result, which will not be described here; in addition, it can be understood that the relationship between the classification category of the strategic emerging industry classification corresponding to the enterprise and the corresponding information in the third classification result, the fourth classification result and the fifth classification result recorded below is also similar to the corresponding relationship in the first classification result.

[0086] As an optional implementation, the classification of the related information based on the classification rule to obtain the classification result includes: determining a classification category of the strategic emerging industry classification based on the semantic similarity model and the similarity of the related information satisfying a preset condition, and obtaining the classification result according to the classification category; and / or, determining at least one frequent item set satisfying the classification requirement according to the classification system of the strategic emerging industry classification; and classifying the related information based on the determined frequent item set to obtain the classification result.

[0087] Here, the semantic similarity model includes, for example, BM25 and / TF-IDF, etc.

[0088] BM25 is usually used as a search relevance split. The main idea is to perform morpheme analysis on Query to generate morphemes q iThen, for each search result D, the relevance score of each morpheme q i with D, and finally, the relevance score of Query with D is obtained by weighted summing the relevance scores of q i with D.

[0089] TF-IDF (Term Frequency-Inverse Document Frequency) is a commonly used weighting technique for information retrieval and data mining. TF is Term Frequency, and IDF is Inverse Document Frequency. Term Frequency refers to the number of times a given word appears in a document. This number is usually normalized (typically by dividing by the total number of words in the document) to prevent it from favoring longer documents. The main idea of inverse document frequency is that if a document containing the term t is less, the IDF is larger, which means that the term has good class distinction ability. The IDF of a certain term can be obtained by dividing the total number of documents by the number of documents containing the term, and then taking the logarithm of the quotient. TF-IDF is equal to the product of TF and IDF.

[0090] The classification category of the strategic emerging industry classification that meets the preset condition in terms of similarity to the relevant information is determined based on a semantic similarity model. Specifically, the classification category of the strategic emerging industry classification with the highest similarity to the relevant information can be determined.

[0091] According to the classification system of the strategic emerging industry classification, at least one frequent item set that meets the classification requirement is determined. That is, by virtue of the classification rule, the relevant information is classified based on the characteristics that certain keyword combinations frequently appear in certain industries but do not frequently appear in other industries, and the classification result is obtained.

[0092] It can be understood that, by using the step of determining the classification category based on the semantic similarity model alone, or by using the step of classifying the relevant information by using the determined frequent item set alone, a part of the strategic emerging industry classification of the corresponding enterprises can be obtained. Of course, there is a possibility of missing identification by using a single step. By using the semantic similarity model and the frequent item set to determine the classification category, the identification probability can be improved.

[0093] As an optional specific implementation, the relevant information is classified based on the classification rule to obtain a classification result, specifically including:

[0094] determine a classification category of the strategic emerging industry classification that meets the preset condition based on the semantic similarity model, and obtain a third classification result according to the classification category; the third classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or third information, and the third information indicates that no classification category of the strategic emerging industry classification corresponding to the enterprise is obtained; in a case where the third classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, the third classification result is determined as the second classification result;

[0095] In a case where the third classification result includes the third information, at least one frequent item set meeting a classification requirement is determined according to the classification system of the strategic emerging industry classification; the related information is classified based on the determined frequent item set, and a fourth classification result is obtained; the fourth classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or fourth information, and the fourth information indicates that no classification category of the strategic emerging industry classification corresponding to the enterprise is obtained; in a case where the fourth classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, the fourth classification result is determined as the second classification result.

[0096] Here, the classification is first performed based on the semantic similarity model, and then the classification is performed in the manner of the frequent item set when no classification category of the strategic emerging industry classification corresponding to the enterprise is obtained. If the classification based on the semantic similarity model obtains the classification category of the strategic emerging industry classification corresponding to the enterprise, the relevant steps of the frequent item set rule classification are not performed, and the obtained classification result is directly determined as the second classification result. Conversely, at least one frequent item set meeting a classification requirement is determined according to the classification system of the strategic emerging industry classification; the related information is classified based on the determined frequent item set, and if a classification category can be obtained, the classification category obtained based on this step is determined as the second classification result.

[0097] For example, the corpus (related information of an enterprise to be classified) simultaneously includes “solar energy”, “photovoltaic”, and “power generation”, and it can be determined that the enterprise belongs to the solar energy industry, code 6.3. For another example, the corpus simultaneously includes “raw pharmaceuticals” and “drugs”, and it can be determined that the corresponding enterprise belongs to the biological medicine industry, code 4.1. Because the combination of the above-mentioned keywords does not commonly appear in other industries, or even does not appear, the frequent item set can be determined, and the related information is classified based on the frequent item set.

[0098] The frequent item set can include a combination of at least two keywords. When determining the frequent item set, whether the combination itself is valid can be determined by setting a threshold, and the set threshold is adjusted by the effect. If the threshold is set too low, the classification result may be inaccurate, noisy, etc. If the threshold is set too high, although the result is accurate, the covered samples are too few, and the classification effect is poor.

[0099] As an optional implementation, the classification category of the strategic emerging industry classification that meets the preset condition based on the semantic similarity model is determined, and a classification result is obtained according to the classification category, including:

[0100] The first classification category of the strategic emerging industry classification that meets the preset condition based on the first semantic similarity model is determined;

[0101] The second classification category of the strategic emerging industry classification that meets the preset condition based on the second semantic similarity model is determined;

[0102] The third classification result is determined according to the first classification category and the second classification category; the third classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or third information, and the third information represents that the classification category of the strategic emerging industry classification corresponding to the enterprise is not obtained;

[0103] Wherein, in the case that there is an intersection corresponding to the first classification category and the second classification category, the third classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, and the classification category of the strategic emerging industry classification corresponding to the enterprise is determined according to the intersection; in the case that there is no intersection corresponding to the first classification category and the second classification category, the third classification result includes the third information;

[0104] Wherein, in the case that the third classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, the third classification result is determined as the second classification result.

[0105] In addition, the embodiment does not exclude the possibility that the first classification category of the strategic emerging industry classification that meets the preset condition with the similarity of the related information is not obtained based on the first semantic similarity model, while the second classification category of the strategic emerging industry classification that meets the preset condition with the similarity of the related information is obtained based on the second semantic similarity model; or the first classification category of the strategic emerging industry classification that meets the preset condition with the similarity of the related information is obtained based on the first semantic similarity model, while the second classification category of the strategic emerging industry classification that meets the preset condition with the similarity of the related information is not obtained based on the second semantic similarity model. In other words, the classification category is obtained based on one of the first semantic similarity model and the second semantic similarity model, while the classification category is not obtained based on the other. Then, for these cases, in the embodiment, the third classification result includes the third information. In this way, when the results of the two semantic similarity models are inconsistent, the classification category that does not obtain the strategic emerging industry classification corresponding to the enterprise is selected as the third classification result, so as to ensure the accuracy of the recognition. Of course, the application also does not exclude the case that the classification category obtained based on one of the above semantic similarity models is determined as the third classification result, so that the accuracy of the recognition may be reduced, but the coverage of the recognition is improved.

[0106] In addition, the embodiment also does not exclude the possibility that the first classification category of the strategic emerging industry classification that meets the preset condition with the similarity of the related information is not obtained based on the first semantic similarity model, and the second classification category of the strategic emerging industry classification that meets the preset condition with the similarity of the related information is not obtained based on the second semantic similarity model. In other words, the classification category is not obtained based on the first semantic similarity model and the second semantic similarity model. Then, the third classification result includes the third information.

[0107] It can be understood that in the embodiment, at least two different semantic similarity models are used to determine the classification category of the strategic emerging industry classification that meets the preset condition with the similarity of the related information, and finally the intersection is taken, so as to improve the accuracy of the classification. It can be understood that the at least two different semantic similarity models can have different emphases, in other words, have respective advantages and respective disadvantages. The intersection of the respective results can avoid the classification error caused by the disadvantages of a single semantic similarity model as much as possible.

[0108] Of course, the present application does not exclude the case of determining the classification category of the strategic emerging industry classification based on a semantic similarity model. If the classification category of the strategic emerging industry classification that meets the preset condition in terms of the similarity to the relevant information can be determined based on the semantic similarity model, the third classification result includes the classification category; otherwise, if it is determined based on the semantic similarity model that there is no classification category of the strategic emerging industry classification that meets the preset condition in terms of the similarity to the relevant information, the third classification result includes the third information, i.e., no classification category of the strategic emerging industry classification corresponding to the enterprise is obtained.

[0109] The first semantic similarity model and the second semantic similarity model are two different semantic similarity models. For example, the first semantic similarity model is a BM25 semantic similarity model, and the second semantic similarity model is a TF-IDF semantic similarity model; of course, the present application is not limited thereto.

[0110] The similarity to the relevant information that meets the preset condition is specifically, for example, the highest similarity to the relevant information. In addition, the present application does not exclude the case of determining a similarity threshold, and the similarity to the relevant information that meets the preset condition is specifically the similarity to the relevant information that is higher than the similarity threshold.

[0111] Specifically, the classification category of the strategic emerging industry classification that meets the preset condition in terms of the similarity to the relevant information is a based on the first semantic similarity model, and a is taken as a prediction candidate classification category; the classification category of the strategic emerging industry classification that meets the preset condition in terms of the similarity to the relevant information is b based on the second semantic similarity model, and b is also taken as a prediction candidate classification category; the intersection of a and b is taken, and if there is an intersection, i.e., there is the same prediction result, the same prediction result is determined as the second classification result; otherwise, if there is no intersection, such as a and b are different, this step does not obtain the classification category of the strategic emerging industry classification corresponding to the enterprise.

[0112] As an optional specific embodiment, obtaining the classification result further includes:

[0113] The relevant information is classified based on a preset rule to obtain a fifth classification result; the fifth classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise or a fifth information, and the fifth information represents that no classification category of the strategic emerging industry classification corresponding to the enterprise is obtained;

[0114] Corresponding to the case that the fifth classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, the fifth classification result is output as the determined classification result.

[0115] Here, both the step of obtaining a classification result with a trained recognition model and / or obtaining a classification result based on a classification rule is performed, and the step of classifying the relevant information based on a human preset rule is performed. For a case where a classification category has been obtained through the step of obtaining a classification result with a trained recognition model and / or obtaining a classification result based on a classification rule, the classification category can be determined again through the step of classifying the relevant information based on a human preset rule, so as to improve the accuracy of the classification result.

[0116] The human preset rule is a rule summarized by a human being according to experience, which is converted into a language recognizable by a computer to form a preset rule.

[0117] For example, the enterprise name contains "insurance", "securities", and "credit", and the corpus (i.e., the relevant information of the enterprise to be classified) contains other services, but does not contain keywords of other industries such as intelligent machines and new energy, it is predicted that the enterprise belongs to other related services, code 9.2.

[0118] As an optional specific embodiment, the second classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or second information, and the second information represents that the classification category of the strategic emerging industry classification corresponding to the enterprise is not obtained.

[0119] Corresponding to a case where the fifth classification result includes the fifth information and the second classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise, the second classification result is output as a determined classification result.

[0120] Corresponding to a case where the fifth classification result includes the fifth information, the second classification result includes the second information, and the first classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise, the first classification result is output as a determined classification result.

[0121] Corresponding to a case where the fifth classification result includes the fifth information, the second classification result includes the second information, and the first classification result includes the first information, the classification result is that the enterprise does not belong to the strategic emerging industry.

[0122] It can be understood that the classification result obtained through the model, rule algorithm may have a small number of inaccurate cases, such as incorrect classification. By adjusting the model, a better effect cannot be obtained. Therefore, the human preset rule is used to strengthen the rule coverage in the embodiment. For a classification result determined through the human preset rule, the result is directly determined as the final output result. For a case where the classification result is determined through the previous steps, the classification result determined through the human preset rule is also used to cover the previous classification result, so as to make up for the deficiencies of the previous steps.

[0123] Specific example:

[0124] Figure 5 The flowchart of the strategic emerging industry classification method provided as a specific example; as shown in the figure, first, the enterprise name, business scope, and introduction are input into the Bert model as raw data for prediction. The results of the Bert model prediction can include "can be labeled as strategic emerging" (Y) and "cannot be labeled as strategic emerging" (N).

[0125] In specific applications, "can be labeled as strategic emerging" can refer to being able to be labeled with a secondary label; "cannot be labeled as strategic emerging" can refer to being unable to be labeled with a secondary label; wherein the secondary label corresponds to the second level of strategic emerging industry classification.

[0126] The part predicted by the Bert model as "cannot be labeled as strategic emerging" is further predicted using a rule-based model. The rule-based model can have 2 sub-models:

[0127] A. Key word rule model:

[0128] The BM25 semantic similarity model is used to weight the key words of the enterprise's related information, and then compared with the existing "Strategic Emerging Industry Classification (2018)" key words of each second-level industry to select the most similar second-level industry as the prediction candidate a.

[0129] The TF-IDF semantic similarity model is used to score the key words of the enterprise's related information again, and then compared with the existing "Strategic Emerging Industry Classification (2018)" key words of each second-level industry to select the second-level industry with the highest similarity as the prediction candidate b.

[0130] Next, the intersection of prediction candidate a and prediction candidate b is taken, and the same prediction value is taken as the second-level industry prediction value of this part of the enterprise.

[0131] B. Frequent item set rule model:

[0132] The key word combination that frequently appears in a certain industry but does not frequently appear in other industries is selected, and these key word combinations are used to label this part of the enterprise with the second-level label of strategic emerging industry classification.

[0133] Next, the companies predicted by the Bert model as "eligible for the strategic emerging industry label" are combined with those predicted by the rule-based model, and the remaining companies "ineligible for the strategic emerging industry label." Finally, a layer of strong rules is applied. These strong rules are manual rules, primarily specific keywords that have been manually spot-checked and verified. For example, companies that contain certain keywords but not others are directly predicted to belong to a certain industry. Through the predictions of all these models and rules, all companies belonging to strategic emerging industries are given the second-level label of the strategic emerging industry classification. Of course, companies that do not receive the second-level label are not considered to belong to strategic emerging industries.

[0134] As an optional specific implementation method, the relevant information includes: company name, business scope and introduction; the trained recognition model includes a trained Bert model; the trained Bert model includes three different Segment Embeddings corresponding to the company name, business scope and introduction respectively.

[0135] BERT (Bidirectional Encoder Representations from Transformers) is a pre-trained language model. It uses the Transformer, a popular feature extractor in this field, and also implements a bidirectional language model. This combination of advantages enables it to achieve better performance.

[0136] Traditional text classification methods (such as SVM and Bayesian machine learning methods) have limited effectiveness. Furthermore, shallow neural networks (such as Fasttext, LSTM, and TextCNN) require a large number of labeled training samples, making them difficult to apply to the classification of strategic emerging industries.

[0137] Of course, it should not be understood that the embodiments of the present application are limited to the use of the Bert model, and the embodiments of the present application do not exclude ELECTRA (Efficiently Learning an Encoder that Classifies Token Replacements Accurately, an encoder that effectively learns to accurately classify replaced tokens) and the like.

[0138] The trained BERT model includes three different Segment Embeddings, one for the company name, one for the business scope, and one for the introduction. For example, the Segment Embedding for the company name is 0, the one for the introduction is 1, and the one for the business scope is 2. By using different Segment Embeddings, the input company name, business scope, and introduction can be distinguished, allowing for weighted processing based on their importance, improving recognition accuracy.

[0139] As an optional specific implementation method, relevant information is input into a trained recognition model, and a classification result is obtained based on the trained recognition model, including: inputting relevant information into a trained recognition model, and after input embedding, obtaining a vector representation of each text block of the relevant information; processing the vector representation of each text block using an average pooling layer to obtain a processed vector; and performing classification based on the processed vector to obtain a classification result.

[0140] As can be understood, this embodiment uses an embedding layer to obtain a vector representation of each text block of relevant information; uses an average pooling layer to perform average pooling on the vector representation of each text block to obtain DocumentEmbedding, which serves as the vector representation of the company text; and performs classification based on DocumentEmbedding, making full use of the information of the entire sequence and achieving more accurate results.

[0141] As an optional specific implementation manner, the trained recognition model includes a trained Bert model; the pre-training process of the trained Bert model includes: domain adaptive pre-training and / or task adaptive pre-training.

[0142] Furthermore, the pre-training process of the trained BERT model includes task-adaptive pre-training; task-adaptive pre-training is performed based on the label information of the national economic industry classification.

[0143] In addition, the pre-training process of the trained BERT model can also include domain-adaptive pre-training. Specifically, pre-training is performed using the company name, business scope, and profile, making the pre-trained model more domain-specific.

[0144] Figure 6The framework structure diagram of the Bert model provided for a specific example; as shown in the figure, the improved points of the Bert model are as follows compared with the general Bert model:

[0145] In the first aspect, the model optimization algorithm based on the Bert framework in the specific example is used to splice the enterprise name (company name), business scope and brief introduction into a piece of text through [SEP], and the different importance of the three fields is considered by using different Segment Embedding to distinguish them, and then the Bert is used to extract relevant features for strategic emerging industry classification. That is, the Bert model used in the specific example distinguishes different sentence inputs through [SEP] and Segment Embedding, uses the same Transformer layers, and finally uses the Document Embedding as the representation of the company text.

[0146] And the usual way is to use the Embedding corresponding to CLS for text classification, and the specific example is based on the output sequence of Bert for pooling, which fully utilizes the information of the entire sequence.

[0147] In the second aspect, the domain adaptive pre-training is tried. The pre-training is tried by using the million-level enterprise name, business scope and brief introduction, that is, the pre-training model is further pre-trained in the large-scale unlabeled corpus in the domain, so that the pre-training model is more domain-oriented.

[0148] In the third aspect, the task adaptive pre-training is tried. Since the existing enterprises have the label information of major categories, large categories, medium categories and small categories according to the "National Economic Industry Classification", the Bert model is further pre-trained by using the million-level labeled data, and then applied to the downstream task-strategic emerging industry classification, which contains the idea of transfer learning.

[0149] In the fourth aspect, the irrelevant information is filtered by keywords. The keywords of each industry are extracted according to the strategic emerging industry classification, and the irrelevant information in the enterprise name, business scope and brief introduction is filtered. In addition, according to the fine-grained word nature, the hanlp is used to filter place names, position words, conjunctions, agency suffixes, onomatopoeic words and prepositions.

[0150] For example, in the input part of the Bert model, for the obtained relevant information of the enterprise to be classified, on the one hand, the irrelevant information such as "the business scope of this company is…" and other general template contents is removed; on the other hand, the virtual words are removed; so as to form corpus with higher value.

[0151] Correspondingly, the related information of the enterprise to be classified is acquired; and the classification result of the strategic emerging industry classification of the enterprise is obtained according to the related information, specifically including: acquiring the related information of the enterprise to be classified; performing filtering processing on the related information, and obtaining the classification result of the strategic emerging industry classification of the enterprise according to the related information after the filtering processing.

[0152] The embodiment of the present application further provides a strategic emerging industry classification device, please refer to Figure 7 The strategic emerging industry classification device 700 includes:

[0153] The acquisition module 701 is configured to acquire the related information of the enterprise to be classified;

[0154] The classification module 702 is configured to obtain the classification result of the strategic emerging industry classification of the enterprise according to the related information; wherein the obtaining of the classification result includes: inputting the related information into the trained recognition model, and obtaining the classification result based on the trained recognition model; and / or classifying the related information based on the classification rule to obtain the classification result; the classification rule is determined according to the classification system of the strategic emerging industry classification.

[0155] Optionally, the classification module 702 is configured to input the related information into the trained recognition model, and obtain a first classification result based on the trained recognition model; the first classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or first information, and the first information represents that the classification category of the strategic emerging industry classification corresponding to the enterprise is not obtained; corresponding to the case that the first classification result includes the first information, the related information is classified based on the classification rule to obtain a second classification result.

[0156] Optionally, the classification module 702 is configured to determine the classification category of the strategic emerging industry classification which satisfies the preset condition in terms of the similarity of the related information based on a semantic similarity model, and obtain the classification result according to the classification category; and / or determine at least one frequent item set satisfying the classification requirement according to the classification system of the strategic emerging industry classification; and classify the related information based on the determined frequent item set to obtain the classification result.

[0157] Optionally, the classification module 702 is configured to determine a classification category of the strategic emerging industry classification that meets the preset condition based on the semantic similarity model, and obtain a third classification result according to the classification category; the third classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or third information, and the third information represents that the classification category of the strategic emerging industry classification corresponding to the enterprise is not obtained; in a case where the third classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, the third classification result is determined as the second classification result; in a case where the third classification result includes the third information, at least one frequent item set that meets the classification requirement is determined according to the classification system of the strategic emerging industry classification; the related information is classified based on the determined frequent item set, and a fourth classification result is obtained; the fourth classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise or fourth information, and the fourth information represents that the classification category of the strategic emerging industry classification corresponding to the enterprise is not obtained; in a case where the fourth classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, the fourth classification result is determined as the second classification result.

[0158] Optionally, the classification module 702 is configured to determine a first classification category of the strategic emerging industry classification that meets the preset condition based on the first semantic similarity model, and determine a second classification category of the strategic emerging industry classification that meets the preset condition based on the second semantic similarity model; the third classification result is determined according to the first classification category and the second classification category; the third classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or third information, and the third information represents that the classification category of the strategic emerging industry classification corresponding to the enterprise is not obtained; in a case where there is an intersection between the first classification category and the second classification category, the third classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, and the classification category of the strategic emerging industry classification is determined according to the intersection; in a case where there is no intersection between the first classification category and the second classification category, the third classification result includes the third information; in a case where the third classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, the third classification result is determined as the second classification result.

[0159] Optionally, the classification module 702 is configured to classify the related information based on the artificial preset rule, and obtain a fifth classification result; the fifth classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or fifth information, and the fifth information represents that the classification category of the strategic emerging industry classification corresponding to the enterprise is not obtained; in a case where the fifth classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, the fifth classification result is output as the determined classification result.

[0160] Optionally, the second classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or second information, and the second information indicates that the enterprise does not obtain the classification category of the strategic emerging industry classification corresponding to the enterprise. The classification module 702 is configured to, in a case where the fifth classification result includes the fifth information and the second classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, output the second classification result as the determined classification result; in a case where the fifth classification result includes the fifth information, the second classification result includes the second information, and the first classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, output the first classification result as the determined classification result; and in a case where the fifth classification result includes the fifth information, the second classification result includes the second information, and the first classification result includes the first information, the classification result is that the enterprise does not belong to the strategic emerging industry.

[0161] Optionally, the relevant information includes at least one of the following: an enterprise name, a business scope, and a brief introduction.

[0162] Optionally, the relevant information includes the enterprise name, the business scope, and the brief introduction; the trained recognition model includes a trained Bert model; and the trained Bert model includes three different Segment Embeddings corresponding to the enterprise name, the business scope, and the brief introduction, respectively.

[0163] The classification module 702 is configured to input the relevant information into the trained recognition model, embed the relevant information after the input, obtain vector representations of each text block of the relevant information, process the vector representations of each text block using an average pooling layer to obtain a processed vector, and perform classification according to the processed vector to obtain a classification result.

[0164] Optionally, the trained recognition model includes a trained Bert model; and a pre-training process of the trained Bert model includes domain self-adaptive pre-training and / or task self-adaptive pre-training.

[0165] Optionally, the pre-training process of the trained Bert model includes task self-adaptive pre-training; and the task self-adaptive pre-training is based on label information of the national economic industry classification.

[0166] The embodiments of the present application also provide a computer readable storage medium. The computer readable storage medium stores instructions, and when the instructions are executed by a processor of an electronic device, the electronic device can perform the steps in the strategic emerging industry classification method according to any of the above embodiments.

[0167] The embodiments of method, apparatus and computer program product described herein can be part of a system, method and / or computer program product. The computer program product can include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present application. The computer program product can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++, or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server. In the latter scenario, the remote computing device can be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computing device, such as through the Internet using an Internet Service Provider. In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) can execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present application.

[0168] The computer readable storage medium can be a combination of one or more computer readable media. The computer readable storage medium can be a combination of one or more of a readable signal medium and a readable storage medium. A readable storage medium is a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium can include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium include the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or punched tape, a

[0169] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0170] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0171] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0172] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0173] Embodiments of the present application also provide an electronic device. Figure 8 As shown in the figure, the electronic device 800 includes one or more processors 801 and a memory 802; the memory 802 stores computer executable instructions; and the processor 801 executes the computer executable instructions to implement the steps in the strategic emerging industry classification method according to any of the above embodiments.

[0174] The processor 801 can be a central processing unit (CPU) or other form of processing unit that has data processing and / or instruction execution capabilities, and can control other components in the electronic device to perform desired functions.

[0175] The memory 802 can include one or more computer program products that can include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory, for example, can include random access memory (RAM), cache memory, and / or the like. The non-volatile memory, for example, can include read-only memory (ROM), hard disk, flash memory, and / or the like. One or more computer program instructions can be stored on the computer-readable storage media, and the processor 1501 can run the program instructions to implement the steps in the text recognition method of the various embodiments of the present application described above and / or other desired functions.

[0176] In one example, the electronic device 800 can further include input and output devices, which are interconnected through a bus system and / or other forms of connection mechanisms (not shown in the figure).

[0177] In addition, the input device can further include, for example, a keyboard, a mouse, a microphone, and / or the like. The output device can output various information to the outside, and can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and / or the like.

[0178] Of course, in order to simplify, Figure 8 Only a part of the components in the electronic device 800 related to the present application is shown in the figure, and components such as buses, input / output interfaces, and the like are omitted. In addition to this, the electronic device 800 can further include any other appropriate components according to specific application cases.

[0179] It should be noted that the strategic emerging industry classification method embodiment, the strategic emerging industry classification device embodiment, the computer readable storage medium embodiment, and the electronic device embodiment provided by the embodiments of the present application belong to the same concept; the technical features in the technical solutions recorded by each embodiment can be combined arbitrarily without conflict.

[0180] It should be understood that the above embodiments are exemplary and are not intended to include all possible implementations of the claims. Various modifications and changes can also be made to the above embodiments without departing from the scope of the present disclosure. Similarly, any combination of the technical features of the above embodiments can be made to form additional embodiments of the present application that have not been explicitly described. Therefore, the above embodiments only express several implementations of the present application, and do not limit the protection scope of the present application.

Claims

1. A classification method for strategic emerging industries, characterized by: The method comprises: Obtain relevant information about the enterprise to be classified; Obtaining, based on the relevant information, a classification result of the strategic emerging industries classification of the enterprise; wherein obtaining the classification result includes: Inputting the relevant information into a trained recognition model, and obtaining a first classification result based on the trained recognition model; the first classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or first information, wherein the first information indicates that a classification category of the strategic emerging industry classification corresponding to the enterprise has not been obtained; Corresponding to the case where the first classification result includes the first information, the relevant information is classified based on the classification rules to obtain a second classification result; the classification rules are determined according to the classification system of the strategic emerging industries classification; classifying the relevant information based on the classification rules includes: determining at least one frequent item set that meets the classification requirements according to the classification system of the strategic emerging industries classification; classifying the relevant information based on the determined frequent item set, and if a classification category can be obtained, determining the obtained classification category as the second classification result.

2. The strategic emerging industries classification method according to claim 1 is characterized in that: The classifying the relevant information based on the classification rules includes: Based on the semantic similarity model, a classification category of the strategic emerging industry classification whose similarity with the relevant information meets preset conditions is determined, and classification is performed according to the classification category.

3. The strategic emerging industries classification method according to claim 2, characterized in that: Classifying the relevant information based on the classification rules to obtain the classification result specifically includes: Determining, based on a semantic similarity model, a classification category of the strategic emerging industry classification whose similarity to the relevant information meets a preset condition, and obtaining a third classification result according to the classification category; the third classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise or third information, the third information indicating that the classification category of the strategic emerging industry classification corresponding to the enterprise has not been obtained; if the third classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, determining the third classification result as the second classification result; Corresponding to the case where the third classification result includes the third information, at least one frequent item set that meets the classification requirements is determined according to the classification system of the strategic emerging industries classification; the relevant information is classified based on the determined frequent item set to obtain a fourth classification result; the fourth classification result includes the classification category of the strategic emerging industries classification corresponding to the enterprise or the fourth information, and the fourth information represents that the classification category of the strategic emerging industries classification corresponding to the enterprise has not been obtained; corresponding to the case where the fourth classification result includes the classification category of the strategic emerging industries classification corresponding to the enterprise, the fourth classification result is determined as the second classification result.

4. The strategic emerging industries classification method according to claim 2, characterized in that: The determining, based on the semantic similarity model, a classification category of the strategic emerging industries classification whose similarity to the relevant information meets a preset condition, and obtaining the classification result according to the classification category includes: Determine, based on the first semantic similarity model, a first classification category of a strategic emerging industry classification whose similarity to the relevant information meets a preset condition; Determine, based on the second semantic similarity model, a second classification category of the strategic emerging industries classification whose similarity to the relevant information meets a preset condition; Determining a third classification result based on the first classification category and the second classification category; the third classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or third information, wherein the third information indicates that a classification category of the strategic emerging industry classification corresponding to the enterprise has not been obtained; Wherein, corresponding to the case where the first classification category and the second classification category have an intersection, the third classification result includes the classification category of the strategic emerging industries classification corresponding to the enterprise, and the classification category of the strategic emerging industries classification is determined according to the intersection; corresponding to the case where the first classification category and the second classification category do not have an intersection, the third classification result includes the third information; Wherein, corresponding to the case where the third classification result includes the classification category of the strategic emerging industry classification corresponding to the enterprise, the third classification result is determined as the second classification result.

5. The strategic emerging industries classification method according to claim 1 is characterized in that: The obtaining of the classification result further comprises: Classifying the relevant information based on preset rules to obtain a fifth classification result; the fifth classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or fifth information, wherein the fifth information indicates that a classification category of the strategic emerging industry classification corresponding to the enterprise has not been obtained; In response to the situation where the fifth classification result includes a classification category of the strategic emerging industries classification corresponding to the enterprise, the fifth classification result is output as the determined classification result.

6. The strategic emerging industries classification method according to claim 5 is characterized in that: The second classification result includes a classification category of the strategic emerging industries classification corresponding to the enterprise or second information, wherein the second information indicates that the classification category of the strategic emerging industries classification corresponding to the enterprise has not been obtained; corresponding to a case where the fifth classification result includes the fifth information and the second classification result includes a classification category of the strategic emerging industries classification corresponding to the enterprise, outputting the second classification result as the determined classification result; corresponding to a case where the fifth classification result includes the fifth information, the second classification result includes the second information, and the first classification result includes a classification category of the strategic emerging industries classification corresponding to the enterprise, outputting the first classification result as the determined classification result; Corresponding to the situation where the fifth classification result includes the fifth information, the second classification result includes the second information, and the first classification result includes the first information, the classification result is that the enterprise does not belong to a strategic emerging industry.

7. The strategic emerging industries classification method according to claim 1 is characterized in that: The relevant information includes at least one of the following: company name, business scope, and introduction.

8. The strategic emerging industries classification method according to claim 1, characterized in that: The relevant information includes: company name, business scope and introduction; the trained recognition model includes a trained Bert model; the trained Bert model includes three different Segment Embeddings corresponding to the company name, business scope and introduction respectively.

9. The strategic emerging industries classification method according to claim 1, characterized in that: Inputting the relevant information into a trained recognition model and obtaining the classification result based on the trained recognition model includes: Inputting the relevant information into a trained recognition model, and embedding the input to obtain a vector representation of each text block of the relevant information; The vector representation of each text block is processed using an average pooling layer to obtain a processed vector; and classification is performed according to the processed vector to obtain the classification result.

10. The strategic emerging industries classification method according to claim 1, characterized in that: The trained recognition model includes a trained Bert model; the pre-training process of the trained Bert model includes: domain adaptive pre-training and / or task adaptive pre-training.

11. The strategic emerging industries classification method according to claim 10, characterized in that: The pre-training process of the trained BERT model includes task-adaptive pre-training; the task-adaptive pre-training is performed based on label information of the national economic industry classification.

12. A strategic emerging industry classification device, characterized in that: include: an acquisition module configured to acquire relevant information of the enterprise to be classified; A classification module is configured to obtain a classification result of the strategic emerging industries classification of the enterprise based on the relevant information; wherein obtaining the classification result includes: Inputting the relevant information into a trained recognition model, and obtaining a first classification result based on the trained recognition model; the first classification result includes a classification category of the strategic emerging industry classification corresponding to the enterprise or first information, wherein the first information indicates that a classification category of the strategic emerging industry classification corresponding to the enterprise has not been obtained; Corresponding to the case where the first classification result includes the first information, the relevant information is classified based on the classification rules to obtain a second classification result; the classification rules are determined according to the classification system of the strategic emerging industries classification; classifying the relevant information based on the classification rules includes: determining at least one frequent item set that meets the classification requirements according to the classification system of the strategic emerging industries classification; classifying the relevant information based on the determined frequent item set, and if a classification category can be obtained, determining the obtained classification category as the second classification result.

13. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, which, when executed by a processor of an electronic device, enable the electronic device to execute the strategic emerging industries classification method described in any one of claims 1 to 11.

14. An electronic device, characterized in that: The electronic device comprises: processor; memory for storing computer-executable instructions; The processor is used to execute the computer-executable instructions to implement the strategic emerging industries classification method described in any one of claims 1 to 11 above.

Images