Industrial chain map intelligent generation method and electronic equipment

Abstract

A method and electronic device for intelligently generating industrial chain maps. This method uses multiple artificial intelligence models to deeply mine the relationship features within complex enterprise and industry information, determining the relationship types and strengths between individual enterprises and / or industries. Then, centered around a target entity word (e.g., an industry word), related industry words with high relationship strengths are connected to the target entity word and displayed, layer by layer expanding to form a visual industrial chain map corresponding to the target entity word. While greatly improving the accuracy of intelligent industrial chain mining, the visual display of the industrial chain map allows business operators to intuitively obtain the information they need, significantly increasing the efficiency of obtaining effective information for business decision-making.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and in particular to methods and electronic equipment for intelligently generating industrial chain maps. Background Art

[0002] The healthy development of a company depends heavily on the current state of its upstream and downstream industries and peers. The state of the industry chain to which a company belongs is a crucial basis for assessing its own development health, its industrial layout, and its research into competitive and cooperative relationships, among other areas of development.

[0003] However, due to the large number of companies and industries, the relevant information is voluminous and complex. When mining enterprise industry chains, there are often issues with unclear industry definitions, ambiguous industry boundaries, and complex industry segments. These factors significantly impact the accuracy of industry chain mining, making the industry chains mined using related technologies often inaccurate and unsuitable for business decision-making. Summary of the Invention

[0004] This application provides an intelligent generation method and electronic equipment for industrial chain maps, which are used to improve the accuracy and visibility of intelligent mining of industrial chains, thereby improving the efficiency of obtaining effective information in corporate business decision-making.

[0005] In the first aspect, the present application provides a method for intelligently generating an industrial chain map, including: using a named entity recognition model to identify a relationship information text set to obtain a specified entity set; the relationship information text set is obtained by extracting text content representing relationship information between industries from a knowledge corpus that stores enterprise information, and the entities in the specified entity set include industry words in the relationship information text set; the relationship information text set and the specified entity set are input into a relationship extraction model to obtain an entity relationship set; the entity relationship set includes multiple entity relationship pairs, and an entity relationship pair includes two entities and the type of relationship between the two entities; based on the specified entity set, the entity relationship set and the knowledge corpus, an entity frequency feature is determined, and the entity frequency feature is used to represent the entity and entity relationship pairs in the knowledge corpus. The frequency of occurrence in the corpus; the entity relationship set and the entity frequency feature are input into the industrial chain relationship strength prediction model to obtain an entity relationship strength set; the entity relationship strength set includes multiple entity pair strength relationships, and an entity pair strength relationship includes two entities, the relationship type of the two entities and the relationship strength values ​​of the two entities; when an industrial chain map generation instruction including a first entity word is received, a first industrial chain map corresponding to the first entity word is generated based on the entity relationship strength set and the first entity word; in the first industrial chain map, the first entity word and the second entity word are connected by a line to show that the first entity word and the second entity word are in the first entity pair strength relationship in the entity relationship strength set, and the relationship strength value in the first entity pair strength relationship is greater than the preset first strength threshold.

[0006] In combination with some embodiments of the first aspect, in some embodiments, the entities in the specified entity set also include enterprise names in the relationship information text set.

[0007] In combination with some embodiments of the first aspect, in some embodiments, the entity frequency features are determined based on the specified entity set, the entity relationship set and the knowledge corpus, specifically including: based on the specified entity set, the entity relationship set and the knowledge corpus, the entity frequency features and the enterprise correlation features are determined, and the enterprise correlation features are used to represent the degree of association between enterprises and industries and / or between enterprises; the entity relationship set and the entity frequency features are input into the industrial chain relationship strength prediction model to obtain the entity relationship strength set, specifically including: the entity relationship set, the entity frequency features and the enterprise correlation features are input into the industrial chain relationship strength prediction model to obtain the entity relationship strength set.

[0008] In combination with some embodiments of the first aspect, in some embodiments, the enterprise relevance characteristics include: the similarity between enterprises corresponding to the enterprise names, the popularity of enterprises corresponding to the enterprise names, the associated exposure of enterprises corresponding to the enterprise names, and the number of enterprises covered by the words in various regions and industries.

[0009] In combination with some embodiments of the first aspect, in some embodiments, the entity frequency feature includes: a word frequency feature and a co-occurrence feature. The word frequency feature is used to represent the number of times industry words in the entity relationship set appear in the knowledge corpus, and the co-occurrence feature is used to represent the number of times entity relationship pairs appear together in the knowledge corpus.

[0010] In combination with some embodiments of the first aspect, in some embodiments, the relationship type is one of a preset relationship type set, which includes a plurality of relationship types representing upstream and downstream relationships between industries or enterprises, and a plurality of relationship types representing competitive or cooperative relationships between industries or enterprises.

[0011] In combination with some embodiments of the first aspect, in some embodiments, when the industrial chain graph is set to expand only two layers, when an industrial chain graph generation instruction including a first entity word is received, a first industrial chain graph corresponding to the first entity word is generated based on the entity relationship strength set and the first entity word, specifically including: when an industrial chain graph generation instruction including the first entity word is received, multiple first-layer entity pair strength relationships including the first entity word in the entity relationship strength set and whose relationship strength values ​​are greater than a preset first strength threshold are determined; with the first entity word as the center, another entity in each of the first-layer entity pair strength relationships is connected with the first entity word, and the corresponding relationship type and relationship strength value are displayed on each connection line; for the second entity word in any first-layer entity pair strength relationship, multiple second-layer entity pair strength relationships including the second entity word in the entity relationship strength set and whose relationship strength values ​​are greater than a preset second strength threshold are determined; another entity in each second-layer entity pair strength relationship is connected with the corresponding second entity word, and the corresponding relationship type and relationship strength value are displayed on each connection line to form the first industrial chain graph corresponding to the first entity word.

[0012] In combination with some embodiments of the first aspect, in some embodiments, the preset second intensity threshold is greater than the preset first intensity threshold.

[0013] In a second aspect, an embodiment of the present application provides an electronic device, comprising: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the electronic device to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0014] In a third aspect, an embodiment of the present application provides a computer-readable storage medium comprising instructions, which, when executed on an electronic device, enables the electronic device to execute the method described in the first aspect and any possible implementation of the first aspect.

[0015] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0016] 1. The relationship extraction model extracts the relationship types between entities related to industry terms. The industry chain relationship strength prediction model then determines the relationship strength between these entities. Finally, upon receiving an instruction to generate an industry graph for an industry term, the entity terms with relationship strengths greater than a preset first threshold are connected to the industry term and expanded layer by layer, thereby obtaining the industry chain graph for the industry term. By deeply mining industry and enterprise relationship information, the accuracy and visibility of intelligent industry chain mining are greatly improved, and the efficiency of obtaining effective information for enterprise management decision-making is enhanced.

[0017] 2. Since the specified entity set includes both industry terms and company names, the relationship between companies and industries can be deeply explored. Therefore, not only can the representative companies of the industry nodes be displayed in the formed industrial chain map, but the industrial chain map corresponding to the company name can also be directly generated, which greatly improves the accuracy of the information displayed in the industrial chain map.

[0018] 3. When expanding entities layer by layer in the process of generating the industrial chain map, the preset strength threshold for determining the strength relationship of entity pairs in the next layer can be set higher than the preset strength threshold for determining the strength relationship of entity pairs in the previous layer (for example, the preset second strength threshold is greater than the preset first strength threshold). In this way, the generation process of the industrial chain map can converge naturally, thereby improving the completeness and accuracy of the industrial chain map information and the generation efficiency of the industrial chain map. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of the method for intelligently generating an industrial chain map in an embodiment of this application;

[0020] Figure 2AThis is a schematic diagram for understanding the functions and training methods of the named entity recognition model in the embodiment of the present application;

[0021] Figure 2B This is a schematic diagram for understanding the functions and training methods of the relationship extraction model in the embodiment of the present application;

[0022] Figure 2C This is a schematic diagram for explaining the functions and training methods of the industrial chain relationship strength prediction model in the embodiment of this application;

[0023] Figure 3A This is an exemplary schematic diagram of the industrial chain map generated in the embodiments of this application;

[0024] Figure 3B This is another exemplary schematic diagram of the industrial chain map generated in the embodiments of this application;

[0025] Figure 3C This is another exemplary schematic diagram of the industrial chain map generated in the embodiments of this application;

[0026] Figure 4 This is another flowchart of the method for intelligently generating an industrial chain map in an embodiment of this application;

[0027] Figure 5 It is a schematic diagram of an exemplary structure of an electronic device 500 provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The terms used in the following examples of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification and appended claims of this application, the singular expressions "a," "an," "said," "above," "the," and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in this application refers to any or all possible combinations comprising one or more of the listed items.

[0029] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0030] The intelligent generation method of the industrial chain map provided in the embodiment of the present application can deeply mine the relationship characteristics in the complex enterprise and industry information through multiple artificial intelligence models to obtain the relationship type and relationship strength between each enterprise and / or industry, and then, with a target entity word (such as an industry word) as the center, connect the related industry words with high relationship strength with the target entity word and display them, expanding layer by layer to form a visual industrial chain map corresponding to the target entity word.

[0031] While greatly improving the accuracy of intelligent mining of the industrial chain, the visual display of the industrial chain map allows business operators to intuitively obtain the required information, greatly improving the efficiency of obtaining effective information in business decision-making.

[0032] The following is a detailed description of the method for intelligently generating an industrial chain map in the embodiment of this application:

[0033] See also Figure 1 , which is a flow chart of the method for intelligently generating an industrial chain map in an embodiment of this application.

[0034] S101, using a named entity recognition model to identify a relationship information text set to obtain a specified entity set;

[0035] The relationship information text set is obtained by extracting text content representing relationship information between industries from a knowledge corpus storing enterprise information. The entities in the specified entity set include industry words in the relationship information text set.

[0036] Users can save the collected company or industry profiles, news, consultations and other company-related content into a knowledge corpus. The knowledge corpus can be created by the user himself or provided by other entities, which is not limited here.

[0037] Before generating the industrial chain map, users can extract text content representing the relationship information between industries from the knowledge corpus to form a relationship information text collection.

[0038] In some embodiments, in addition to extracting text content representing relationship information between industries, users can also extract text content representing relationship information between industries and enterprises and between enterprises to form a relationship information text set.

[0039] The purpose of extracting this set of relationship information text from the knowledge corpus is to make the model's identification more targeted, thereby making the generated industrial chain map more accurate. In some embodiments, if the majority of the content in the knowledge corpus describes the relationships between industries and / or enterprises, or if the knowledge corpus is specifically established for generating an industrial chain map (its content is pre-selected), the entire content in the knowledge corpus can also be used as the set of relationship information text, which is not limited here.

[0040] Electronic devices can use a named entity recognition model to identify this set of relationship information text and obtain a set of designated entities. It is understood that an industrial chain is composed of industry nodes, each of which can be represented by a corresponding industry term. Therefore, in general, the designated entities in this setting include industry terms. Therefore, after identifying this set of relationship information text, a large number of industry terms can be obtained.

[0041] In some embodiments, in addition to using industry terms as designated entities, other entities may also be used as designated entities, such as company names. Thus, after identifying the relationship information text set, a large number of industry terms and company names may be obtained.

[0042] For example, for a text in the relationship information text set that reads "mainly engaged in the research and development, production, and sales of semiconductor laser butterfly packages and coaxial packages, whose products are used in optical communications, optoelectronic instruments, optoelectronic sensors, etc.", the electronic equipment industry entities can be identified using the named entity recognition model: semiconductor lasers, optical communications, optoelectronic instruments, and optoelectronic sensors.

[0043] S102, inputting the relationship information text set and the specified entity set into a relationship extraction model to obtain an entity relationship set;

[0044] The electronic device inputs the relationship information text set input to the named entity recognition model and the specified entity set output from the named entity recognition model into a relationship extraction model to obtain an entity relationship set. The entity relationship set includes multiple entity relationship pairs, where each entity relationship pair includes two entities and the type of relationship between the two entities. The relationship type is one of a set of preset relationship types.

[0045] In some embodiments, the preset relationship type set includes relationship types such as supply relationships and dependency relationships, which represent upstream and downstream relationships between industries or enterprises. In some embodiments, the preset relationship type set may also include relationship types such as competitive relationships, cooperative relationships, production relationships, sales relationships, and agency relationships, which represent competitive or cooperative relationships between industries or enterprises. The preset relationship type set may also include other relationship types that can represent related relationships between industries, between industries and enterprises, or between enterprises, which are not limited here.

[0046] In some embodiments, the entity-relationship pair can be stored in the form of a triple, where two elements of the triple are two entities identified as having a relationship by the relationship extraction model, and the other element of the triple is the type of relationship between the two entities. For example, the entity-relationship set may include an entity-relationship pair: [semiconductor laser, industry term, optical communication, industry term, dependency relationship].

[0047] Here, "semiconductor laser, industry term" is an element of a triple, where "industry term" indicates the entity type of the entity "semiconductor laser." In some embodiments, if the specified entity set only includes "industry term" as the entity type, the entity element of the triple may not include data indicating the entity type of that entity. However, if the specified entity set includes not only "industry term" but also "company name," the entity element of the triple will include data indicating the entity type of that entity to facilitate differentiation.

[0048] S103, determining entity frequency features based on the specified entity set, the entity relationship set, and the knowledge corpus;

[0049] The entity frequency feature is used to represent the frequency of entities and entity relationship pairs in the entity relationship set appearing in the knowledge corpus.

[0050] In some embodiments, the entity frequency feature may include a word frequency feature and a co-occurrence feature.

[0051] The word frequency feature is used to represent the number of times an industry term in an entity relationship set appears in the knowledge corpus. For example, the word frequency feature may include the number of companies covered by each industry term (i.e., the number of entity relationship pairs that include the industry term and a company name); another example is the number of relationships between each industry term (i.e., the number of entity relationship pairs that include the industry term). The word frequency feature may also include other data indicating the number of times an industry term appears in the knowledge corpus, which is not limited here.

[0052] The co-occurrence feature is used to represent the number of times an entity-relationship pair co-occurs in the knowledge corpus. For example, the co-occurrence feature may include statistical information on the global co-occurrence of all entity-relationship pairs; another example is the co-occurrence count of entity-relationship pairs with different entity categories in the knowledge corpus. The co-occurrence feature may also include other data indicating the number of times an entity-relationship pair co-occurs in the knowledge corpus, which is not limited here.

[0053] It is understandable that, in addition to the word frequency feature and co-occurrence feature, in some embodiments, the entity frequency feature may also include other features used to represent the frequency of occurrence of entities and entity relationships in the knowledge corpus, which is not limited here.

[0054] S104, inputting the entity relationship set and the entity frequency feature into the industry chain relationship strength prediction model to obtain an entity relationship strength set;

[0055] The electronic device inputs the entity relationship set output by the relationship extraction model and the entity frequency characteristics calculated by the electronic device into the industry chain relationship strength prediction model to obtain an entity relationship strength set. This entity relationship strength set includes multiple entity-pair strength relationships. Each entity-pair strength relationship includes two entities, the relationship type between the two entities, and the relationship strength value between the two entities.

[0056] Compared with the entity relationship pairs in the entity relationship set, the entity pair strength relationships in the entity relationship strength set have an additional relationship strength value.

[0057] In some embodiments, if a quaternary entity-pair strength relationship is represented by a quaternary entity-pair strength relationship, in addition to the three elements of the entity-pair relationship, the quaternary entity-pair strength relationship also includes a relationship strength value element. For example, the entity-pair strength relationship set may include an entity-pair strength relationship as follows: [semiconductor laser, industry term, optical communication, industry term, dependency relationship, X]. X is the relationship strength value, which can vary within a preset range for different entity-pair strength relationships.

[0058] In some embodiments, if the industry chain relationship strength prediction model normalizes the relationship strength value, the relationship strength value in all entity pair strength relationships can be between 0 and 1, and a larger value indicates a higher relationship strength.

[0059] For example, the word frequency and co-occurrence characteristics of negative electrode materials and lithium-ion batteries are relatively prominent, and the related companies have a certain scale and strong R&D capabilities and innovation capabilities. The industrial chain relationship strength prediction model predicts an entity pair strength relationship as [negative electrode materials, industry terms, lithium-ion batteries, industry terms, dependency, 0.95], where the relationship strength value is 0.95, indicating that the model determines that the correlation strength of the upstream and downstream relationships in the industrial chain is very high.

[0060] S105. Upon receiving an instruction to generate an industrial chain graph including a first entity word, generate a first industrial chain graph corresponding to the first entity word based on the entity relationship strength set and the first entity word.

[0061] After obtaining the entity relationship strength set, if the electronic device receives an instruction to generate an industrial chain graph including a first entity word, it can generate a first industrial chain graph corresponding to the first entity word based on the entity relationship strength set and the first entity word.

[0062] In the first industrial chain map, the line connecting the first entity word and the second entity word shows that the first entity word and the second entity word are in the first entity pair strength relationship in the entity relationship strength set, and the relationship strength value in the first entity pair strength relationship is greater than the preset first strength threshold.

[0063] It can be understood that, with the first entity word as the center, entities with a relationship strength value greater than a preset first strength threshold can be displayed. Assuming these entities are called first-level entities, then based on these first-level entities, entity pairs with relationship strength values ​​greater than a certain strength threshold can be displayed. This can be extended in this way.

[0064] In this first industry chain map, when expanding each layer, only the entity pair strength relationships with relationship strength values ​​greater than a preset strength threshold are displayed. This can more clearly identify industry terms with strong associations, thereby facilitating the determination of the relationships between industry terms in the industry chain map. The specific number of layers to be expanded can be set by default or specified by the user, and is not limited here. Similarly, the preset strength threshold corresponding to the expansion of each layer can be set by default or specified by the user. The preset strength thresholds corresponding to each layer can be the same or different, and are not limited here.

[0065] In the embodiment of the present application, the relationship types between entities related to industry terms are extracted through a relationship extraction model, and the relationship strength values ​​between these entities are determined through an industry chain relationship strength prediction model. Finally, when an instruction is received to generate an industry map for an industry word, entity words with relationship strength values ​​greater than a preset first threshold value and the industry word can be connected and expanded layer by layer to obtain the industry chain map where the industry word is located. By deeply mining industry and enterprise relationship information, the accuracy and visibility of intelligent mining of the industry chain are greatly improved, and the efficiency of obtaining effective information in enterprise management decision-making is improved.

[0066] In the above embodiments, electronic devices deeply mine information using multiple artificial intelligence models. The initial framework of each model can be directly adopted from an open source artificial intelligence model, or it can be obtained by combining multiple open source artificial intelligence models. Then, based on the specific functional requirements of each model, corresponding data training is performed to obtain the aforementioned named entity recognition model, relationship extraction model, and industry chain relationship strength prediction model.

[0067] For ease of understanding, the following briefly describes the functions and training methods of the named entity recognition model, relationship extraction model, and industry chain relationship strength prediction model in the embodiments of this application:

[0068] Figure 2A This is a schematic diagram for understanding the functions and training methods of the named entity recognition model in the embodiment of this application.

[0069] like Figure 2A As shown in FIG, after inputting the relationship information text set into the named entity recognition model, the specified type of entity can be output, such as industry terms, or industry terms and company names.

[0070] The named entity recognition model can be obtained by training the initial framework of the named entity recognition model using training data consisting of a large amount of original text and corresponding text with entity annotation completed.

[0071] For example, in some embodiments, the BERT classification model can be used as the initial framework of the named entity recognition model. In some embodiments, the BERT classification model and the CRF model can be combined to form the initial framework of the named entity recognition model. Other entity recognition model frameworks can also be used, which are not limited here.

[0072] Figure 2B This is a schematic diagram for understanding the functions and training methods of the relationship extraction model in the embodiment of this application.

[0073] like Figure 2BAs shown in the figure, after the named entity recognition model's input relation information text set and output industry terms and company names are fed into the relation extraction model, an entity relationship set is output. This entity relationship set includes multiple identified entity relationship pairs. An entity relationship pair can be represented by a triple, such as [entity 1, category, entity 2, category, relationship type].

[0074] The relationship extraction model can be obtained by training the initial framework of the relationship extraction model after forming training data from a set of relationship information texts, industry terms, company names, and a large number of entity relationship pairs with manually labeled relationship types.

[0075] For example, in some embodiments, the initial framework of the relationship extraction model can be a BERT classification model, a convolutional neural network (CNN) model, or other relationship extraction model frameworks, which are not limited here.

[0076] Figure 2C This is a schematic diagram for understanding the functions and training methods of the industrial chain relationship strength prediction model in the embodiment of this application.

[0077] like Figure 2C As shown in Figure 1, after inputting entity frequency features and entity relationship pairs into the industry chain relationship strength prediction model, an entity relationship strength set can be output. This entity relationship strength set includes multiple predicted entity pair strength relationships. An entity pair strength relationship can be represented by a four-tuple, such as [entity 1, category, entity 2, category, relationship type, relationship strength value].

[0078] The industry chain relationship strength prediction model can be obtained by training the initial framework of the relationship extraction model after the entity frequency features, entity relationship pairs and a large number of entity pair strength relationships with manually labeled relationship strength values ​​form training data.

[0079] In some embodiments, the initial framework BERT classification model of the industrial chain relationship strength prediction model can also be a recurrent neural network RNN ​​model, or other classification prediction model frameworks, which are not limited here.

[0080] In the above embodiment, the entity types specified by the named entity recognition model may include only industry terms, or may include both industry terms and company names.

[0081] If the entity type specified for recognition is only industry terms, then both the entity relationship pairs and the entity pair strength relationships will only contain relationships between industry terms. Consequently, the corresponding industry chain graph can only be generated based on one industry term.

[0082] In some embodiments, only the relationships between industry words can be mined and established, and an industrial chain map containing only industry words can be generated. Figure 3A This is an exemplary schematic diagram of the industrial chain map generated in the embodiments of this application.

[0083] like Figure 3A As shown, if the electronic device receives an instruction to generate an industrial chain map including industry entity A, the electronic device can determine that there are three entity pair strength relationships including the industry entity A and whose relationship strength value is higher than the preset first strength threshold, wherein the other entities respectively included are: industry entity AB, industry entity AC and industry entity AD.

[0084] The electronic device can display the three entities connected to the industry entity A respectively, and display the relationship type and relationship strength value of the corresponding entity pair strength relationship on the connection line.

[0085] Furthermore, the electronic device may further expand and display the intensity relationships of the three first-layer entities:

[0086] like Figure 3A As shown, for industry entity AC, it can be determined that there are three entity-pair strength relationships that include industry entity AC and have relationship strength values ​​greater than a preset second strength threshold. These three entities each include another entity: industry entity C1, industry entity C2, and industry entity C3. The electronic device can display these three entities connected to industry entity AC, and display the relationship type and relationship strength value of the corresponding entity-pair strength relationship on the connection line.

[0087] like Figure 3A As shown, for industry entity AB, it can be determined that there is only one entity pair strength relationship that includes industry entity AB and has a relationship strength value greater than a preset second strength threshold. The other entity included is industry entity B1. The electronic device can display a line connecting industry entity B1 and industry entity AB, and display the relationship type and relationship strength value of the corresponding entity pair strength relationship on the line.

[0088] like Figure 3A As shown, for industry entity AD, it can be determined that there are five entity-pair strength relationships that include industry entity AD and have relationship strength values ​​greater than a preset second strength threshold. These five entities are industry entity D1, industry entity D2, industry entity D3, industry entity D4, and industry entity D5. The electronic device can display these five entities connected to industry entity AD, and display the relationship type and relationship strength value of the corresponding entity-pair strength relationship on the connection lines.

[0089] In this way, expanding to the second layer, it forms the following Figure 3A The industry chain map corresponding to industry entity A is shown. It is understandable that in actual applications, the entities in the second layer can be further expanded, which is not limited here.

[0090] For the industrial chain, in addition to expressing industrial nodes through industry terms, the representative enterprises of each industrial node can also be displayed in the industrial chain map. These representative enterprises can be represented by their company names in the industrial chain map.

[0091] In some embodiments, if the entity types specified by the named entity recognition model include both industry terms and company names, then the entity relationship pairs and entity pair strength relationships include relationships between industry terms, between industry terms and company names, and between company names. The generated industrial chain map will include representative companies of the industrial nodes corresponding to the industry terms, that is, there will be company names connected to them. Accordingly, a corresponding industrial chain map can be generated based on an industry term, or based on a company name.

[0092] Take the example of generating a corresponding industrial chain map based on an industry word. Figure 3B This is another exemplary schematic diagram of the industrial chain map generated in the embodiments of this application.

[0093] like Figure 3B As shown, if an electronic device receives an instruction to generate an industrial chain map that includes industry entity A, it can determine that there are four entity pair strength relationships that include industry entity A and have relationship strength values ​​greater than a preset first strength threshold. These relationships correspond to three industry word entities: industry entity AB, industry entity AC, and industry entity AD, and one enterprise name entity: enterprise entity AE. Enterprise entity AE can be a representative enterprise of industry entity A, allowing users to have a more intuitive understanding of the scope of various industries.

[0094] The electronic device can respectively connect these four entities with the industry entity A and display the relationship type and relationship strength value in the corresponding entity pair strength relationship on the connection line.

[0095] Furthermore, the electronic device may further expand and display the intensity relationships of the three first-layer entities:

[0096] like Figure 3B As shown, for industry entity AC, it can be determined that there are four entity-pair strength relationships that include industry entity AC and have relationship strength values ​​greater than a preset second strength threshold. These correspond to three industry word entities: industry entity C1, industry entity C2, and industry entity C3, and one enterprise name entity: enterprise entity C4. The electronic device can display these four entities with lines connected to industry entity AC, and display the relationship type and relationship strength value of the corresponding entity-pair strength relationships on the lines.

[0097] like Figure 3BAs shown, for industry entity AB, it can be determined that there are two entity-pair strength relationships that include industry entity AB and have a relationship strength value greater than a preset second strength threshold. These relationships correspond to an industry term entity, industry entity B1, and a business name entity, business entity B2. The electronic device can display lines connecting industry entity B1 and business entity B2 to industry entity AB, respectively, and display the relationship type and relationship strength value of the corresponding entity-pair strength relationships on the lines.

[0098] like Figure 3B As shown, for industry entity AD, it can be determined that there are seven entity-pair strength relationships that include industry entity AD and have relationship strength values ​​greater than a preset second strength threshold. These correspond to five industry word entities: industry entity D1, industry entity D2, industry entity D3, industry entity D4, and industry entity D5, and two enterprise name entities: enterprise entity D6 and enterprise entity D7. The electronic device can display these seven entities in connection with industry entity AD, and display the relationship type and relationship strength value of the corresponding entity-pair strength relationship on the connection lines.

[0099] like Figure 3B As shown, for enterprise entity AE, it can be determined that there are two entity-pair strength relationships that include industry entity AE and have a relationship strength value greater than a preset second strength threshold. These relationships correspond to two enterprise name entities: enterprise entity E1 and enterprise entity E2. The electronic device can display these two entities connected to enterprise entity AE, and the relationship type and relationship strength value of the corresponding entity-pair strength relationship can be displayed on the connection line.

[0100] In this way, expanding to the second layer, it forms the following Figure 3B The industry chain map corresponding to industry entity A is shown. It is understandable that in actual applications, the entities in the second layer can be further expanded, which is not limited here.

[0101] Take the example of generating a corresponding industrial chain map based on a company name. Figure 3C This is another exemplary schematic diagram of the industrial chain map generated in the embodiments of this application.

[0102] like Figure 3C As shown, if an electronic device receives an instruction to generate an industrial chain map including enterprise entity A, the electronic device may determine that there are three entity pair strength relationships including enterprise entity A and having a relationship strength value greater than a preset first strength threshold, corresponding to one industry word entity: industry entity AB, and two enterprise name entities: enterprise entity AC and enterprise entity AD. Industry entity AB may be the industry of enterprise entity A.

[0103] The electronic device may respectively connect these three entities with the enterprise entity A and display the relationship type and relationship strength value of the corresponding entity pair strength relationship on the connection line.

[0104] Furthermore, the electronic device may further expand and display the intensity relationships of the three first-layer entities:

[0105] like Figure 3C As shown, for industry entity AB, it can be determined that there are four entity-pair strength relationships that include industry entity AB and have relationship strength values ​​greater than a preset second strength threshold. These correspond to three industry word entities: industry entity B1, industry entity B2, and industry entity B3, and one enterprise name entity: enterprise entity B4. The electronic device can display these four entities connected to industry entity AB, and the relationship types and relationship strength values ​​of the corresponding entity-pair strength relationships are displayed on the connection lines.

[0106] like Figure 3C As shown, for business entity AC, it can be determined that there are three entity-pair strength relationships that include business entity AC and have relationship strength values ​​greater than a preset second strength threshold. These correspond to two industry-term entities: industry entity C1 and industry-term entity C2, and one business name entity: business entity C3. The electronic device can display these three entities connected to business entity AC, and display the relationship type and relationship strength value of the corresponding entity-pair strength relationship on the connection lines.

[0107] like Figure 3C As shown, for business entity AD, it can be determined that there are three entity-pair strength relationships that include business entity AD and have relationship strength values ​​greater than a preset second strength threshold. These correspond to one industry term entity: industry entity D1, and two business name entities: business entity D2 and business entity D3. The electronic device can display these three entities in a line connection with business entity AD, and display the relationship type and relationship strength value of the corresponding entity-pair strength relationship on the line connection.

[0108] In this way, expanding to the second layer, it forms the following Figure 3C The industry chain map corresponding to the enterprise entity A is shown. It is understandable that in actual applications, the entities in the second layer can be further expanded, which is not limited here.

[0109] In some embodiments, when determining an entity pair strength relationship with a relationship strength value higher than a preset second strength threshold during relationship expansion of entities in a next layer, the entity pair strength relationship including entities in a previous layer may be ignored.

[0110] In some embodiments, if there are entity pair strength relationships on different branches of the graph with relationship strength values ​​higher than a preset strength threshold, they can also be connected and displayed. Figure 2B Taking the industrial chain diagram shown above as an example, if there is an entity pair strength relationship between industry entity AD and enterprise entity B2, and the relationship strength value is greater than a preset strength threshold (e.g., preset strength threshold A), a line connecting industry entity AD and enterprise entity B2 can be displayed, and the relationship type and relationship strength value of the corresponding entity pair strength relationship can be displayed on the line. In some embodiments, this can be omitted to simplify the diagram, and this is not limited here.

[0111] It is understandable that the first strength threshold and the second strength threshold can be the same or different. In some embodiments, the second strength threshold can be higher than the first strength threshold. And as the number of expansion layers increases, the strength threshold corresponding to each layer can also increase layer by layer. For example, if the strength relationship of the second layer entity is expanded to the third layer, it will be determined that the entity in the second layer of the industry chain map (for example Figure 3A For an entity-to-entity strength relationship (industry entity B1) with a relationship strength value exceeding a preset third strength threshold, the preset third strength threshold is higher than the second strength threshold. Simply setting the strength thresholds to increase layer by layer eliminates the need to limit the number of layers, and the number of layers in the industry chain graph will naturally converge.

[0112] It should be noted that, for ease of description and understanding, the above example only expands the entities in the first layer and assumes that very few entity-pair strength relationships meet the preset strength threshold. In actual applications, this is generally expanded to the third layer, and can be expanded to further layers based on user needs. Furthermore, for each entity, the number of entity-pair strength relationships that meet the preset strength threshold can be quite large, for example, more than ten or even thirty or forty, though this is not a limitation here.

[0113] The following combination Figure 3B and Figure 3C The exemplary industrial chain map shown is taken as an example, in which the entity types specified by the named entity recognition model include industry terms and company names, to specifically describe the method for intelligently generating the industrial chain map in the embodiment of this application:

[0114] Figure 4 This is another flow chart of the method for intelligently generating an industrial chain map in an embodiment of this application.

[0115] S401, using a named entity recognition model to identify a relationship information text set to obtain a specified entity set;

[0116] The relationship information text set is obtained by extracting text content representing relationship information between enterprises and / or industries from a knowledge corpus storing enterprise information. The entities in the specified entity set include enterprise names and industry terms in the relationship information text set.

[0117] S402: Input the relationship information text set and the specified entity set into a relationship extraction model to obtain an entity relationship set;

[0118] The entity relationship set includes multiple entity relationship pairs, and one entity relationship pair includes two entities and the relationship type between the two entities.

[0119] S403: Determine entity frequency features and enterprise relevance features based on the specified entity set, the entity relationship set, and the knowledge corpus;

[0120] The entity frequency feature is used to represent the frequency of occurrence of entities and / or entity relationship pairs in the knowledge corpus, and the enterprise relevance feature is used to represent the degree of association between enterprises and industries and / or between enterprises.

[0121] The enterprise relevance characteristics may include data such as similarity between enterprises, enterprise popularity, enterprise related exposure, and the number of enterprises covered by industry terms in various regions.

[0122] S404: Input the entity relationship set, the entity frequency feature, and the enterprise correlation feature into an industry chain relationship strength prediction model to obtain an entity relationship strength set;

[0123] The entity relationship strength set includes multiple entity pair strength relationships, and one entity pair strength relationship includes two entities, the relationship types of the two entities, and the relationship strength values ​​of the two entities.

[0124] Steps S401~S404 and Figure 1 In the illustrated embodiment, steps S101 to S104 are similar. For common features, please refer to the description of steps S101 to S104 and will not be repeated here.

[0125] S405: upon receiving an instruction to generate an industrial chain graph including a first entity word, determining a plurality of first-level entity pair strength relationships in the entity relationship strength set including the first entity word and having a relationship strength value greater than a preset first strength threshold;

[0126] S406 , with the first entity word as the center, display lines connecting another entity in each first-level entity pair strength relationship with the first entity word, and displaying corresponding relationship types and relationship strength values ​​on each line;

[0127] S407: For any second entity word in a first-level entity pair strength relationship, determine a plurality of second-level entity pair strength relationships in the entity relationship strength set that include the second entity word and have a relationship strength value greater than a preset second strength threshold;

[0128] S408 : Display another entity in each second-level entity pair strength relationship with the corresponding second entity word by connecting lines, and display the corresponding relationship type and relationship strength value on each connecting line.

[0129] Please refer to the above Figure 3B and Figure 3C The steps S405 to S408 can be understood by referring to the description of the generation process of the industrial chain map shown in FIG.

[0130] It is understandable that the third-level entity pair strength relationship can be expanded based on the second-level entity pair strength relationship, and the fourth-level entity pair strength relationship can be expanded based on the third-level entity pair strength relationship, and so on, and all of them can be connected and displayed in the industry map. The specific number of layers can be based on user settings and is not limited here.

[0131] In some embodiments, in order to make it easier for users to understand the industrial chain map, the entity pair strength relationship corresponding to the relationship type representing the upstream relationship of the industrial chain can be displayed in the upper part of the display screen, and the entity pair strength relationship corresponding to the relationship type representing the downstream relationship of the industrial chain can be displayed in the lower part of the display screen.

[0132] In the embodiment of the present application, since the specified entity set includes both industry terms and company names, the relationship between enterprises and industries can be deeply explored. Therefore, not only can representative enterprises of industry nodes be displayed in the formed industrial chain map, but the industrial chain map corresponding to the company name can also be directly generated, which greatly improves the accuracy of the information displayed in the industrial chain map.

[0133] In addition, when expanding entities layer by layer in the process of generating the industrial chain map, the preset strength threshold for determining the strength relationship of entity pairs in the next layer can be set higher than the preset strength threshold for determining the strength relationship of entity pairs in the previous layer. This allows the generation process of the industrial chain map to converge naturally, thereby improving the integrity and accuracy of the industrial chain map information and the generation efficiency of the industrial chain map.

[0134] The above describes the method for intelligently generating an industrial chain map in an embodiment of the present application. The following introduces an exemplary electronic device 500 provided in an embodiment of the present application.

[0135] Figure 5 It is a schematic diagram of an exemplary structure of an electronic device 500 provided in an embodiment of the present application.

[0136] In some embodiments, the electronic device 500 is a computer device, which may be a terminal device or a server, and includes a processor, a memory, and a network interface connected via a system bus.

[0137] The processor of the computer device is used to provide computing and control capabilities. The processor may include one or more processing units, for example, the processor may include one or more of an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU). Different processing units may be independent devices or integrated into one or more processors.

[0138] If the processor includes an NPU as one of its processing units, it can help improve the efficiency of deep learning processing. The NPU is a neural network (NN) computing processor that quickly processes input information by drawing on the structure of biological neural networks, such as the transmission mode between neurons in the human brain, and can also continuously self-learn. The NPU can realize intelligent cognition applications such as image recognition, face recognition, voice recognition, and text comprehension in the electronic device 500.

[0139] The computer device's memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operating system and computer program in the non-volatile storage medium to run. The computer device's database is used to store data.

[0140] The network interface of the computer device is used for communicating with other external terminals or servers via a network connection. In some embodiments, the network interface can be a wired network interface, and in some embodiments, the network interface can also be a wireless network interface.

[0141] When the computer program is executed by the processor, it implements the method for intelligently generating the industrial chain map in the embodiment of the present application.

[0142] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0143] In some embodiments of the present application, a computer-readable storage medium is also provided, including instructions. When the instructions are executed on the electronic device 500, the electronic device 500 can execute the method for intelligently generating an industrial chain map in the embodiment of the present application.

[0144] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0145] As used in the above embodiments, the term “when” may be interpreted to mean “if” or “after” or “in response to determining that” or “in response to detecting that”, depending on the context. Similarly, the phrases “upon determining that” or “if (stated condition or event) is detected” may be interpreted to mean “if determining that” or “in response to determining that” or “upon detecting (stated condition or event)” or “in response to detecting (stated condition or event)”, depending on the context.

[0146] In the above embodiments, all or part of the embodiments can be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., a floppy disk, hard disk, tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive).

[0147] Those skilled in the art will appreciate that all or part of the process steps in the above-described method embodiments can be implemented by a computer program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and when executed, the program can include the process steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for intelligently generating an industrial chain map, characterized in that: include: S101. Using a named entity recognition model, identify a set of relationship information texts to obtain a designated entity set; the set of relationship information texts is obtained by extracting text content representing relationship information between industries from a knowledge corpus storing enterprise information, and the entities in the designated entity set include industry terms and enterprise names in the set of relationship information texts; S102: Input the relationship information text set and the specified entity set into a relationship extraction model to obtain an entity relationship set; the entity relationship set includes multiple entity relationship pairs, and an entity relationship pair includes two entities and the relationship type between the two entities; S103. Determine entity frequency features and enterprise relevance features based on the specified entity set, the entity relationship set, and the knowledge corpus, wherein the entity frequency features are used to represent the frequency of occurrence of entity and entity relationship pairs in the knowledge corpus; The enterprise relevance feature is used to represent the degree of association between an enterprise and an industry and / or between enterprises; The enterprise relevance features include: similarity between enterprises corresponding to enterprise names, popularity of enterprises corresponding to enterprise names, exposure of enterprises corresponding to enterprise names, and the number of enterprises covered by the words in each region and industry; S104: Input the entity relationship set, the entity frequency feature, and the enterprise relevance feature into an industry chain relationship strength prediction model to obtain an entity relationship strength set; the entity relationship strength set includes multiple entity pair strength relationships, and an entity pair strength relationship includes two entities, the relationship type of the two entities, and the relationship strength values ​​of the two entities; S105. Upon receiving an instruction to generate an industrial chain graph including a first entity word, generate a first industrial chain graph corresponding to the first entity word based on the entity relationship strength set and the first entity word; in the first industrial chain graph, the first entity word and the second entity word are displayed as a connected line, the first entity word and the second entity word are in a first entity pair strength relationship in the entity relationship strength set, and the relationship strength value in the first entity pair strength relationship is greater than a preset first strength threshold.

2. The method according to claim 1, characterized in that The enterprise relevance features include: the similarity between enterprises corresponding to the enterprise names, the popularity of enterprises corresponding to the enterprise names, the associated exposure of enterprises corresponding to the enterprise names, and the number of enterprises covered by the words in each region and industry.

3. The method according to claim 1, characterized in that The entity frequency features include: word frequency features and co-occurrence features. The word frequency features are used to represent the number of times industry words in the entity relationship set appear in the knowledge corpus, and the co-occurrence features are used to represent the number of times entity relationship pairs appear together in the knowledge corpus.

4. The method according to claim 1, wherein The relationship type is one of a preset relationship type set, which includes multiple relationship types representing upstream and downstream relationships between industries or enterprises, and multiple relationship types representing competitive or cooperative relationships between industries or enterprises.

5. The method according to any one of claims 1 to 4, characterized in that In the case where the industrial chain graph is set to be expanded only by two layers, upon receiving an instruction to generate an industrial chain graph including a first entity word, generating a first industrial chain graph corresponding to the first entity word based on the entity relationship strength set and the first entity word specifically includes: Upon receiving an instruction to generate an industrial chain graph including a first entity word, determining a plurality of first-level entity pair strength relationships in the entity relationship strength set including the first entity word and having a relationship strength value greater than a preset first strength threshold; With the first entity word as the center, display lines connecting another entity in each first-level entity pair strength relationship with the first entity word, and display the corresponding relationship type and relationship strength value on each line; For any second entity word in a first-level entity pair strength relationship, determining a plurality of second-level entity pair strength relationships in the entity relationship strength set that include the second entity word and have a relationship strength value greater than a preset second strength threshold; Another entity in each second-layer entity pair strength relationship is displayed with a corresponding second entity word, and the corresponding relationship type and relationship strength value are displayed on each connection line to form a first industrial chain map corresponding to the first entity word.

6. The method according to claim 5, characterized in that The preset second intensity threshold is greater than the preset first intensity threshold.

7. An electronic device, characterized in that: include: one or more processors and memory; The memory is coupled to the one or more processors, and is configured to store computer program codes, where the computer program codes include computer instructions. The one or more processors call the computer instructions to enable the electronic device to execute the method according to any one of claims 1 to 6.

8. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on an electronic device, the electronic device is caused to execute the method according to any one of claims 1 to 6.

Images