An enterprise identification method and device based on an enterprise knowledge graph

Abstract

The application discloses an enterprise identification method and device based on an enterprise knowledge graph, and the method comprises the following steps: acquiring Chinese keywords in public opinion information of each enterprise entity; performing semantic splitting on the keywords according to pictographic elements, and outputting word vectors of the enterprise entities according to the semantic splitting results; identifying the word vectors through a semantic correlation degree interpretation model to obtain public opinion identification results of the enterprise entities; constructing an enterprise knowledge graph according to the public opinion identification results of the enterprise entities and economic relations between the enterprise entities; and determining the levels of the enterprise entities according to the enterprise knowledge graph. The application can extract enterprise entity word vectors from public opinion information of small and micro enterprises, combine the economic relations between the small and micro enterprises, mine more accurate conditions of the small and micro enterprises and relations between the small and micro enterprises, and construct an accurate enterprise knowledge graph, so that the small and micro enterprises can be protected, and the data and property safety of the application can be ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer information processing, in particular to an enterprise identification method and device based on an enterprise knowledge graph, an electronic device and a computer readable medium. BACKGROUND

[0002] According to statistics, there are more than 40 million small and micro enterprises in China, and the proportion of informationization is less than 10%. Due to the lack of informationization support, the management of most small and micro enterprises is in a state of extensive and chaotic, which makes them at a disadvantage in market competition and easily bankrupted by large and medium-sized enterprises. Because of this, small and micro enterprises often need financing to maintain operation due to various business or social environment problems.

[0003] In actual work process, for the support of small and micro enterprises, not only the self situation of small and micro enterprises needs to be accurately determined, but also the business relationship between small and micro enterprises needs to be determined. The visible account data is often unreliable and easy to be fake, and the customer group of small and micro enterprises is not stable. The transaction between small and micro enterprises or the external business situation is often very complex. Therefore, the existing technology is not accurate enough to describe the enterprises and the relationship between enterprises. If the enterprise with data security and property security problems is identified as a safe white list enterprise in the identification process, it will cause great hidden troubles to the data security and property security of the enterprise itself. SUMMARY

[0004] Therefore, the present application mainly aims to provide an enterprise identification method and device based on an enterprise knowledge graph, an electronic device and a computer readable medium, so as to at least partially solve at least one of the above technical problems, such as the difficulty in determining the enterprise state of small and micro enterprises.

[0005] In order to solve the above technical problems, the first aspect of the present application provides an enterprise identification method based on an enterprise knowledge graph, which comprises:

[0006] Obtaining public opinion information of each enterprise entity and obtaining Chinese keywords in the public opinion information;

[0007] According to the semantic splitting of the pictographic elements, the keywords are respectively split, and the word vector of the enterprise entity is output according to the semantic splitting result;

[0008] The word vector is identified through a semantic correlation degree explanation model to obtain an opinion recognition result of the enterprise entity, the semantic correlation degree explanation model is used to identify semantic correlation degrees of the keyword corresponding to the word vector and each word or Chinese character in text training data, and a word or Chinese character with a semantic correlation degree meeting a preset condition is used as an opinion recognition result to explain and describe the keyword in a semantic layer;

[0009] Economic information of each enterprise entity is acquired, and economic relationships between the enterprise entities are determined according to the economic information;

[0010] An enterprise knowledge graph is constructed according to the opinion recognition results of each enterprise entity and the economic relationships between the enterprise entities;

[0011] The enterprise entities are classified according to the enterprise knowledge graph to determine the levels of each enterprise entity.

[0012] According to a preferred embodiment of the present application, the semantic splitting of the keyword according to the pictographic elements includes:

[0013] The keyword is converted into a traditional Chinese character;

[0014] The traditional Chinese character is split and mapped according to the pictographic elements to obtain a semantic splitting result.

[0015] According to a preferred embodiment of the present application, the pictographic elements include Chinese five-stroke radicals, and the split and mapping of the traditional Chinese character according to the pictographic elements includes:

[0016] The traditional Chinese character is split and mapped through the Chinese five-stroke radicals to obtain a plurality of English mapping units, each English mapping unit corresponds to a pictographic element and represents a semantic unit;

[0017] The plurality of English mapping units are combined to obtain a plurality of combined features representing different semantics of the keyword.

[0018] According to a preferred embodiment of the present application, the pictographic elements include minimum segmentation elements of the traditional Chinese character, and the split and mapping of the traditional Chinese character according to the pictographic elements includes:

[0019] The traditional Chinese character is split according to the pictographic elements;

[0020] The split result is mapped to obtain a plurality of mapping units, each mapping unit corresponds to a pictographic element and represents a semantic unit;

[0021] The plurality of mapping units are combined to obtain a plurality of combined features representing different semantics of the keyword.

[0022] According to a preferred embodiment of the present application, the outputting of the word vector of the enterprise entity according to the semantic segmentation result comprises:

[0023] The semantic segmentation result is one-hot encoded to obtain the word vector of the enterprise entity.

[0024] According to a preferred embodiment of the present application, the constructing of the enterprise knowledge graph according to the public opinion recognition result of each enterprise entity and the economic relationship between the enterprise entities comprises:

[0025] The node corresponding to each enterprise entity is constructed, and the public opinion recognition result of the enterprise entity and the economic relationship between the enterprise entities are taken as attribute information of the node.

[0026] The connection edge between the nodes is determined according to the economic relationship between the enterprise entities.

[0027] The enterprise knowledge graph is constructed based on the connection edge and the node.

[0028] According to a preferred embodiment of the present application, the economic relationship comprises transaction information and economic association relationship between the enterprise entities.

[0029] Taking the public opinion recognition result of the enterprise entity and the economic relationship between the enterprise entities as the attribute information of the node comprises:

[0030] The transaction information between the enterprise entities is adjusted through the economic association relationship, so that the value of the transaction information between the enterprise entities with the preset relationship in the economic association relationship is reduced.

[0031] The public opinion recognition result of the enterprise entity and the transaction information between the enterprise entities are taken as the attribute information of the node.

[0032] The connection edge between the nodes is determined according to the economic relationship between the enterprise entities, comprising:

[0033] The connection edge between the nodes is determined according to the transaction information between the enterprise entities.

[0034] According to a preferred embodiment of the present application, the economic association relationship comprises at least one of investment relationship, debt relationship, guarantee relationship and upstream / downstream relationship.

[0035] The transaction information of each node is adjusted through the economic association relationship, comprising:

[0036] The value y of the transaction information between the enterprise entities after adjustment i is determined by the following formula:

[0037]

[0038] wherein: x i is the original value of the transaction information between enterprise entities, a i is the percentage of guarantee or the percentage of equity, investment relationship=true indicates that the enterprise entities include investment relationship, upstream and downstream relationship=true indicates that the enterprise entities include upstream and downstream relationship, debt relationship=true indicates that the enterprise entities include debt relationship, and guarantee relationship=true indicates that the enterprise entities include guarantee relationship.

[0039] To solve the above technical problems, the second aspect of the present application provides an enterprise identification device based on an enterprise knowledge graph, which comprises:

[0040] A first acquisition module is configured to acquire public opinion information of each enterprise entity and acquire Chinese keywords in the public opinion information.

[0041] A word vector processing module is configured to perform semantic splitting on the keywords respectively according to pictographic elements, and output word vectors of the enterprise entities according to the semantic splitting results.

[0042] An identification module is configured to identify the word vectors through a semantic correlation degree interpretation model to obtain public opinion identification results of the enterprise entities, wherein the semantic correlation degree interpretation model is used to identify semantic correlation degrees between the keywords corresponding to the word vectors and each word or Chinese character in text training data, and words or Chinese characters with semantic correlation degrees meeting preset conditions are used as public opinion identification results to explain and describe the keywords at a semantic level.

[0043] A second acquisition module is configured to acquire economic information of each enterprise entity and determine economic relationships between the enterprise entities according to the economic information.

[0044] A construction module is configured to construct an enterprise knowledge graph according to public opinion identification results of each enterprise entity and economic relationships between the enterprise entities.

[0045] A determination module is configured to grade the enterprise entities according to the enterprise knowledge graph and determine levels of each of the enterprise entities.

[0046] According to a preferred embodiment of the present application, the word vector processing module comprises:

[0047] A conversion module is configured to convert the keywords into traditional Chinese characters.

[0048] A splitting and mapping module is configured to perform splitting and mapping processing on the traditional Chinese characters according to pictographic elements to obtain semantic splitting results.

[0049] According to a preferred embodiment of the present application, the pictographic elements comprise Chinese five-stroke radicals, and the splitting and mapping module comprises:

[0050] A first splitting and mapping module is configured to split and map the traditional Chinese characters into Chinese five-stroke radicals to obtain a plurality of English mapping units, each of which corresponds to a pictographic element and represents a semantic unit.

[0051] A first combination module is configured to combine the plurality of English mapping units to obtain a plurality of combined features representing different semantics of the keyword.

[0052] According to a preferred embodiment of the present application, the pictographic elements comprise the smallest segmentation elements of the traditional Chinese characters, and the splitting and mapping module comprises:

[0053] A splitting module is configured to split the traditional Chinese characters according to the pictographic elements.

[0054] A mapping module is configured to map the splitting result to obtain a plurality of mapping units, each of which corresponds to a pictographic element and represents a semantic unit.

[0055] A second combination module is configured to combine the plurality of mapping units to obtain a plurality of combined features representing different semantics of the keyword.

[0056] According to a preferred embodiment of the present application, the word vector processing module further comprises:

[0057] An encoding module is configured to perform one-hot encoding on the semantic splitting result to obtain the word vector of the enterprise entity.

[0058] According to a preferred embodiment of the present application, the construction module is specifically configured to construct nodes corresponding to each of the enterprise entities, and take the public opinion recognition result of the enterprise entity and the economic relationship between the enterprise entities as attribute information of the nodes; determine connection edges between the nodes according to the economic relationship between the enterprise entities; and construct the enterprise knowledge graph based on the connection edges and the nodes.

[0059] According to a preferred embodiment of the present application, the economic relationship comprises transaction information and economic association relationship between the enterprise entities.

[0060] The construction module is specifically configured to adjust the transaction information between the enterprise entities through the economic association relationship, so that the value of the transaction information between the enterprise entities in the economic association relationship that comprises a preset relationship is reduced; and take the public opinion recognition result of the enterprise entity and the transaction information between the enterprise entities as attribute information of the nodes.

[0061] The construction module is further configured to determine the connection edges between the nodes according to transaction information between the enterprise entities.

[0062] According to a preferred embodiment of the present application, the economic relationship includes at least one of an investment relationship, a debt relationship, a guarantee relationship, and an upstream / downstream relationship.

[0063] The construction module determines the adjusted value y of the transaction information between the enterprise entities by the following formula i :

[0064]

[0065] wherein x i is an original value of the transaction information between the enterprise entities, a i is a guarantee percentage or a stock percentage, investment relationship=true indicates that the investment relationship is included between the enterprise entities, upstream / downstream relationship=true indicates that the upstream / downstream relationship is included between the enterprise entities, debt relationship=true indicates that the debt relationship is included between the enterprise entities, and guarantee relationship=true indicates that the guarantee relationship is included between the enterprise entities.

[0066] To solve the above technical problems, the third aspect of the present application provides an electronic device, comprising:

[0067] a processor; and

[0068] a memory storing computer executable instructions that, when executed, cause the processor to perform the above method.

[0069] To solve the above technical problems, the fourth aspect of the present application provides a computer readable storage medium, wherein the computer readable storage medium stores one or more programs, when the one or more programs are executed by a processor, the above method is implemented.

[0070] This invention takes into account the pictographic characteristics of Chinese characters. It semantically segments keywords in the public opinion information of various enterprise entities based on pictographic elements, ensuring that each semantic segmentation result accurately explains the different meanings of the keyword. Then, it processes each semantic segmentation result into word vectors, outputting the corresponding word vectors. A semantic correlation degree interpretation model identifies the semantic correlation between the keyword and each word or Chinese character in the text training data. Words or Chinese characters whose semantic correlation degree meets preset conditions are used as public opinion recognition results to provide semantic explanations for the keyword. Therefore, this invention can extract enterprise entity word vectors from the public opinion information of micro and small enterprises, and combine this with the economic relationships between micro and small enterprises to uncover more accurate information about micro and small enterprises and the relationships between them, constructing an accurate enterprise knowledge graph. This provides support for serving micro and small enterprises and protects their data and property security. Attached Figure Description

[0071] To make the technical problems solved by this invention, the technical means adopted, and the technical effects achieved clearer, specific embodiments of this invention will be described in detail below with reference to the accompanying drawings. However, it should be noted that the drawings described below are merely drawings of exemplary embodiments of this invention. Those skilled in the art can obtain drawings of other embodiments based on these drawings without any creative effort.

[0072] Figure 1 This is a flowchart illustrating an enterprise identification method based on an enterprise knowledge graph according to an embodiment of the present invention;

[0073] Figure 2a This is a schematic diagram of a process for splitting and mapping traditional Chinese characters based on pictographic elements according to an embodiment of the present invention;

[0074] Figure 2b This is a schematic diagram of another embodiment of the present invention for splitting and mapping traditional Chinese characters based on pictographic elements;

[0075] Figure 3 This is a schematic diagram of the structural framework of an enterprise identification device based on an enterprise knowledge graph according to an embodiment of the present invention;

[0076] Figure 4 This is a structural block diagram of an exemplary embodiment of an electronic device according to the present invention;

[0077] Figure 5 This is a schematic diagram of an embodiment of a computer-readable medium according to the present invention. Detailed Implementation

[0078] Exemplary embodiments of the present application will now be described more fully with reference to the accompanying drawings, in which exemplary embodiments can be embodied in various forms. The embodiments described are not intended to be exhaustive or to be limited to the precise form disclosed. Rather, the embodiments are intended to be given on an illustrative basis and to provide insight into the scope of the present application, which should be limited only by the attached claims.

[0079] Structures, properties, effects, or other features described in connection with a certain exemplary embodiment can be combined with one or more other exemplary embodiments in any suitable manner.

[0080] In the course of introducing specific embodiments, the detailed description of structures, properties, effects, or other features is intended to enable a full understanding of the embodiments by those skilled in the art. However, it is not excluded that those skilled in the art can implement the present application in a specific case without the above-mentioned structures, properties, effects, or other features.

[0081] The flowcharts in the drawings are only exemplary flow demonstrations, and do not necessarily mean that all contents, operations, and steps in the flowcharts must be included in the solutions of the present application, nor does it mean that the execution must be performed in the order shown in the drawings. For example, some operations / steps in the flowcharts can be divided, some operations / steps can be combined or partially combined, etc. The execution order shown in the flowcharts can be changed according to actual conditions without departing from the essential spirit of the present application.

[0082] The blocks in the drawings Figure 1 generally represent functional entities, which do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0083] The same reference numbers in different drawings represent the same or similar elements, components, or parts, and thus the repeated description of the same or similar elements, components, or parts can be omitted hereinafter. It should also be understood that although the first, second, third, etc. representative numerals are used to describe various devices, elements, components, or parts, these devices, elements, components, or parts should not be limited by these representative numerals. That is, these representative numerals are only used to distinguish one from another. For example, a first device can also be referred to as a second device without departing from the essential technical solutions of the present application. In addition, the terms "and / or", "and / or" mean all combinations of one or more of the listed items.

[0084] Please refer to Figure 1 , Figure 1The application provides an enterprise identification method based on an enterprise knowledge graph, as shown in the accompanying drawings. Figure 2a The method comprises the following steps.

[0085] S1, obtaining public opinion information of each enterprise entity and obtaining Chinese keywords in the public opinion information.

[0086] In the embodiment of the application, the public opinion information of the enterprise entity can be extracted from text data related to enterprise macro-environment analysis, micro-environment analysis and / or enterprise public opinion analysis, and the Chinese keywords can be description text obtained by analyzing the public opinion information, which can be Chinese characters or words.

[0087] In the embodiment of the application, the Internet is a collection place of public opinion, and the information sources of the Internet are very extensive, including information of various regions, various fields and even countries around the world. Internet information, especially network media information, has no too many turnover links and layer-by-layer approval procedures, and the problems are exposed more sharply and timely, especially the information published by emerging self-media, which is updated every second in front of us. We can find many valuable public opinion points from it. The public opinion information can be related public opinion news of the enterprise, such as negative news or negative reports related to the enterprise; the public opinion information can also be the prompt content in the public financial statements of the enterprise or the risk items determined by the financial institutions.

[0088] For example, the keywords can be single Chinese characters or words obtained by describing the enterprise through macro-environment analysis and / or micro-environment analysis and / or enterprise public opinion analysis; the obtained keywords can be used as node attributes of the enterprise entity to explain and describe the enterprise entity, so as to accurately depict the enterprise portrait. The macro-environment analysis can include industry analysis, industry analysis, supply chain analysis, etc.; the micro-environment analysis can include financial analysis, business analysis, customer analysis, etc. in a predetermined time period.

[0089] In the embodiment of the application, the keywords in the public opinion information can be obtained by text cleaning of the public opinion information, word segmentation of the public opinion information, removal of stop words and auxiliary words, etc., and then determining the keywords in the public opinion information, and the keywords in the public opinion information can also be determined by counting the word frequency of semantically similar words in the cleaned data, and the keywords in the public opinion information can also be manually marked or determined by big data processing.

[0090] S2, according to the semantic splitting of the pictographic elements, the keywords are respectively split, and the word vectors of the enterprise entity are output according to the semantic splitting results.

[0091] According to the characteristics of the Chinese pictographic characters, the key word is semantically split according to a pictographic element, and a word vector of the enterprise entity is output according to a semantic splitting result; the pictographic element can be a minimum structure of Chinese characters representing complete semantics, in an example, considering that each Wubi root can represent complete semantics, the Wubi root can be used as the pictographic element. In another example, considering that the minimum segmentation elements of the traditional Chinese characters can also represent complete semantics, the minimum segmentation elements of the traditional Chinese characters can be used as the pictographic element. In addition, a Chinese character structure can also be specified as the pictographic element according to semantics, for example, the Chinese character "zhi" is used as a pictographic element, representing knowledge, and the like.

[0092] In the embodiment of the application, the key word can be directly semantically split according to the pictographic element. Considering that the traditional Chinese characters can contain more pictographic elements and have more semantic information than the simplified Chinese characters, the key word is preferably converted into the traditional Chinese characters, and then the traditional Chinese characters are split and mapped according to the pictographic element to obtain a semantic splitting result, the semantic splitting result contains multiple different semantics of the key word, and finally the semantic splitting result is subjected to word vector extraction to obtain word vectors corresponding to the multiple different semantics of the key word.

[0093] In an example, the pictographic element includes a Wubi root of Chinese characters, as shown in the following table. Figure 2b The splitting and mapping of the traditional Chinese characters according to the pictographic element includes:

[0094] S21, the traditional Chinese characters are split and mapped by the Wubi root of Chinese characters to obtain multiple English mapping units;

[0095] Each English mapping unit corresponds to a pictographic element and represents a semantic unit; for example, the Chinese character "nan" is split and mapped by the Wubi root to obtain English mapping units "l", "l" and "b", and the Chinese character "yong" is split and mapped by the Wubi root to obtain English mapping units "c", "e", "l" and "b". After the splitting and mapping, the input order of the semantics, word formation method and combination all have good explanatory meanings.

[0096] S22, the multiple English mapping units are combined to obtain multiple combination features representing different semantics of the key word.

[0097] In the embodiment, each English mapping unit represents a semantic unit, and after the multiple English mapping units are combined, different semantics can be obtained.

[0098] If the splitting maps N English mapping units, the combination in this embodiment means that m English mapping units are taken out from the N English mapping units for combination, where m is less than or equal to N. Preferably, during the combination process, the n-gram model can be used to perform n-gram processing on the plurality of English mapping units to obtain the n-gram features of the keyword. For example, if 3-gram is selected, the keyword "male" will be processed to obtain the 3-gram features: ^ll, llb, lb^(where ^ represents a placeholder).

[0099] In another example, the pictographic elements include the smallest segmented elements of traditional Chinese characters, as shown in Figure 3 The splitting and mapping of the traditional Chinese characters according to the pictographic elements includes:

[0100] S201, splitting the traditional Chinese characters according to the pictographic elements, and mapping the splitting results to obtain a plurality of mapping units,

[0101] Each mapping unit corresponds to a pictographic element and represents a semantic unit.

[0102] In this step, the splitting of the traditional Chinese characters according to the pictographic elements obtains the pictographic elements contained in the traditional Chinese characters, and the mapping of the pictographic elements to characters can be achieved by characterizing the pictographic elements contained in the traditional Chinese characters. One pictographic element is represented by a character, and the splitting results are finally mapped to characters. Wherein, the character can be a number, and the pictographic element is digitized to represent a pictographic element. The character can also be an English character, and the pictographic element is characterized to represent a pictographic element.

[0103] For example, the traditional Chinese character for fog is mist, which can be split into rain, spear and duty according to the pictographic elements. After digitizing, rain corresponds to the number 2, spear corresponds to the number 4, and duty corresponds to the number 1. Therefore, mist will be split and mapped to the mapping units "2", "4" and "1".

[0104] S202, combining the plurality of mapping units to obtain a plurality of combination features representing different semantics of the keyword.

[0105] In this embodiment, each mapping unit represents a semantic unit, and after the combination of the plurality of mapping units, different semantics can be obtained.

[0106] If the splitting maps N mapping units, the combination in the embodiment refers to taking m mapping units from the N mapping units for combination, wherein: m is less than or equal to N. Preferably, in the combination process, the plurality of mapping units can be subjected to n-gram processing by using an n-gram model to obtain n-gram features of the keyword. For example: selecting 3-gram, the keyword "mist" will be processed to obtain 3-gram features: 241, 124, 412.

[0107] At this point, a plurality of combination features representing different semantics of the keyword are obtained as a semantic splitting result. In the embodiment, the semantic splitting result can be subjected to one-hot coding to obtain a word vector of the enterprise entity.

[0108] S3, identifying the word vector by using a semantic correlation degree explanation model to obtain an opinion recognition result of the enterprise entity;

[0109] The semantic correlation degree explanation model is used to identify semantic correlation degrees of the keyword corresponding to the word vector and each word or Chinese character in the text training data, and words or Chinese characters with semantic correlation degrees meeting preset conditions are used as opinion recognition results to explain and describe the keyword in the semantic level. For example, each word vector obtained by one-hot coding can be input into the semantic correlation degree explanation model for identification, and word vectors of words or Chinese characters with semantic correlation degrees meeting preset conditions are output as target word vectors. Further, the semantic correlation degree explanation model can also perform embedding processing on the identified target word vectors, and directly output the embedded target word vectors. In this way, the embedded target word vectors can be directly used as node attributes in the enterprise knowledge graph, facilitating subsequent application of the enterprise knowledge graph.

[0110] In the embodiment of the application, the words or Chinese characters with semantic correlation degrees meeting preset conditions can be determined by determining semantic correlation degrees of the keyword and each word or Chinese character in the text training data, and words or Chinese characters ranked in the top preset positions in the order of semantic correlation degrees from large to small are used as words or Chinese characters meeting preset conditions. The words or Chinese characters with semantic correlation degrees greater than a preset correlation degree threshold can also be used as words or Chinese characters meeting preset conditions.

[0111] It should be noted that before the word vector is identified by using the semantic correlation degree explanation model, the semantic correlation degree explanation model needs to be trained so that it can distinguish the word vector of the enterprise entity from a plurality of semantics of the keyword. The training process can include:

[0112] S11, extracting training text from a corpus;

[0113] The corpus can be a Chinese disassembled dictionary database, a Chinese news database, or the like. The training text can be a complete article in the corpus. The more articles extracted in this step, the better the effect of the semantic correlation explanation model.

[0114] S12. Disassemble the Chinese characters in each training text according to the pictographic elements to obtain semantic disassembly results.

[0115] For example, for a single Chinese character in a training text, the Chinese character can be converted into a traditional Chinese character first, and then the traditional Chinese character can be disassembled and mapped according to the pictographic elements to obtain a semantic disassembly result.

[0116] The pictographic elements can be Chinese Wubi roots, and the disassembling and mapping of the traditional Chinese character according to the pictographic elements can include:

[0117] The traditional Chinese character is disassembled and mapped by Chinese Wubi roots to obtain a plurality of English mapping units, each English mapping unit corresponding to a pictographic element and representing a semantic unit.

[0118] The plurality of English mapping units are combined to obtain a plurality of combined features representing different semantics of the key words.

[0119] The pictographic elements can also be the smallest segmentation elements of the traditional Chinese character, and the disassembling and mapping of the traditional Chinese character according to the pictographic elements can include:

[0120] The traditional Chinese character is disassembled according to the pictographic elements.

[0121] The disassembly result is mapped to obtain a plurality of mapping units, each mapping unit corresponding to a pictographic element and representing a semantic unit.

[0122] The plurality of mapping units are combined to obtain a plurality of combined features representing different semantics of the key words.

[0123] S13. The word vector obtained by one-hot encoding the semantic disassembly result of the Chinese character in each training text is used as a training set to train the semantic correlation explanation model, so that the model is used to confirm the correlation or correlation degree between different key words in an article.

[0124] In the embodiment, the correlation or relevance between different keywords in the article can be obtained by the repetition of each keyword in different articles to obtain the relevance between keywords; for example, the word "fog" can form different words or be used alone, when forming "fog", the articles related to "fog" are likely to mostly describe the words related to "fog governance" or "fog prevention", and when used alone, most of them are describing fog weather, and no words related to governance appear, at this time, according to the repetition times of the words in the article and the categories described in the articles containing "fog", the correlation or relevance between "fog" and other "keywords" is further determined; similarly, when the word "haze" appears, the related articles are basically describing the governance of fog and haze, so the correlation or relevance of keywords corresponding to such words will be more accurate.

[0125] Preferably, in step S12, after obtaining the semantic segmentation result, a word library can be constructed according to the semantic segmentation result, the word library containing the semantic segmentation result corresponding to each pictographic element, and in subsequent step S2, the key word can be converted into a traditional Chinese character, the target pictographic element contained in the traditional Chinese character is segmented, and the semantic segmentation result corresponding to the target pictographic element can be directly queried from the word library. The mapping and combination process is omitted.

[0126] S4, obtaining economic information of each enterprise entity, and determining economic relations between the enterprise entities according to the economic information;

[0127] In the embodiment of the application, the economic relations can be production, distribution, exchange and consumption relations occurring in the total social production process. Considering the operating characteristics of the enterprise entities, in the embodiment, the economic relations include at least one of transaction information, production relations, sales relations, investment relations, financing relations, debt relations, guarantee relations and upstream and downstream relations between the enterprise entities.

[0128] In the embodiment of the application, for small and micro enterprises, transaction information and upstream and downstream industries between the small and micro enterprises can be determined according to their operating information, and debt relations, investment relations, equity penetration and other relations can be determined through their public information. The relations between different two enterprises can be determined by obtaining the equity relations and enterprise operating information of the enterprises disclosed in the enterprise query website, and the enterprise debt situation and creditors can be determined through the debt information disclosed by the enterprises, and the company holding information, equity mortgage and other information can be determined according to the information disclosed in the securities market.

[0129] S5, constructing an enterprise knowledge graph according to the public opinion recognition results of each enterprise entity and the economic relations between the enterprise entities;

[0130] In the embodiment of the present application, the enterprise is accurately described through the public opinion recognition result of the enterprise entity, and meanwhile, more accurate enterprise and enterprise relationship are mined by combining the economic relationship between enterprises, so that a more accurate enterprise knowledge graph is constructed, thereby effectively determining the risk situation of the enterprise and guaranteeing data security and property safety.

[0131] In an example, the enterprise knowledge graph is constructed according to the public opinion recognition result of each enterprise entity and the economic relationship between the enterprise entities, and the method comprises the following steps:

[0132] S51, constructing a node corresponding to each of the enterprise entities, and taking the public opinion recognition result of the enterprise entity and the economic relationship between the enterprise entities as attribute information of the node;

[0133] In the embodiment of the present application, a corresponding node is constructed for each enterprise entity, and the public opinion recognition result and the economic relationship between the enterprise entities determined in the above step are taken as attribute information of the node. Of course, the attribute information of the node can also include the name of the enterprise entity, the industry to which the enterprise entity belongs and other related information, which will not be described herein.

[0134] S52, determining the connection edge between the nodes according to the economic relationship between the enterprise entities;

[0135] In the embodiment of the present application, it is determined whether there is an association between two enterprise entities according to the economic relationship between the enterprise entities, and whether the nodes are associated is determined based on this. If there is an association, the nodes corresponding to the two enterprise entities can be connected, thereby improving the classification accuracy of the knowledge graph.

[0136] S53, constructing the enterprise knowledge graph based on the connection edge and the node;

[0137] In the embodiment of the present application, the corresponding enterprise knowledge graph is constructed according to the generated node and the determined connection edge between the nodes, so that the system or the user can quickly determine the relationship between different enterprise entities based on the enterprise knowledge graph, thereby improving the data processing efficiency.

[0138] Further, in a specific example, the economic relationship includes transaction information and economic association relationship between enterprise entities.

[0139] The transaction information can reflect the transaction between enterprises, for example, two enterprise entities A and B, A purchases raw materials from B and pays the corresponding payment m to B. The transaction information in the node corresponding to the enterprise entity A is to pay the payment m to B, and the transaction information in the node corresponding to the enterprise entity B is to receive the payment m paid by A.

[0140] The economic association relationship can reflect whether the enterprise has financial hidden dangers and fraud hidden dangers, and the economic association relationship can be various relationships between enterprises such as relative relationship, shareholding relationship, mortgage loan relationship, investment relationship, debt relationship, guarantee relationship, upstream and downstream relationship and the like.

[0141] In step S51, the public opinion recognition result of the enterprise entity and the economic relationship between the enterprise entities are taken as attribute information of the node, including:

[0142] S511, adjusting the transaction information between the enterprise entities through the economic association relationship, so that the value of the transaction information between the enterprise entities with the preset relationship in the economic association relationship is reduced;

[0143] S512, taking the public opinion recognition result of the enterprise entity and the transaction information between the enterprise entities as attribute information of the node.

[0144] In the embodiment of the application, if there is an economic association relationship between the enterprise entities, the transaction between the enterprise entities may have certain fraud hidden dangers, for example, when there is cross equity or economic relationship between two companies, the transaction information between the two companies may have authenticity doubts or internal transaction problems due to various reasons, at this time, the value of the transaction information between the enterprise entities with the preset relationship in the economic association relationship is adjusted and reduced, thereby improving the judgment of the revenue capacity of the enterprise entity. Wherein: the preset relationship can be at least two relationships in the economic association relationship, or a specified number of relationships in the economic association relationship. For example, the value of the transaction information between the enterprise entities with the guarantee relationship and the investment relationship in the economic association relationship is adjusted and reduced.

[0145] Accordingly, the transaction information between different enterprise entities is added with attribute information in the generated node. In the present scheme, only the transaction information between the enterprise entities is retained in the attribute information of the node, purely from the revenue capacity of the company itself, the business status of the enterprise entity is determined, the business capacity of the enterprise entity is wrongly estimated because of the existence of the holding company or the guarantee company, and at the same time, the business capacity of the enterprise is wrongly overestimated because of the internal transaction, the classification accuracy of the enterprise entity is improved.

[0146] The connection edge between the nodes is determined according to the economic relationship between the enterprise entities in S52, and the connection edge between the nodes is determined according to transaction information between the enterprise entities. For example, the transaction relationship between two enterprise entities can be determined according to the transaction information between the enterprise entities, and then the connection edge between the nodes is determined according to the transaction relationship between the two enterprise entities. For example, whether the two enterprise entities exist in a transaction relationship is determined according to the transaction information between the enterprise entities, and whether the nodes are associated is determined according to the transaction relationship. If the two enterprise entities are associated, the nodes corresponding to the two enterprise entities are connected, otherwise, the nodes corresponding to the two enterprise entities are not connected.

[0147] In the embodiment of the application, the connection relationship of the enterprise knowledge graph is optimized based on the above scheme, so that the nodes with transaction relationship between the enterprise entities are connected, the misjudgment caused by connection due to other economic relationships is reduced, the information amount of the enterprise knowledge graph is reduced, and the application efficiency of the enterprise knowledge graph is improved.

[0148] In the embodiment of the application, the establishment of each company is actually to realize revenue and good operation, and the economic association relationship between different companies will not improve the quality of the products or the service of the company itself, but is easy to cause internal transaction or internal support. The property information of the enterprise entity is adjusted according to the transaction information and the economic association relationship of the enterprise entity, the actual situation of the enterprise entity is determined from the revenue capacity of the enterprise entity, and the competitiveness of the enterprise entity in the actual operation process is determined. Especially for small and micro enterprises, determining the competitiveness of the small and micro enterprises is the core standard for providing support to the small and micro enterprises.

[0149] In order to more accurately reflect the complex economic relationship between enterprises, the economic association relationship includes at least one of investment relationship, debt relationship, guarantee relationship and upstream and downstream relationship; the value y of the transaction information between the adjusted enterprise entities i The value y can be determined by the following formula:

[0150]

[0151] Wherein: x i is the original value of the transaction information between the enterprise entities, a i is the guarantee percentage or the stock ownership percentage, investment relationship=true indicates that the enterprise entities include investment relationship, upstream and downstream relationship=true indicates that the enterprise entities include upstream and downstream relationship, debt relationship=true indicates that the enterprise entities include debt relationship, and guarantee relationship=true indicates that the enterprise entities include guarantee relationship. The percentage is 100 times the ratio value, and if the guarantee ratio of the enterprise entity is N%, a i =N.

[0152] In the embodiment of the present application, if the two companies are in an upstream-downstream relationship and there is an investment relationship between the two companies, it is possible to directly purchase without considering the product quality of the company. This not only does not improve the competitiveness of the company's products, but also makes the product be eliminated because it does not need to be updated. In order to avoid misjudgment of the enterprise entity due to some internal transactions, the value of the transaction information in the formula is adjusted in the present solution to improve the accuracy of the judgment of the enterprise entity.

[0153] In this way, the relationship between the enterprise entities is no longer independent or linear. For enterprises that have cross relationships (such as: enterprises that have both investment relationships and guarantee relationships), the transaction relationship between them is adjusted to weaken the possibility of increasing data volume through related transactions between enterprises with specific relationships, thereby improving the accuracy of the relationship between enterprise entities.

[0154] For example, in the process of constructing an enterprise knowledge graph, an enterprise knowledge graph can be constructed based on M enterprise entities and N transaction information between the M entities. For example: knowledge fusion is performed on M enterprise entities and N transaction information, and an enterprise knowledge graph is constructed using the knowledge fused data. The enterprise knowledge graph includes M nodes, M node attributes, and directed edges between nodes with transaction information; the M nodes correspond one-to-one to the M enterprise entities, and the public opinion recognition result of each enterprise entity can be used as attribute information of the corresponding node.

[0155] S6, classifying the enterprise entities according to the enterprise knowledge graph to determine the level of each enterprise entity.

[0156] In the present solution, the number of nodes connected by each node in the enterprise knowledge graph and the actual operating conditions can be used to classify enterprise entities and determine enterprises with core competitiveness or development potential. At the same time, enterprises with insufficient competitiveness can also be screened out, and even pure small and micro enterprises that simply extract funds can be determined. Through the present solution, it is convenient to determine the intensity of support provided to different small and micro enterprises, and economic losses caused by aid are effectively avoided.

[0157] For example, on the constructed enterprise knowledge graph, the attribute information of each node and the edges between each node can be queried. Using the enterprise knowledge graph, other enterprise entities associated with the target enterprise can be queried. Through the connection relationship in the graph, the enterprise entities can be classified according to the public opinion recognition result in the enterprise knowledge graph, the transaction information between the enterprise entities, and the associated enterprise conditions. Alternatively, a classification model for classifying enterprise entities can be trained using the data in the enterprise knowledge graph, and the classification model can be used to determine the level of the enterprise entities.

[0158] Figure 3 An enterprise identification device based on an enterprise knowledge graph, as shown in Figure 4 The device comprises:

[0159] A first acquisition module 31 is configured to acquire public opinion information of each enterprise entity and acquire Chinese keywords in the public opinion information.

[0160] A word vector processing module 32 is configured to perform semantic splitting on the keywords respectively according to pictographic elements and output word vectors of the enterprise entities according to the semantic splitting results.

[0161] An identification module 33 is configured to identify the word vectors by a semantic correlation degree interpretation model to obtain public opinion identification results of the enterprise entities, wherein the semantic correlation degree interpretation model is configured to identify semantic correlation degrees of the keywords corresponding to the word vectors and each word or Chinese character in text training data, and perform semantic-level interpretation and explanation on the keywords by taking words or Chinese characters with semantic correlation degrees meeting preset conditions as the public opinion identification results.

[0162] A second acquisition module 34 is configured to acquire economic information of each enterprise entity and determine economic relationships between the enterprise entities according to the economic information.

[0163] A construction module 35 is configured to construct an enterprise knowledge graph according to the public opinion identification results of each enterprise entity and the economic relationships between the enterprise entities.

[0164] A determination module 36 is configured to grade the enterprise entities according to the enterprise knowledge graph and determine levels of each of the enterprise entities.

[0165] In an embodiment, the word vector processing module 32 comprises:

[0166] A conversion module is configured to convert the keywords into traditional Chinese characters.

[0167] A splitting and mapping module is configured to perform splitting and mapping processing on the traditional Chinese characters according to pictographic elements to obtain semantic splitting results.

[0168] Optionally, the pictographic elements comprise Chinese Wubi roots, and the splitting and mapping module comprises:

[0169] A first splitting and mapping module is configured to perform splitting and mapping on the traditional Chinese characters by Chinese Wubi roots to obtain a plurality of English mapping units, each of which corresponds to a pictographic element and represents a semantic unit.

[0170] A first combination module is configured to combine the plurality of English mapping units to obtain a plurality of combination features representing different semantics of the keywords.

[0171] Optionally, the pictographic elements include the smallest segmentation elements of traditional Chinese characters, and the splitting mapping module includes:

[0172] The splitting module is used to split traditional Chinese characters based on pictographic elements;

[0173] The mapping module is used to map the splitting results; it yields multiple mapping units, each corresponding to a pictographic element and representing a semantic unit.

[0174] The second combination module is used to combine the multiple mapping units to obtain multiple combined features representing different semantics of the keywords.

[0175] Furthermore, the word vector processing module 32 also includes:

[0176] The encoding module is used to perform one-hot encoding on the semantic segmentation results to obtain word vectors for enterprise entities.

[0177] Preferably, the construction module 35 is specifically used to construct nodes corresponding to each enterprise entity, and to use the public opinion identification results of the enterprise entity and the economic relationship between the enterprise entities as the attribute information of the nodes; to determine the connection edges between the nodes according to the economic relationship between the enterprise entities; and to construct the enterprise knowledge graph based on the connection edges and the nodes.

[0178] The economic relationships include: transaction information and economic connections between business entities;

[0179] Preferably, the construction module 35 is specifically used to adjust the transaction information between enterprise entities through the economic relationship, so that the value of the transaction information between enterprise entities with the economic relationship including the preset relationship is reduced; and to use the public opinion identification results of the enterprise entities and the transaction information between enterprise entities as the attribute information of the node;

[0180] Preferably, the construction module 35 is further configured to determine the connection edges between the nodes based on the transaction information between the enterprise entities.

[0181] The economic relationships mentioned include at least one of the following: investment relationships, debt relationships, guarantee relationships, and upstream and downstream relationships;

[0182] The construction module 35 determines the adjusted value y of the transaction information between the enterprise entities using the following formula. i :

[0183]

[0184] Where: x ia is the original attribute value of the enterprise entity. i This is to guarantee a percentage of ownership or equity.

[0185] Those skilled in the art will understand that the modules in the above-described device embodiments can be distributed throughout the device as described, or they can be modified accordingly and distributed in one or more devices different from the above embodiments. The modules in the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0186] The following describes embodiments of the electronic device of the present invention, which can be considered as implementations of the physical form of the methods and apparatus embodiments of the present invention described above. Details described in the embodiments of the electronic device of the present invention should be considered as supplements to the methods or apparatus embodiments described above; details not disclosed in the embodiments of the electronic device of the present invention can be implemented with reference to the methods or apparatus embodiments described above.

[0187] Figure 4 This is a structural block diagram of an exemplary embodiment of an electronic device according to the present invention. Figure 4 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0188] like Figure 1 As shown, the electronic device 400 of this exemplary embodiment is manifested in the form of a general data processing device. The components of the electronic device 400 may include, but are not limited to: at least one processing unit 410, at least one storage unit 420, a bus 430 connecting different electronic device components (including storage unit 420 and processing unit 410), a display unit 440, etc.

[0189] The storage unit 420 stores a computer-readable program, which may be source code or read-only program code. The program can be executed by the processing unit 410, causing the processing unit 410 to perform the steps of various embodiments of the present invention. For example, the processing unit 410 can perform actions such as... Figure 4 The steps are shown.

[0190] The storage unit 420 may include a readable medium in the form of a volatile storage unit, such as a random access memory (RAM) unit 4201 and / or a cache memory unit 4202, and may further include a read-only memory unit (ROM) unit 4203. The storage unit 420 may also include a program / utility 4204 having a set (at least one) of program modules 4205, such program modules 4205 including, but not limited to: operating electronic devices, one or more application programs, other program modules, and program data; each or some combination of these examples may include an implementation of a network environment.

[0191] Bus 430 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0192] Electronic device 400 can also communicate with one or more external devices 100 (e.g., keyboard, display, network device, Bluetooth device, etc.), enabling users to interact with electronic device 400 via these external devices 100, and / or enabling electronic device 400 to communicate with one or more other data processing devices (e.g., router, modem, etc.). This communication can be made via input / output (I / O) interface 450, and also via network adapter 460 to one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet). Network adapter 460 can communicate with other modules of electronic device 400 via bus 430. It should be understood that, although... Figure 5 As not shown, other hardware and / or software modules may be used in electronic device 400, including but not limited to: microcode, device drivers, redundancy processing units, external disk drive arrays, RAID electronics, tape drives, and data backup storage electronics.

[0193] Figure 5 This is a schematic diagram of a computer-readable medium embodiment of the present invention. ​As shown, the computer program can be stored on one or more computer-readable media. A computer-readable medium can be a readable signal medium or a readable storage medium. A readable storage medium can be, for example, but not limited to, an electronic device, apparatus, or device that is electrical, magnetic, optical, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. When the computer program is executed by one or more data processing devices, it enables the computer-readable medium to implement the above-described method of the present invention, namely: acquiring public opinion information of various enterprise entities and acquiring Chinese keywords in the public opinion information; semantically decomposing the keywords according to pictographic elements and outputting word vectors of the enterprise entities according to the semantic decomposition results; identifying the word vectors through a semantic relevance interpretation model to obtain the public opinion recognition results of the enterprise entities, wherein the semantic relevance interpretation model is used to identify the semantic relevance between the keywords corresponding to the word vectors and each word or Chinese character in the text training data, and uses words or Chinese characters whose semantic relevance meets preset conditions as public opinion recognition results to provide semantic explanations for the keywords; constructing an enterprise knowledge graph based on the public opinion recognition results of various enterprise entities and the economic relationships between the enterprise entities; classifying the enterprise entities according to the enterprise knowledge graph to determine the level of each enterprise entity.

[0194] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described in this invention can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this invention can be embodied in the form of a software product, which can be stored in a computer-readable storage medium (such as a CD-ROM, USB flash drive, portable hard drive, etc.) or on a network, including several instructions to cause a data processing device (such as a personal computer, server, or network device, etc.) to execute the methods described above according to this invention.

[0195] The computer-readable storage medium may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, capable of transmitting, propagating, or transmitting programs for use by or in conjunction with an electronic device, apparatus, or device that executes instructions. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0196] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as "C" or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0197] In summary, the present invention can be implemented by methods, apparatus, electronic devices, or computer-readable media that execute computer programs. In practice, some or all of the functions of the present invention can be implemented using general-purpose data processing devices such as microprocessors or digital signal processors (DSPs).

[0198] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the present invention is not inherently related to any specific computer, virtual device, or electronic device, and various general-purpose devices can also implement the present invention. The above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for enterprise identification based on enterprise knowledge graph, characterized in that, The method includes: Obtain public opinion information for each enterprise entity, and extract Chinese keywords from the public opinion information; Convert keywords into traditional Chinese characters; Traditional Chinese characters are decomposed and mapped using the Wubi input method to obtain multiple English mapping units. Each English mapping unit corresponds to a Wubi input method root and represents a semantic unit. These multiple English mapping units are combined to obtain multiple combined features representing different semantics of the keyword. Alternatively, traditional Chinese characters are decomposed based on their smallest segmentation element. The decomposition results are mapped to obtain multiple mapping units, each corresponding to a smallest segmentation element of a traditional Chinese character and representing a semantic unit. These multiple mapping units are combined to obtain multiple combined features representing different semantics of the keyword. One-hot encoding is performed on multiple combined features representing different semantics of the keyword to obtain word vectors corresponding to multiple different semantics of the keyword; The semantic relevance interpretation model is trained so that it determines the semantic relevance between different keywords based on the repetition of each keyword in the text training data and the category to which the text containing the keyword belongs. The model then distinguishes the word vector that most accurately explains the enterprise entity from the word vectors corresponding to multiple different semantics of the keyword. The word vectors are identified by a semantic relevance interpretation model to obtain the public opinion recognition result of the enterprise entity. The semantic relevance interpretation model is used to identify the semantic relevance between the keywords corresponding to the word vectors and each word or Chinese character in the text training data, and to use the words or Chinese characters whose semantic relevance meets the preset conditions as the public opinion recognition result to provide a semantic explanation of the keywords. Obtain economic information of each enterprise entity and determine the economic relationships between the enterprise entities based on the economic information; A corporate knowledge graph is constructed based on the public opinion identification results of each corporate entity and the economic relationships between the corporate entities. The enterprise entities are classified according to the enterprise knowledge graph to determine the level of each enterprise entity.

2. The method according to claim 1, characterized in that, The construction of the enterprise knowledge graph based on the public opinion identification results of each enterprise entity and the economic relationships between the enterprise entities includes: Construct a node corresponding to each of the aforementioned enterprise entities, and use the public opinion identification results of the enterprise entities and the economic relationships between the enterprise entities as the attribute information of the nodes; The connection edges between the nodes are determined based on the economic relationships between the enterprise entities; The enterprise knowledge graph is constructed based on the connecting edges and the nodes.

3. The method according to claim 2, characterized in that, The economic relationships include: transaction information and economic connections between business entities; The public opinion identification results of the enterprise entities and the economic relationships between the enterprise entities are used as the attribute information of the nodes, including: The transaction information between corporate entities is adjusted through the aforementioned economic relationships, thereby reducing the value of transaction information between corporate entities with pre-defined economic relationships. The public opinion identification results of the enterprise entities and the transaction information between the enterprise entities are used as the attribute information of the nodes; Determining the connection edges between the nodes based on the economic relationships between the enterprise entities includes: The connection edges between the nodes are determined based on the transaction information between the enterprise entities.

4. The method according to claim 3, characterized in that, The economic relationships mentioned include at least one of the following: investment relationships, debt relationships, guarantee relationships, and upstream and downstream relationships; Adjusting the transaction information of each node based on the aforementioned economic relationships includes: The adjusted value y of the transaction information between the aforementioned enterprise entities i Determined by the following formula: Where: x i a represents the raw numerical value of transaction information between business entities. i For guarantee percentage or equity percentage, investment relationship = true indicates that there is an investment relationship between corporate entities, upstream and downstream relationship = true indicates that there is an upstream and downstream relationship between corporate entities, debt relationship = true indicates that there is a debt relationship between corporate entities, and guarantee relationship = true indicates that there is a guarantee relationship between corporate entities.

5. A corporate identification device based on enterprise knowledge graph, characterized in that, The device includes: The first acquisition module is used to acquire public opinion information of various enterprise entities and to acquire Chinese keywords in the public opinion information; The conversion module is used to convert keywords into traditional Chinese characters. The splitting and mapping module is used to split and map traditional Chinese characters using the Chinese Wubi input method to obtain multiple English mapping units. Each English mapping unit corresponds to a Chinese Wubi input method root and represents a semantic unit. The multiple English mapping units are combined to obtain multiple combined features representing different semantics of the keyword. Alternatively, traditional Chinese characters are split according to their minimum segmentation element. The splitting results are mapped to obtain multiple mapping units, each corresponding to a minimum segmentation element of a traditional Chinese character and representing a semantic unit. The multiple mapping units are combined to obtain multiple combined features representing different semantics of the keyword. The encoding module is used to perform one-hot encoding on multiple combined features representing different semantics of the keyword to obtain word vectors corresponding to multiple different semantics of the keyword; The training module is used to train the semantic relevance interpretation model, so that the semantic relevance interpretation model determines the semantic relevance between different keywords based on the repetition of each keyword in the text training data and the category to which the text containing the keyword belongs, and distinguishes the word vector that can most accurately explain the enterprise entity from the word vectors corresponding to multiple different semantics of the keyword. The identification module is used to identify the word vectors through a semantic association degree interpretation model to obtain the public opinion identification result of the enterprise entity. The semantic association degree interpretation model is used to identify the semantic association degree between the keywords corresponding to the word vectors and each word or Chinese character in the text training data, and to use the words or Chinese characters whose semantic association degree meets the preset conditions as the public opinion identification result to provide a semantic explanation of the keywords. The second acquisition module is used to acquire economic information of each enterprise entity and determine the economic relationship between the enterprise entities based on the economic information. The construction module is used to build an enterprise knowledge graph based on the public opinion identification results of each enterprise entity and the economic relationship between the enterprise entities; The determination module is used to classify the enterprise entities according to the enterprise knowledge graph and determine the level of each enterprise entity.

6. The apparatus according to claim 5, characterized in that, The construction module is specifically used to construct nodes corresponding to each enterprise entity, and to use the public opinion identification results of the enterprise entity and the economic relationship between the enterprise entities as the attribute information of the nodes. The connection edges between the nodes are determined based on the economic relationships between the enterprise entities; The enterprise knowledge graph is constructed based on the connecting edges and the nodes.

7. The apparatus according to claim 6, characterized in that, The economic relationships include: transaction information and economic connections between business entities; The construction module is specifically used to adjust the transaction information between enterprise entities through the economic relationship, so that the value of the transaction information between enterprise entities with the economic relationship, including the preset relationship, is reduced; and the public opinion identification results of the enterprise entities and the transaction information between enterprise entities are used as the attribute information of the nodes. The construction module is also specifically used to determine the connection edges between the nodes based on the transaction information between the enterprise entities.

8. The apparatus according to claim 7, characterized in that, The economic relationships mentioned include at least one of the following: investment relationships, debt relationships, guarantee relationships, and upstream and downstream relationships; The construction module determines the adjusted value y of the transaction information between the enterprise entities using the following formula. i : Where: x i a represents the raw numerical value of transaction information between business entities. i For guarantee percentage or equity percentage, investment relationship = true indicates that there is an investment relationship between corporate entities, upstream and downstream relationship = true indicates that there is an upstream and downstream relationship between corporate entities, debt relationship = true indicates that there is a debt relationship between corporate entities, and guarantee relationship = true indicates that there is a guarantee relationship between corporate entities.

9. An electronic device, comprising: processor; as well as A memory storing computer-executable instructions, which, when executed, cause the processor to perform the method according to any one of claims 1-4.

10. A computer-readable storage medium, wherein, The computer-readable storage medium stores one or more programs that, when executed by a processor, implement the method of any one of claims 1-4.

Images