A Method for Constructing Industry and Enterprise Profiles Based on Public Opinion Events

Abstract

This invention provides a method for constructing industry and enterprise profiles based on public opinion events, comprising: correcting public opinion entities and entity relationships according to historical event records in an event detail table to obtain corrected public opinion entities and corrected entity relationships; constructing real-time event records of real-time public opinion events based on the corrected public opinion entities and corrected entity relationships; extracting industry data based on historical and real-time event records to obtain an industry detail table, and extracting enterprise events based on historical and real-time event records to obtain an enterprise event table; constructing industry profiles based on the industry detail table, constructing enterprise profiles based on the enterprise event table, and providing services to target enterprises based on the industry and enterprise profiles. This method improves the accuracy of the constructed enterprise profiles and enhances the suitability of the provided services for the enterprise.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for constructing industry profiles and enterprise profiles based on public opinion events. Background Technology

[0002] In the process of providing services to enterprises, it is often necessary to use enterprise profiles to provide personalized services. However, in the process of building enterprise profiles, relevant data about the enterprise is often directly extracted from the collected data and then filled into the enterprise profile. Although this method can obtain an enterprise profile, the accuracy of the enterprise profile is low because the names and terms of the collected data may be different and the collected data is not classified. At the same time, providing services to enterprises based solely on enterprise profiles may result in low service suitability for the enterprise. Summary of the Invention

[0003] This invention provides a method for constructing industry profiles and enterprise profiles based on public opinion events, in order to solve the technical problems of inaccurate enterprise profiles and incompatible services provided by existing technologies. This method improves the accuracy of the constructed enterprise profiles and enhances the compatibility of the provided services with the enterprises.

[0004] This invention provides a method for constructing industry profiles and enterprise profiles based on public opinion events, including:

[0005] Entity identification is performed on real-time public opinion events to obtain the public opinion entities included in the real-time public opinion events, and relationship extraction is performed on the real-time public opinion events to obtain the entity relationships between the public opinion entities;

[0006] Based on the historical event records in the event details table, the public opinion entity and the entity relationship are corrected to obtain the corrected public opinion entity and the corrected entity relationship. Each historical event record in the event details table corresponds to a historical public opinion event. The historical event record includes at least one of the following fields: event identifier, company name, full name of associated company, industry level four classification, product term, event subject, activity, event object, event occurrence time, and event source.

[0007] Based on the corrected public opinion entities and the corrected entity relationships, a real-time event record of the real-time public opinion event is constructed;

[0008] Based on the historical event records and the real-time event records, industry data is extracted to obtain an industry detail table, and based on the historical event records and the real-time event records, enterprise events are extracted to obtain an enterprise event table.

[0009] An industry profile is constructed based on the industry details table, an enterprise profile is constructed based on the enterprise event table, and services are provided to the target enterprise based on the industry profile and the enterprise profile.

[0010] According to a method for constructing industry and enterprise profiles based on public opinion events provided by the present invention, the step of correcting the public opinion entity and the entity relationship according to the historical event records recorded in the event details table to obtain the corrected public opinion entity and the corrected entity relationship includes: based on the public opinion entity, performing entity retrieval from the event details table to obtain the entity name corresponding to the public opinion entity in the event details table, and updating the public opinion entity to the entity name as the corrected public opinion entity; based on the entity relationship, performing relationship retrieval from the event details table to obtain the relationship name of the entity relationship in the event details table, and updating the entity relationship to the relationship name as the corrected entity relationship.

[0011] According to a method for constructing industry profiles and enterprise profiles based on public opinion events provided by the present invention, the step of constructing real-time event records of the real-time public opinion events based on the modified public opinion entities and the modified entity relationships includes: when the row sequence corresponding to the entity name in the event details table is the same as the row sequence corresponding to the relationship name, the event record corresponding to the row sequence corresponding to the entity name is taken as the real-time event record of the real-time public opinion event.

[0012] According to a method for constructing industry and enterprise profiles based on public opinion events provided by the present invention, the step of extracting industry data based on the historical event records and the real-time event records to obtain an industry detail table includes: extracting historical industry data of at least one dimension from the historical event records, wherein the dimensions of the historical industry data include historical industry policy dimension, historical industry supply and demand dynamics dimension, historical industry price fluctuation dimension, historical industry competitive landscape dimension, and historical industry development dimension; quantifying the historical industry data to obtain first quantified data; and determining the industry detail table based on the first quantified data and the real-time event records.

[0013] According to a method for constructing industry and enterprise profiles based on public opinion events provided by the present invention, the step of determining an industry detail table based on the first quantitative data and the real-time event records includes: constructing a candidate industry detail table according to the dimensions to which the first quantitative data belongs; extracting real-time industry data of at least one dimension from the real-time event records, wherein the dimensions of the real-time industry data include real-time industry policy dimension, real-time industry supply and demand dynamics dimension, real-time industry price fluctuation dimension, real-time industry competitive landscape dimension, and real-time industry development dimension; quantifying the real-time industry data to obtain second quantitative data; and updating the first quantitative data in the candidate industry detail table according to the dimensions to which the second quantitative data belongs to obtain the industry detail table.

[0014] According to the present invention, a method for constructing industry profiles and enterprise profiles based on public opinion events is provided. The step of extracting enterprise events based on historical event records and real-time event records to obtain an enterprise event table includes: extracting historical enterprise data of at least one dimension from the historical event records, wherein the dimensions of the historical enterprise data include historical business activity, historical compliance level, historical financial status, historical risk level, historical development stage, and historical service demand; quantifying the historical enterprise data to obtain third quantified data; and determining the enterprise event table based on the third quantified data and the real-time event records.

[0015] According to a method for constructing industry and enterprise profiles based on public opinion events provided by the present invention, the step of determining an enterprise event table based on the third quantitative data and the real-time event records includes: constructing a candidate enterprise event table according to the dimensions to which the third quantitative data belongs; extracting real-time enterprise data of at least one dimension from the real-time event records, wherein the dimensions of the real-time enterprise data include real-time business activity dimension, real-time compliance level dimension, real-time financial status dimension, real-time risk level dimension, real-time development stage dimension, and real-time service demand dimension; quantifying the real-time enterprise data to obtain fourth quantitative data; and updating the third quantitative data in the candidate enterprise event table according to the dimensions to which the fourth quantitative data belongs to obtain the enterprise event table.

[0016] According to the present invention, a method for constructing industry profiles and enterprise profiles based on public opinion events, wherein providing services to target enterprises based on the industry profiles and enterprise profiles includes: extracting target industry data of at least one dimension from the industry profile, wherein the dimensions of the target industry data include industry policy dimension, industry supply and demand dynamics dimension, industry price fluctuation dimension, industry competitive landscape dimension, and industry development degree dimension; extracting target enterprise data of at least one dimension from the enterprise profile, wherein the dimensions of the target enterprise data include business activity dimension, compliance level dimension, financial status dimension, risk level dimension, development stage dimension, and service demand dimension; and providing services to the target enterprises based on the target industry data and the target enterprise data.

[0017] According to a method for constructing industry and enterprise profiles based on public opinion events provided by the present invention, the step of providing services to the target enterprise based on the target industry data and the target enterprise data includes: determining the enterprise type of the target enterprise based on the target industry data and the target enterprise data; when the enterprise type indicates that the target enterprise is a shrinking enterprise, providing risk services to the target enterprise, the risk services including legal consulting services, debt optimization services, operational compliance services, and environmental rectification services; when the enterprise type indicates that the target enterprise is an expanding enterprise, providing expansion services to the target enterprise, the expansion services including subsidy application services and industry fund matching services.

[0018] According to the present invention, a method for constructing industry profiles and enterprise profiles based on public opinion events is provided. The step of providing services to target enterprises based on the industry profiles and enterprise profiles includes: extracting real-time industry information from the industry profiles and extracting real-time enterprise information of the target enterprise from the enterprise profiles; when the real-time industry information is trending information and the real-time enterprise information indicates that the target enterprise is an enterprise with idle production capacity, pushing instant services to the target enterprise, wherein the trending information includes at least one of policy implementation, commodity price fluctuation exceeding a change threshold, and target orders exceeding an order threshold being signed; the instant services include at least one of subsidy application services, order matching services, and financial credit services; when the real-time industry information is risk information and the real-time enterprise information indicates that the target enterprise is a risky enterprise, pushing risk control services to the target enterprise, wherein the risk control services include industrial transformation consulting services, production reduction suggestion services, and environmental rectification suggestion services.

[0019] This invention also provides an apparatus for constructing industry profiles and enterprise profiles based on public opinion events, comprising the following modules:

[0020] The identification module is used to perform entity identification on real-time public opinion events, obtain the public opinion entities included in the real-time public opinion events, and extract the relationships between the public opinion entities.

[0021] The correction module is used to correct the public opinion entity and the entity relationship according to the historical event records recorded in the event details table, so as to obtain the corrected public opinion entity and the corrected entity relationship. Each historical event record in the event details table corresponds to a historical public opinion event. The historical event record includes at least one of the following fields: event identifier, company name, full name of associated company, industry four-level classification, product term, event subject, activity, event object, event occurrence time, and event source.

[0022] A construction module is used to construct a real-time event record of the real-time public opinion event based on the corrected public opinion entity and the corrected entity relationship;

[0023] The extraction module is used to extract industry data based on the historical event records and the real-time event records to obtain an industry detail table, and to extract enterprise events based on the historical event records and the real-time event records to obtain an enterprise event table.

[0024] The construction module is also used to construct an industry profile based on the industry details table, construct an enterprise profile based on the enterprise event table, and provide services to the target enterprise based on the industry profile and the enterprise profile.

[0025] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for constructing industry profiles and enterprise profiles based on public opinion events as described above.

[0026] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for constructing industry profiles and enterprise profiles based on public opinion events as described above.

[0027] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the method for constructing industry profiles and enterprise profiles based on public opinion events as described above.

[0028] This invention provides a method for constructing industry and enterprise profiles based on public opinion events. First, it extracts public opinion entities and entity relationships from text, transforming unstructured real-time public opinion events into structured data. Then, by referring to historical event records in the event details table, the identified public opinion entities and entity relationships are corrected, ensuring consistency and coherence between newly extracted knowledge and historical knowledge. Based on the corrected entities and relationships, real-time event records are constructed, and this structured description is stored in the event details table. Based on the historical and real-time event records, industry details tables and enterprise event tables are extracted and generated, transforming the original events into industry and enterprise data categorized by dimensions. Next, industry profiles are constructed based on the industry details tables, and enterprise profiles are constructed based on the enterprise event tables. This enables a three-dimensional and quantitative model of the industry's current status, trends, and multi-dimensional characteristics of enterprises, improving the accuracy of the constructed enterprise and industry profiles. Finally, based on the industry and enterprise profiles, service strategies suitable for the current scenario can be accurately matched, providing targeted business decision-making assistance or risk response services to target enterprises, improving the adaptability of the services provided to the enterprises. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0030] Figure 1 This is a schematic diagram of the first process of a method for constructing industry profiles and enterprise profiles based on public opinion events provided by the present invention.

[0031] Figure 2 This is a schematic diagram of the second process of a method for constructing industry profiles and enterprise profiles based on public opinion events provided by the present invention.

[0032] Figure 3 This is a schematic diagram of the third process of a method for constructing industry profiles and enterprise profiles based on public opinion events provided by the present invention.

[0033] Figure 4 This is a schematic diagram of the fourth process of a method for constructing industry profiles and enterprise profiles based on public opinion events provided by the present invention.

[0034] Figure 5 This is a schematic diagram of a device for constructing industry profiles and enterprise profiles based on public opinion events, provided by the present invention.

[0035] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0037] The following is combined with Figures 1-4 This invention describes a method for constructing industry and enterprise profiles based on public opinion events.

[0038] Figure 1 This is a schematic diagram of the first process of a method for constructing industry profiles and enterprise profiles based on public opinion events provided by the present invention, as shown below. Figure 1 As shown, the method includes the following:

[0039] In step 101, entity recognition is performed on the real-time public opinion events to obtain the public opinion entities included in the real-time public opinion events, and relationship extraction is performed on the real-time public opinion events to obtain the entity relationships between the public opinion entities.

[0040] As an example, public opinion entities can be named entities that can be identified from the unstructured text of real-time public opinion events. The types of public opinion entities include at least company names, personal names, product terms, industry category names, and region names. For example, in the real-time public opinion event "Company G announces its entry into the new energy vehicle battery field", the identified public opinion entities are "Company G" and "new energy vehicle battery".

[0041] Entity relationships can be semantic associations between two public opinion entities in a real-time public opinion event. Entity relationships are used to describe the behaviors or states that occur between public opinion entities. The types of entity relationships can include investment relationships, cooperation relationships, acquisition relationships, competition relationships, penalty relationships, raw material supply relationships, etc. For example, based on the above public opinion entities "Company G" and "new energy vehicle battery", the extracted entity relationship is "entry / deployment", which represents an expansionary business action.

[0042] As an example, the server can first input the text content of real-time public opinion events into a pre-trained entity recognition model. The entity recognition model can adopt a sequence labeling architecture to predict the label of each character or word in the text. The label identifies the position of the character or word in the public opinion entity (beginning, internal or end position). The server combines the predicted continuous label segments to obtain the text segment of the public opinion entity and its corresponding entity type identifier. For example, the server labels "Company G" as a public opinion entity of the organization name type and "new energy vehicle battery" as a public opinion entity of the product word type.

[0043] Then, the server can combine all the identified public opinion entities into pairs to form several candidate entity pairs. For each candidate entity pair, the server locates the sentence fragments in the real-time public opinion event text that contain both public opinion entities. The server inputs the location and type information of the located sentence fragments and candidate entity pairs into a pre-trained relation extraction model. The relation extraction model can determine the semantics of the sentence based on the attention mechanism and output the probability value of the candidate entity pair belonging to each predefined entity relation. The server selects the entity relation with the highest probability value that exceeds the confidence threshold as the entity relation extracted from the real-time public opinion event, such as "Company G - Entering - New Energy Vehicle Batteries".

[0044] In step 102, the public opinion entity and the entity relationship are corrected according to the historical event records in the event details table to obtain the corrected public opinion entity and the corrected entity relationship. Each historical event record in the event details table corresponds to a historical public opinion event. The historical event record includes at least one of the following fields: event identifier, company name, full name of associated company, industry four-level classification, product term, event subject, activity, event object, event occurrence time, and event source.

[0045] As an example, the event details table can be a structured data table maintained by the server in the database to persistently store processed historical public opinion event information. Each row of the event details table corresponds to a historical event record. For example, the values ​​of each field of a historical event record in the event details table may include event identifier (E20250101001), company name (A Technology Co., Ltd.), full name of associated company (B New Energy Co., Ltd.), industry level four classification (lithium-ion battery manufacturing), product term (ternary lithium battery), event subject (A Technology Co., Ltd.), activity (capacity expansion), event object (ternary lithium battery production line), event occurrence time (2025-01-01 10:00:00), and event source (National Enterprise Credit Information Publicity System).

[0046] The corrected public opinion entity can be the entity name obtained by standardizing and aligning the text of the public opinion entity extracted from the real-time public opinion event according to the existing historical event records in the event details table. For example, the public opinion entity "Company A" extracted from the real-time public opinion event is corrected to the company name "A Technology Co., Ltd." that already exists in the event details table.

[0047] The revised entity relationship can be the entity relationship extracted from real-time public opinion events, and the relationship name obtained by unifying the corresponding field's standardized expression in the existing historical event records in the event details table. For example, the entity relationship "reached cooperation" extracted from real-time public opinion events is revised to the standardized expression "established strategic cooperative relationship" already used in the activity field of the event details table.

[0048] As an example, firstly, for each public opinion entity, the server can initiate a search in multiple specified fields of the event details table using the text of the public opinion entity as the query condition. The specified fields include at least the company name, the full name of the associated company, and product terms. The server can calculate the text similarity between the text of the public opinion entity and the text of the stored values ​​in each specified field. When the text similarity exceeds the preset entity matching threshold, the server determines that the search has been successful, extracts the normalized entity name stored in the corresponding field of the event details table, and replaces the original public opinion entity text with the normalized entity name to obtain the corrected public opinion entity.

[0049] For each entity relationship, the server can use the text of that entity relationship as the query condition to search within the active field of the event details table. The server determines the semantic similarity between the entity relationship text and each existing value in the active field. When the semantic similarity exceeds the preset relationship matching threshold, the server extracts the corresponding existing value in the active field as the corrected entity relationship.

[0050] As an example, when correcting public opinion entities, the server can introduce an industry-level four-tier classification field as an auxiliary disambiguation method. Specifically, when multiple candidate records of highly similar public opinion entities are retrieved in the event details table, the server can further obtain the contextual industry keywords of the real-time public opinion event in which the public opinion entity is located, and match them with the industry-level four-tier classification field values ​​corresponding to each candidate record. If the similarity between the industry-level four-tier classification of the candidate record and the contextual industry keywords exceeds the similarity threshold, the server will prioritize selecting the standardized entity name in the candidate record as the corrected public opinion entity. In this way, the problem of confusion between homophones or similar company names in different industries can be solved, and the accuracy of entity collection in the company profile can be improved.

[0051] By using existing historical event records in the event details table as a standardization basis, the heterogeneous public opinion entities and entity relationships from multiple sources are corrected into standardized semantic expressions, thereby achieving seamless alignment between incremental event information and historical event information. This enables enterprise profiles to form a coherent and complete view of enterprise behavior trajectory based on continuous and consistent entity and relationship data, significantly improving the long-term maintainability and data reliability of enterprise profiles.

[0052] In some embodiments, step 102, which modifies the public opinion entity and the entity relationship according to the historical event records in the event details table to obtain the modified public opinion entity and the modified entity relationship, can be implemented by the following technical solution: based on the public opinion entity, perform entity retrieval from the event details table to obtain the entity name corresponding to the public opinion entity in the event details table, and update the public opinion entity to the entity name as the modified public opinion entity; based on the entity relationship, perform relationship retrieval from the event details table to obtain the relationship name of the entity relationship in the event details table, and update the entity relationship to the relationship name as the modified entity relationship.

[0053] As an example, entity retrieval can be a server performing a matching query in an event details table to locate the corresponding record's processing action. The scope of entity retrieval can cover predefined fields in the event details table related to entity names such as enterprise, product, and brand. Entity retrieval can be used to discover standardized expressions of the same entity that already exist in historical data. For example, the server performs entity retrieval with the public opinion entity "Jia Technology" as input conditions and matches the record "Jia Technology Group Co., Ltd." in the enterprise name field of the event details table.

[0054] An entity name can be a standardized named text that matches the public opinion entity in the matching records of the event details table when the server performs entity retrieval. For example, the enterprise name field value "Jia Technology Group Co., Ltd." extracted from the matching records in the event details table is the entity name.

[0055] Relationship retrieval can be a process where the server performs a matching query in the event details table to locate the corresponding relationship description. The scope of relationship retrieval is mainly the activity field in the event details table. Relationship retrieval can be used to discover the standard naming conventions for this type of business behavior that already exist in historical data. For example, the server performs a relationship retrieval with the entity relationship "equity participation" as the input condition and matches the standard description "equity investment" in the activity field of the event details table.

[0056] The relation name can be a normalized relation description text that matches the entity relation in the matching records of the event details table when the server performs relation retrieval. For example, the activity field value "equity investment" extracted from the record hit in the event details table is the relation name.

[0057] As an example, the server first performs entity retrieval from the event details table based on the public opinion entity. Specifically, the server can iterate through the enterprise name field and the associated enterprise full name field of each historical event record in the event details table, calculate the character similarity between the text of the public opinion entity and each field, and when the character similarity exceeds the preset entity matching threshold, the server can use the value of the corresponding field in the event details table as the entity name. The server can then update the original public opinion entity with this entity name, and the updated result is the corrected public opinion entity.

[0058] Meanwhile, the server can iterate through the activity field of each historical event record in the event details table and calculate the semantic similarity between the text of the entity relationship and the value of the activity field. When the calculated semantic similarity exceeds the preset relationship matching threshold, the server can use the value of the activity field in the event details table as the relationship name. The server can update the original entity relationship to this relationship name, and the updated result is the corrected entity relationship.

[0059] As an example, when performing entity retrieval, the server can use the product term field and the industry fourth-level classification field for auxiliary verification. Specifically, when multiple candidate entity names with the same high similarity are retrieved in the enterprise name field and the full name field of the associated enterprise in the event details table, the server further obtains the product term or industry keyword in the real-time public opinion event in which the public opinion entity is located, and retrieves the product term or industry keyword in the public opinion event in the corresponding field of the event details table. The server selects the entity name in the candidate record that best matches the context of the real-time public opinion event with the product term or industry fourth-level classification as the final corrected public opinion entity.

[0060] By performing entity retrieval and relationship retrieval in the event details table based on public opinion entities and entity relationships respectively, and updating the names of public opinion entities and entity relationships with the retrieved entity names and relationship names, the problems caused by ambiguity in entity names and inconsistency in relationship descriptions are eliminated, thereby improving the data quality and analytical credibility of enterprise profiling.

[0061] In step 103, a real-time event record of the real-time public opinion event is constructed based on the corrected public opinion entity and the corrected entity relationship.

[0062] As an example, real-time event logs can be structured event data entries corresponding to real-time public opinion events. The data structure of real-time event logs can be the same as that of historical event logs in the event details table. The fields of real-time event logs should at least include the entity fields and activity fields corresponding to the corrected public opinion entity and the corrected entity relationship, respectively. Real-time event logs are used to synchronize real-time public opinion events to the historical database in a standardized form to ensure the timeliness of updating enterprise profiles. For example, for the real-time public opinion event "Jia Technology Group announced its investment in Yi New Energy Company's lithium battery project", the real-time event log built by the server can include fields such as the event subject "Jia Technology Group Co., Ltd.", the activity "equity investment", the event object "Yi New Energy Co., Ltd.", the event occurrence time "2025-03-15", and the event source "Caijing.com".

[0063] As an example, the server can create a real-time event record to be populated with the same field structure as each historical event record in the event details table, and generate a unique event identifier for the real-time event record.

[0064] Then, the server can fill the event subject field and event object field in the real-time event record with the corrected public opinion entities according to their positions in the real-time public opinion event. Specifically, when the corrected public opinion entities are associated through the corrected entity relationship, the server can determine the public opinion entity as the action initiator and the public opinion entity as the action recipient based on the directionality of the entity relationship. After that, the entity name of the initiator can be filled into the event subject field, and the entity name of the recipient can be filled into the event object field.

[0065] Next, the server can obtain the event occurrence time and event source from the metadata of the real-time public opinion event. The event occurrence time can be the release timestamp of the real-time public opinion event itself, and the event source can be the collection channel identifier of the real-time public opinion event. The server can fill the event occurrence time into the event occurrence time field of the real-time event record and fill the event source into the event source field.

[0066] Finally, the server fills the complete real-time event log as a new entry and writes it into the event details table in a structured manner, completing the storage operation.

[0067] In some embodiments, the construction of the real-time event record of the real-time public opinion event based on the modified public opinion entity and the modified entity relationship in step 103 can be achieved by the following technical solution: when the row sequence corresponding to the entity name in the event details table is the same as the row sequence corresponding to the relationship name, the event record corresponding to the row sequence corresponding to the entity name is taken as the real-time event record of the real-time public opinion event.

[0068] As an example, the row sequence can be the index number of the row position of the historical event record in the event details table. For example, the historical event record stored in the 5th row of the event details table has a row sequence of 5.

[0069] As an example, when the server extracts the entity name in the entity retrieval, it can record the row sequence of that entity name in the event details table, that is, the row sequence corresponding to the entity name. Specifically, when the entity retrieval matches the enterprise name or associated enterprise full name field value in a specific row in the event details table, the server can extract the row sequence of that row and store it in association with the entity name.

[0070] When the server extracts a relation name during relation retrieval, it can also record the row sequence of that relation name in the event details table, i.e., the row sequence corresponding to the relation name. Specifically, when the relation retrieval matches the activity field value in a specific row in the event details table, the server can extract the row sequence of that row and store it in association with the relation name.

[0071] Subsequently, the server compares the row sequence corresponding to the entity name with the row sequence corresponding to the relation name. When the two row sequences are the same, it means that the entity name and the relation name both originate from the same historical event record in the event details table. At this time, the server determines that the core information of the current real-time public opinion event has been fully expressed by the historical event record, and the server can directly use the event record corresponding to the row sequence of the entity name as the real-time event record of the real-time public opinion event.

[0072] As an example, when the server determines that the row sequence corresponding to the entity name is the same as the row sequence corresponding to the relation name, it can further verify the time difference between the event occurrence time in the real-time public opinion event and the event occurrence time in the event record corresponding to the row sequence. If the time difference is less than the preset duplicate event window threshold, the server can use the existing event record as the real-time event record; if the time difference exceeds the threshold, even if the three elements overlap, they are considered to be two independent events of the same kind, and the server still needs to construct a new real-time event record normally.

[0073] As an example, after the server uses an existing event record as a real-time event record, it can supplement and update the event source field of the existing event record. Specifically, the server can add a new source identifier for the real-time public opinion event to the event source field, indicating that the event record has been verified by at least two independent sources.

[0074] By comparing the row sequence corresponding to the entity name with the row sequence corresponding to the relation name, existing historical event records in the event details table can be directly reused as real-time event records. This avoids the problem of generating duplicate records when real-time public opinion events have been covered by historical data, thus improving the efficiency of building real-time event records.

[0075] In step 104, industry data is extracted based on the historical event records and the real-time event records to obtain an industry detail table, and enterprise events are extracted based on the historical event records and the real-time event records to obtain an enterprise event table.

[0076] In some embodiments, the industry data extraction in step 104 based on the historical event records and the real-time event records to obtain the industry detail table can be performed as follows: Figure 2 Steps 1041 to 1043 shown are implemented as follows:

[0077] In step 1041, at least one dimension of historical industry data is extracted from the historical event records. The dimensions of the historical industry data include historical industry policy dimension, historical industry supply and demand dynamics dimension, historical industry price fluctuation dimension, historical industry competitive landscape dimension, and historical industry development dimension.

[0078] As an example, historical industry data can be structured data from historical event records used to describe the overall state of the industry in which a company operates during a specific historical period. Historical industry data can contain multiple dimensions and be used to provide industry background analysis for profiling a company.

[0079] The historical industry policy dimension can be a collection of categories from historical event records that involve industry regulations, government plans, tax incentives, entry conditions, subsidy policies, etc. For example, the "release of the new energy vehicle purchase subsidy reduction plan" extracted from historical event records is data from the historical industry policy dimension.

[0080] Historical industry supply and demand dynamics can be a collection of categories from historical event records involving industry capacity expansion, overcapacity, inventory levels, changes in downstream demand, and operating rates. Data from historical industry supply and demand dynamics is used to assess the supply and demand balance of the industry in which a company operates. For example, "the capacity utilization rate of the entire lithium battery industry dropped to 60%" extracted from historical event records is a piece of data from historical industry supply and demand dynamics.

[0081] Historical industry price fluctuation dimension can be a collection of categories from historical event records involving product ex-factory prices, raw material prices, freight rates, terminal retail prices, and price indices. The data of historical industry price fluctuation dimension is used to track the cyclical changes in prices in the industry in which a company operates. For example, "the average spot price of polysilicon has risen for three consecutive weeks" extracted from historical event records is a data point of historical industry price fluctuation dimension.

[0082] The historical industry competitive landscape dimension can be a collection of categories from historical event records that involve industry market concentration, new entrants, mergers and acquisitions, and changes in the market share of leading companies. Data from the historical industry competitive landscape dimension is used to analyze the competitive intensity and structural evolution of the industry in which a company operates. For example, "the combined market share of the top five companies in the industry exceeds 80% for the first time" is a data point from the historical industry competitive landscape dimension extracted from historical event records.

[0083] The historical industry development dimension can be a collection of categories from historical event records that involve industry technology maturity, standardization, industry life cycle stage, and the emergence of new business models. The data of the historical industry development dimension is used to locate the evolutionary stage of the industry in which a company operates. For example, "the first international technical standard in the industry was officially released" extracted from historical event records is a data point of the historical industry development dimension.

[0084] As an example, the server has a pre-defined dimensional classification model. The server can use the title field and content summary field of the historical event record as input to the dimensional classification model, and the dimensional classification model will output the matching probability of the historical event record on each pre-defined dimension.

[0085] The server can set recognition thresholds for each of the five dimensions. When the matching probability for a dimension exceeds the threshold, the server determines that the historical event record belongs to that dimension. Then, the server can extract key entities, action phrases, numerical indicators, and timestamps from the historical event record and restructure them according to the required format for each dimension. Specifically, for the historical industry policy dimension, the server can extract the issuing department, policy name, and implementation date; for the historical industry supply and demand dynamics dimension, the server can extract supply-side indicators, demand-side indicators, and capacity data; for the historical industry price fluctuation dimension, the server can extract price indicator names, direction of change, and magnitude of change; for the historical industry competitive landscape dimension, the server can extract a set of company names, market share values, and industry ranking; and for the historical industry development dimension, the server can extract technical keywords, standardization organizations, and development stage tags.

[0086] As an example, after extracting historical industry data, the server can establish a mapping between enterprise attribute tags and historical industry data dimensions. Specifically, the server can vectorize keywords in historical industry data and calculate the similarity with existing product and technology tags in the enterprise profile. When the similarity exceeds the mapping threshold, the server records the historical industry data as an indirect impact event for the enterprise.

[0087] As an example, the server can preset weights for each dimension, normalize the numerical indicators in the historical industry data for each dimension, and obtain an industry prosperity curve with time as the horizontal axis after weighted summation. The server will then overlay and compare the industry prosperity curve with the revenue or profit change curve in the enterprise profile, and automatically mark the time window when the enterprise's performance deviates from the industry trend.

[0088] In step 1042, the historical industry data is quantized to obtain the first quantized data.

[0089] As an example, data quantization can be the process by which a server converts historical industry data from qualitative descriptions or unstructured numerical values ​​into numerical data. Data quantization includes operations such as assigning intensity values ​​to text descriptions, extracting numerical indicators from natural language and calibrating units, and performing one-hot encoding on discrete categorical values. The purpose of data quantization is to make historical industry data from different dimensions and sources numerically comparable and computable. For example, "significant price increases" in the dimension of historical industry price fluctuations can be quantified into a numerical range feature of "price change rate: +15% or more".

[0090] As an example, the server can preset different quantification rules for different dimensions. The quantification rules include a text intensity mapping table and a numerical extraction template. Specifically, for the historical industry policy dimension, the server extracts policy nature keywords from historical industry data and looks up the corresponding policy strength score in the text intensity mapping table. The text intensity mapping table is pre-set with a one-to-one correspondence between keywords such as "strong support", "general support", "neutral", "general tightening" and "strong tightening" and scores in the range [-1, 1]. The server uses the matched score as the quantification value of the historical industry policy dimension and combines the quantification value with the policy name, issuing department, and effective time to form the first quantification data.

[0091] For the dimensions of historical industry supply and demand dynamics, historical industry price fluctuations, historical industry competitive landscape, and historical industry development, the server can use numerical extraction templates to match numerical expressions from the text content of historical industry data. These templates include regular expression rules to identify values ​​such as "increased by X%", "decreased by X%", "reached X billion yuan", and "market share X%". When a match is successful, the server uses the extracted value and its corresponding unit identifier as the base value. When a match fails, the server can map the text description to an intensity score using a text intensity mapping table. Subsequently, the server normalizes the values ​​to unify the units to a standard unit and maps the values ​​to a preset scalar range.

[0092] As an example, the server can calculate a confidence value between 0 and 1 based on the authority level of the source of historical industry data and the clarity of the numerical expression. For example, the confidence value of the first quantitative data of historical industry data released by the National Bureau of Statistics is set to 0.95; the confidence value of the first quantitative data of data from informal channels and obtained through text mapping is set to 0.6.

[0093] After obtaining the first quantized data, the server can aggregate the first quantized data of the same dimension according to the time series to generate a continuous quantized curve of that dimension. The server uses the timestamp in the first quantized data as the horizontal axis coordinate and the quantized value as the vertical axis coordinate, and obtains the quantized trend line of historical industry data over time through interpolation and smoothing.

[0094] In step 1043, an industry detail table is determined based on the first quantified data and the real-time event record.

[0095] In some embodiments, determining the industry detail table based on the first quantified data and the real-time event records in step 1043 can be achieved through the following technical solution: constructing a candidate industry detail table according to the dimensions to which the first quantified data belongs; extracting at least one dimension of real-time industry data from the real-time event records, wherein the dimensions of the real-time industry data include real-time industry policy dimension, real-time industry supply and demand dynamics dimension, real-time industry price fluctuation dimension, real-time industry competitive landscape dimension, and real-time industry development dimension; quantifying the real-time industry data to obtain second quantified data; updating the first quantified data in the candidate industry detail table according to the dimensions to which the second quantified data belongs, to obtain the industry detail table.

[0096] As an example, the second quantitative data can be a numerical data set generated by the server after performing data quantification on real-time industry data. Both the second and first quantitative data can contain dimension identifiers, quantitative indicator names, quantitative values, and timestamps. The difference is that the values ​​of the second quantitative data come from real-time event records, representing the latest quantitative status of the industry. For example, after performing data quantification on the real-time industry price fluctuation dimension data, the second quantitative data is obtained as "Dimension: Price Fluctuation, Indicator: Average Spot Price, Value: 8% Month-on-Month Increase, Time: First Week of January 2025".

[0097] As an example, the server can read the generated first quantitative data set, traverse each first quantitative data record in the first quantitative data set, and for each first quantitative data record, the server reads the dimension identifier field of the first quantitative data and assigns the first quantitative data to the corresponding dimension partition according to the dimension identifier. The server can create a candidate industry detail table, which is internally divided into five sub-tables according to dimensions, corresponding to the historical industry policy dimension, historical industry supply and demand dynamics dimension, historical industry price fluctuation dimension, historical industry competitive landscape dimension, and historical industry development dimension. The server fills the first quantitative data under each dimension into the corresponding sub-tables to complete the construction of the candidate industry detail table.

[0098] Simultaneously, the server can extract real-time industry data for at least one dimension from real-time event logs. The server can perform dimension recognition processing on the real-time event logs. Specifically, the server can input the title and summary fields of the real-time event logs into a dimension classification model, outputting the matching probability of the real-time event logs on the dimensions of real-time industry policy, real-time industry supply and demand dynamics, real-time industry price fluctuations, real-time industry competitive landscape, and real-time industry development. When the matching probability of a certain dimension exceeds the recognition threshold, the server extracts the structured real-time industry data corresponding to that dimension from the real-time event logs, using the same extraction method as extracting historical industry data from historical event logs.

[0099] The server can invoke the same data quantization rules used to generate the first quantized data, perform text intensity mapping and numerical extraction processing on real-time industry data, and obtain the second quantized data after normalization. Each piece of second quantized data includes a dimension identifier, a quantization indicator name, a quantization value, and a timestamp.

[0100] The server can update the first quantitative data in the candidate industry detail table according to the dimension to which the second quantitative data belongs, thus obtaining the industry detail table. The server can traverse each second quantitative data in the second quantitative data set, read the dimension identifier and quantitative indicator name of the second quantitative data. In the candidate industry detail table, the server can locate the sub-table with the same dimension identifier as the second quantitative data, and search for the first quantitative data record with the same quantitative indicator name as the second quantitative data in the sub-table. When a matching first quantitative data record is found, the server can replace the quantitative value and timestamp of the second quantitative data with the original quantitative value and timestamp in the first quantitative data record to complete the data update. When no matching first quantitative data record is found, the server inserts the second quantitative data as a new record into the corresponding sub-table. After the server completes the traversal processing of all second quantitative data, it saves the updated candidate industry detail table as the industry detail table.

[0101] As an example, when the server performs data updates, it can retain the historical quantitative values ​​of the first quantitative data that match the quantitative indicator name of the second quantitative data. The server can then perform a weighted average of the quantitative values ​​of the first and second quantitative data according to the time interval, assigning a higher weight coefficient to data with newer timestamps and a lower weight coefficient to data with older timestamps. The server writes the weighted and merged values ​​to the corresponding positions in the candidate industry detail table. In this way, the values ​​in the industry detail table can absorb the latest changes in real-time industry data without causing drastic fluctuations due to extreme values ​​in a single real-time event record, thus maintaining the smoothness and stability of the industry quantitative data.

[0102] After the server updates the second quantitative data to the sub-table of the candidate industry details table, the server can retrieve other sub-tables of dimensions that are causally related to the dimension based on the preset dimension association graph. For example, when the real-time industry policy dimension generates a second quantitative data and its quantitative value indicates that the policy is tightening strongly, the server automatically retrieves the historical industry supply and demand dynamic dimension sub-table and applies a downward adjustment correction coefficient to the quantitative value of the supply and demand dynamic indicators of related industry products.

[0103] By extracting real-time industry data from real-time event logs and quantifying it to obtain second-order quantified data, and then updating the first-order quantified data in the candidate industry detail table according to the dimensions, an industry detail table that is continuously synchronized with the current industry status is generated. This can accurately measure the relative position of enterprises in the latest industry coordinate system, significantly improving the dynamic accuracy and timeliness of enterprise profiles.

[0104] In some embodiments, the enterprise event extraction in step 104, based on the historical event records and the real-time event records, to obtain the enterprise event table can be achieved through methods such as... Figure 3 Steps 1044 to 1046 shown are implemented as follows:

[0105] In step 1044, at least one dimension of historical enterprise data is extracted from the historical event records. The dimensions of the historical enterprise data include historical business activity, historical compliance level, historical financial status, historical risk level, historical development stage, and historical service demand.

[0106] As an example, historical enterprise data can be structured data from historical event records used to describe the state of a single enterprise during a specific historical period. Historical enterprise data can contain multiple dimensions to provide a basis for analyzing the historical performance of the enterprise in creating a corporate profile.

[0107] The historical business activity dimension can be a collection of the frequency, scope, and scale of a company's participation in business activities from historical event records. Data in the historical business activity dimension is used to measure the degree to which a company actively participates in market competition. For example, "the company has won 15 bids in the past year" extracted from historical event records is a piece of historical business activity dimension data for a company.

[0108] The historical compliance level dimension can be a collection of historical event records involving administrative penalties, litigation judgments, tax audits, environmental inspections, product recalls, etc., which are related to the company. The data in the historical compliance level dimension is used to assess the company's legal compliance and regulatory risks. For example, "the company was fined 500,000 yuan for environmental violations" is a historical corporate data point in the historical compliance level dimension, which is extracted from the historical event records.

[0109] The historical financial status dimension can be a collection of historical event records involving a company's financing, guarantees, equity pledges, fund freezes, accounts receivable collection, bond issuance, etc. The data in the historical financial status dimension is used to judge the company's financial adequacy and liquidity. For example, "the company completed a Series B financing of RMB 200 million" is a piece of historical corporate data in the historical financial status dimension.

[0110] The historical risk level dimension can be a collection of historical event records involving abnormal business operations, bankruptcy reorganization, default on judgments, major safety accidents, departure of key personnel, loss of major customers, etc. The data of the historical risk level dimension is used to comprehensively evaluate the various risk exposures faced by enterprises. For example, "the enterprise is listed in the list of abnormal business operations" extracted from the historical event records is a piece of historical enterprise data in the historical risk level dimension.

[0111] The historical development stage dimension can be a collection of historical event records involving aspects such as company startup, first round of financing, product launch, market expansion, team scaling, external investment, and IPO preparation. Data in the historical development stage dimension is used to locate a company's position in its life cycle. For example, the "Initial Public Offering Prospectus" extracted from historical event records constitutes a piece of historical company data in the historical development stage dimension.

[0112] The historical service demand dimension can be a collection of historical event records involving enterprises seeking financing, finding strategic investors, recruiting key technical positions, seeking digital transformation solutions, seeking intellectual property agency, and seeking legal services. The data in the historical service demand dimension is used to detect the enterprise's current and future potential service needs. For example, "the enterprise issued a tender notice for a comprehensive digital transformation solution" extracted from historical event records is a piece of historical enterprise data in the historical service demand dimension.

[0113] As an example, the server first obtains a set of historical event records. For each historical event record, the server performs enterprise entity identification processing. Specifically, the server has a pre-set enterprise name database and enterprise identifier index. It links the title field and content summary field of the historical event record to the specific enterprise entity. When a historical event record can be associated with a specific enterprise, the server can include the historical event record in the historical enterprise data extraction process.

[0114] For historical event records associated with specific enterprises, the server can perform dimensional classification processing. Specifically, the server has a preset enterprise dimension classification model, which contains six classifiers, corresponding to the dimensions of historical business activity, historical compliance level, historical financial status, historical risk level, historical development stage, and historical service demand. The server can input the text content of the historical event record into the enterprise dimension classification model, and the enterprise dimension classification model outputs the matching probability of the historical event record in each dimension. When the matching probability of a certain dimension exceeds the preset enterprise dimension recognition threshold, the server determines that the historical event record belongs to that dimension.

[0115] The server extracts information from historical event records that are determined to belong to a certain dimension. For the historical business activity dimension, the server can extract elements such as activity type, participation frequency, and winning bid amount; for the historical compliance level dimension, the server can extract elements such as the penalizing authority, type of incident, penalty amount, and penalty time; for the historical financial status dimension, the server can extract elements such as financing round, financing amount, investor name, and pledge ratio; for the historical risk level dimension, the server can extract elements such as risk event type, amount involved, and current status; for the historical development stage dimension, the server can extract elements such as milestone event type, occurrence time, and scope of impact; and for the historical service demand dimension, the server can extract elements such as demand type, demand description keywords, and demand release time.

[0116] The server can reorganize the extracted elements according to the predefined structure of the corresponding dimension to generate historical enterprise data records. Each historical enterprise data record can include an enterprise identifier, dimension identifier, key element fields and timestamp. The server can then store the generated historical enterprise data records into the historical enterprise data set.

[0117] As an example, after extracting historical enterprise data, the server can construct the trend of change in that dimension based on multiple historical enterprise data records under the same dimension. The server can summarize the activity frequency under the historical business activity dimension by time series to generate an enterprise business activity curve; and accumulate the number of penalties and penalty amounts under the historical compliance level dimension by quarter to generate a compliance risk trend chart.

[0118] The server can establish cross-dimensional correlation analysis rules to discover patterns with causal or co-occurring relationships from historical enterprise data across different dimensions. For example, when the server detects a company's financing arrival time in the historical financial status dimension, it automatically retrieves historical operational activity data from the same time window to determine whether the financing led to business expansion. Similarly, when the server detects a record of default in the historical risk level dimension, it automatically retrieves historical financial status and compliance level data before and after that point in time to trace the precursors of the risk event.

[0119] By extracting historical enterprise data from historical event records, covering dimensions such as historical business activity, historical compliance level, historical financial status, historical risk level, historical development stage, and historical service needs, the server can build a structured historical archive for the enterprise that includes dimensions such as business activity, compliance level, financial status, risk level, development stage, and service needs, thereby improving the accuracy of subsequent enterprise profile construction.

[0120] In step 1045, the historical enterprise data is quantified to obtain third quantified data.

[0121] As an example, the server can obtain the generated historical enterprise data set and perform the appropriate quantitative processing flow according to the dimensions of the historical enterprise data. The server has a set of enterprise quantitative rules for six dimensions of historical enterprise data. Each dimension of the enterprise quantitative rule set includes a text intensity mapping table, a numerical normalization template, and a classification label assignment table.

[0122] For the historical business activity dimension, the enterprise quantitative rule set can include mapping rules for conversion of participation frequency to activity score and a value assignment table for monetary tiers. The server can read the activity type and participation frequency fields extracted from historical enterprise data. The server matches the activity type with a preset activity weight table, and each activity type is assigned a weight coefficient. For example, the weight of bidding activities is 1.0, the weight of exhibition participation is 0.6, and the weight of government procurement registration is 0.8. The server multiplies the occurrence frequency of each activity type by the corresponding weight coefficient and sums them to obtain the original activity score. The server performs normalization processing on the original activity score. The normalization template maps the original activity score to the [0, 1] interval to generate a quantitative value of activity. The server can combine the enterprise identifier, dimension identifier, indicator name, quantitative value, normalization interval, and timestamp into third quantitative data.

[0123] For the historical compliance level dimension, the enterprise quantitative rule set includes a penalty type level mapping table and an amount normalization template. The server can read the penalty authority, cause type, and penalty amount fields from historical enterprise data. The server looks up the initial risk score corresponding to the cause type in the penalty type level mapping table. The penalty type level mapping table pre-maps penalty types such as "criminal judgment," "license revocation," "high fine," "general fine," "warning," and "interview" within the range [0, 1] to obtain the initial risk score. The server substitutes the penalty amount into the amount normalization template, converting the absolute amount into an amount risk coefficient within the range [0, 1]. The server can then perform a weighted sum of the initial risk score and the amount risk coefficient to obtain a quantitative value for the compliance level. The weighting is dynamically determined by the level of the penalty authority and the proximity of the penalty time. The server can combine the enterprise identifier, dimension identifier, indicator name, quantitative value, normalization range, and timestamp into third quantitative data.

[0124] For the historical funding status dimension, the enterprise quantitative rule set can include a financing round score mapping table and an amount normalization template. The server reads the financing round, financing amount, and investor name fields from historical enterprise data. The server matches the financing round with the round score mapping table, mapping rounds such as seed round, angel round, Series A, Series B, and Series C to round scores that increase from smallest to largest. The server converts the financing amount into an amount score within the range of [0, 1] using the amount normalization template. The server can query the investor background database based on the investor name to obtain the investor's industry status coefficient. The round score, amount score, and investor coefficient are weighted and summed to obtain the quantitative value of funding status. The server can combine the enterprise identifier, dimension identifier, indicator name, quantitative value, normalization range, and timestamp into third quantitative data.

[0125] For the historical risk level dimension, the enterprise quantitative rule set can include a risk event type assignment table and a risk exposure amount normalization template. The server can read the risk event type, the amount involved, and the current status field from the historical enterprise data. The server looks up the risk base score corresponding to the risk event type in the risk event type assignment table. The risk event type assignment table pre-maps types such as bankruptcy reorganization, defaulting on judgments, major safety accidents, departure of key personnel, and loss of major customers to the risk base score in the range [0, 1]. The server can multiply the statuses such as "in progress", "closed", and "withdrawn" in the current status field by status coefficients of 1.0, 0.5, and 0.1, respectively. The server substitutes the amount involved into the risk exposure amount normalization template to obtain the exposure score. The server multiplies the risk base score, status coefficient, and exposure score to obtain the risk level quantitative value. The server combines the enterprise identifier, dimension identifier, indicator name, quantitative value, normalization range, and timestamp into the third quantitative data.

[0126] For the historical development stage dimension, the enterprise quantitative rule set includes an event stage mapping table. The server reads milestone events and their occurrence time fields from historical enterprise data. The milestone event mapping table pre-maps events such as enterprise establishment, first round of financing, product launch, market expansion, team scaling, external investment, IPO guidance, and successful IPO into stage positioning values ​​within the range of 0 to 1, according to the enterprise lifecycle. After matching the stage positioning value, the server can apply time decay adjustment to the stage positioning value based on the interval between the occurrence time and the current time to generate a quantitative value for the development stage. The server can combine the enterprise identifier, dimension identifier, indicator name, quantitative value, normalized interval, and timestamp into third-party quantitative data.

[0127] For the historical service demand dimension, the enterprise quantification rule set can include a demand type assignment table and a demand intensity template. The server can read the demand type, demand description keywords, and demand release time fields from historical enterprise data. The server can match the demand type with the demand type assignment table to obtain a basic demand type score. The server counts the urgency words and scale indicators in the demand description keywords and determines the demand intensity coefficient based on the combination of urgency words and scale indicators. The server calculates the demand timeliness coefficient based on the interval between the demand release time and the current time; the more recent the timeliness coefficient, the higher it is. The server multiplies the basic demand type score, the demand intensity coefficient, and the demand timeliness coefficient to obtain the quantified service demand value. The server combines the enterprise identifier, dimension identifier, indicator name, quantified value, normalized interval, and timestamp into third-party quantified data.

[0128] The server aggregates all processed third-dimensional quantized data and generates a third-dimensional quantized data set.

[0129] In step 1046, an enterprise event table is determined based on the third quantified data and the real-time event records.

[0130] In some embodiments, determining the enterprise event table based on the third quantitative data and the real-time event records in step 1046 can be achieved through the following technical solution: constructing a candidate enterprise event table according to the dimensions to which the third quantitative data belongs; extracting at least one dimension of real-time enterprise data from the real-time event records, wherein the dimensions of the real-time enterprise data include real-time operational activity dimension, real-time compliance level dimension, real-time financial status dimension, real-time risk level dimension, real-time development stage dimension, and real-time service demand dimension; quantifying the real-time enterprise data to obtain fourth quantitative data; updating the third quantitative data in the candidate enterprise event table according to the dimensions to which the fourth quantitative data belongs, to obtain the enterprise event table.

[0131] As an example, real-time enterprise data can be structured data from real-time event logs used to describe the latest status of an enterprise. Real-time enterprise data corresponds to historical enterprise data in terms of dimensional division, including real-time operational activity dimension, real-time compliance level dimension, real-time financial status dimension, real-time risk level dimension, real-time development stage dimension, and real-time service demand dimension. For example, "Enterprise A completed a Series C financing of 500 million yuan today" extracted from real-time event logs is a piece of real-time enterprise data in the real-time financial status dimension.

[0132] The real-time business activity dimension can be a classification of the latest dynamics of a company's business activities in the vicinity of the current moment, as reflected in real-time event records. For example, "Company B won the bid for a smart city project" extracted from real-time event records is real-time enterprise data in the real-time business activity dimension.

[0133] The real-time compliance level dimension can be a classification of the latest dynamics of a company being penalized by regulators, involved in litigation, or other compliance events in the real-time event log. For example, "Company C is under investigation for data security issues" extracted from the real-time event log is real-time enterprise data in the real-time compliance level dimension.

[0134] The real-time financial status dimension can be a dynamic classification of the latest financial changes of an enterprise in the current time frame, such as financing, guarantees, and pledges, as reflected in real-time event records. For example, "Enterprise D obtains a bank credit line of 1 billion yuan" extracted from real-time event records is real-time enterprise data in the real-time financial status dimension.

[0135] The real-time risk level dimension can be the latest dynamic classification of new risk events exposed by an enterprise in the vicinity of the current moment, as reflected in the real-time event log. For example, "A fire accident occurred at Enterprise E's factory" extracted from the real-time event log is the real-time enterprise data of the real-time risk level dimension.

[0136] The real-time development stage dimension can be a classification of the latest dynamics of an enterprise in real-time event records that reflect the achievement of new milestones near the current moment. For example, "Enterprise F releases its overseas market strategy for the first time" extracted from real-time event records is real-time enterprise data in the real-time development stage dimension.

[0137] The real-time service demand dimension can be the latest dynamic classification of new service demands generated by an enterprise around the current moment, as reflected in real-time event records. For example, "Enterprise G urgently recruits a Chief Information Security Officer" extracted from real-time event records is real-time enterprise data in the real-time service demand dimension.

[0138] As an example, the server reads the generated third quantitative data set, which contains quantitative records belonging to multiple enterprises and multiple dimensions. The server groups the third quantitative data set using the enterprise identifier as the primary key to obtain a subset of the third quantitative data for each enterprise. For each enterprise's subset of the third quantitative data, the server creates a candidate enterprise event table for each enterprise. The candidate enterprise event table is internally divided into six sub-tables according to dimensions, corresponding to the dimensions of historical business activity, historical compliance level, historical financial status, historical risk level, historical development stage, and historical service demand. The server reads the dimension identifier field of the third quantitative data and allocates the third quantitative data to the corresponding sub-tables in the candidate enterprise event table according to the dimension identifier, thus completing the construction of the candidate enterprise event table.

[0139] Simultaneously, the server can perform joint enterprise entity dimension identification processing on real-time event records. The server has a pre-set enterprise dimension classification model. It identifies the specific enterprise to which the real-time event record belongs from the title and summary fields, and obtains the enterprise identifier. Then, the server can input the text content of the real-time event record into the enterprise dimension classification model. This model can contain six classifiers, corresponding to real-time operational activity, real-time compliance level, real-time financial status, real-time risk level, real-time development stage, and real-time service demand. The enterprise dimension classification model outputs the matching probability of the real-time event record in each dimension. When the matching probability of a certain dimension exceeds a pre-set real-time enterprise dimension identification threshold, the server determines that the real-time event record belongs to that dimension. The server then performs information extraction processing on real-time event records determined to belong to a certain dimension to extract structured real-time enterprise data from them.

[0140] The server can quantify real-time enterprise data to obtain fourth quantified data. The server calls the same enterprise quantification rule set generated by the third quantification data. The enterprise quantification rule set includes a text intensity mapping table for six dimensions, a numerical normalization template, and a classification label assignment table. The server inputs the real-time enterprise data into the corresponding enterprise quantification rule set according to its dimension, performs quantification processing, and outputs the fourth quantified data. Each fourth quantified data item contains an enterprise identifier, dimension identifier, quantification indicator name, quantification value, normalization interval, and timestamp.

[0141] The server updates the third-level quantitative data in the candidate enterprise event table according to the dimension to which the fourth-level quantitative data belongs, thus obtaining the enterprise event table. Specifically, the server can traverse each fourth-level quantitative data record in the fourth-level quantitative data set, read the enterprise identifier and dimension identifier of the fourth-level quantitative data, and locate the corresponding candidate enterprise event table based on the enterprise identifier. In the candidate enterprise event table, the server locates the corresponding sub-table based on the dimension identifier. In the sub-table, the server searches for a third-level quantitative data record with the same quantitative indicator name as the fourth-level quantitative data. When a matching third-level quantitative data record is found, the server can overwrite the quantitative value and timestamp of the fourth-level quantitative data into the quantitative value field and timestamp field of the third-level quantitative data record to complete the data update. When no matching third-level quantitative data record is found, the server can insert the fourth-level quantitative data as a new record into the corresponding sub-table. After completing the traversal processing of all fourth-level quantitative data, the server saves the updated candidate enterprise event table as the enterprise event table.

[0142] As an example, when performing data updates, the server can set a real-time event evaluation window. For instance, the real-time event evaluation window can be 24 hours prior to the current time. The server aggregates multiple real-time enterprise data points for the same enterprise and the same dimension within this real-time event evaluation window, and quantifies each of these data points to obtain multiple fourth-quantified data points. The server calculates the average or median of these fourth-quantified data points and then uses this aggregation result to update the third-quantified data in the candidate enterprise event table.

[0143] By extracting real-time enterprise data from real-time event logs and quantifying it to obtain fourth quantitative data, and then updating the third quantitative data in the candidate enterprise event table according to dimensions, an enterprise event table that is continuously synchronized with the latest actual status of the enterprise is generated, improving the real-time performance, accuracy and dynamic tracking capability of the enterprise profile.

[0144] In step 105, an industry profile is constructed based on the industry details table, an enterprise profile is constructed based on the enterprise event table, and services are provided to the target enterprise based on the industry profile and the enterprise profile.

[0145] As an example, an industry detail table can be a structured data table generated by a server after quantifying data from multiple industry dimensions. The industry detail table can include industry identifiers, quantitative values ​​for industry business activity, industry compliance level, industry financial status, industry risk level, industry development stage, and industry service demand. For example, in the industry detail table for the semiconductor industry, the quantitative value for industry business activity is 0.82, and the quantitative value for industry compliance level is 0.45.

[0146] An industry profile can be a structured description of an industry as a whole across multiple dimensions. An industry profile can include a set of industry benchmark values, with each benchmark value corresponding to a dimension, reflecting the statistical central trend or overall level of the industry in that dimension. For example, for the new energy battery industry, the industry profile includes benchmark values ​​for industry business activity (0.78), industry compliance level (0.32), industry financial status (0.85), industry risk level (0.40), industry development stage (0.65), and industry service demand (0.70).

[0147] A company profile can be a structured description of a company's characteristics across multiple dimensions. A company profile can include a set of quantitative values ​​for the company itself. Each quantitative value corresponds to a dimension, reflecting the company's actual performance in that dimension. For example, the company profile of company H includes a company's operational activity value of 0.55, a company's compliance level value of 0.20, a company's financial status value of 0.90, a company's risk level value of 0.68, a company's development stage value of 0.40, and a company's service needs value of 0.75.

[0148] As an example, the server can read an industry detail table containing records from multiple industries. Each record contains an industry identifier and quantitative values ​​for each industry dimension. For each industry, the server extracts all the quantitative values ​​for each dimension from the corresponding row in the industry detail table. The server then organizes the extracted quantitative values ​​to generate an industry profile data structure. This structure includes an industry identifier, industry business activity benchmark, industry compliance level benchmark, industry financial status benchmark, industry risk level benchmark, industry development stage benchmark, and industry service demand benchmark. The server directly maps the quantitative values ​​for each industry in the industry detail table to the corresponding industry benchmark values, thus completing the construction of the industry profile.

[0149] The server can read the enterprise event table, which is for a single enterprise and contains six sub-tables. Each sub-table stores the latest quantitative data records for that dimension. For each sub-table in the enterprise event table, the server aggregates the quantitative values ​​of all quantitative records in that sub-table and calculates the enterprise's own quantitative value for that dimension. The aggregation method can be to take the latest value, the average value, or the weighted value. The server combines the calculated enterprise's own quantitative values ​​for the six dimensions to generate an enterprise profile data structure. The enterprise profile data structure can include enterprise identifier, enterprise's own business activity value, enterprise's own compliance level value, enterprise's own financial status value, enterprise's own risk level value, enterprise's own development stage value, and enterprise's own service needs value.

[0150] After receiving a service request from a target company, the server can parse the target company's identifier from the service request. Based on the target company's identifier, the server can search for the target company's corporate profile in the corporate profile set, determine the industry to which the target company belongs, and then obtain the corresponding industry profile of the target company. The server then compares the target company's corporate profile with the industry profile of the target company's industry in a dimension-by-dimensional manner.

[0151] The server calculates the difference between the target company's own business activity value and the industry benchmark business activity value to determine the target company's relative performance in terms of business activity. The server performs difference calculations on all six dimensions to generate a difference analysis vector. Based on the direction and degree of deviation of each dimension in the difference analysis vector, the server identifies the target company's relative strengths and weaknesses.

[0152] The server has a pre-set service decision rule base, which can contain the mapping relationship between dimensional differences and service types. The server inputs the difference analysis vector into the service decision rule base and generates a service recommendation list for the target enterprise. The service recommendation list can include specific service items such as risk warning, financing connection, compliance rectification, market expansion suggestions and digital transformation solutions.

[0153] As an example, after generating the difference analysis vector, the server can also extract the historical time series of the quantitative values ​​of each dimension in the target company's corporate event table, calculate the trend parameters of the target company in six dimensions, and at the same time extract the trend template of the corresponding industry dimension in the industry profile. This template can be obtained by fitting the historical multi-period versions of the industry details table.

[0154] The server compares the target company's trend parameters with industry trend templates to determine whether the target company is accelerating its catch-up, evolving synchronously, or deviating from the downward trajectory in each dimension. The server can adjust the service priority and urgency indicators in the service recommendation list based on the trend comparison results. For example, when the difference in the target company's risk level dimension is already positive and the trend parameters show accelerated deterioration, risk warning services are marked as the highest priority and accompanied by short-term suggestions.

[0155] As an example, when the server receives a request from a service provider to filter all companies within a specific industry, the server can read the industry profile and the company profiles of all companies within the industry. The server calculates the comprehensive deviation for each company. The comprehensive deviation is the sum of the squares of the weighted differences between the values ​​of each dimension of the company profile and the industry benchmark value. The server sorts all companies within the industry according to the comprehensive deviation and presents the sorting results and the service recommendation list to the service provider to help the service provider identify the target companies that need the most intervention or are the most valuable within the industry.

[0156] In some embodiments, providing services to the target enterprise based on the industry profile and the enterprise profile in step 105 can be achieved through, for example... Figure 4 Steps 1051 to 1053 shown are implemented as follows:

[0157] In step 1051, at least one dimension of target industry data of the target enterprise's industry is extracted from the industry profile. The dimensions of the target industry data include industry policy dimension, industry supply and demand dynamics dimension, industry price fluctuation dimension, industry competitive landscape dimension, and industry development dimension.

[0158] As an example, target industry data can be structured numerical values ​​used to describe the target company's industry in a specific external environment dimension. Target industry data is used to supplement the granularity of the industry profile in the quantitative expression of the external environment. For example, the target industry data extracted from the industry profile of "new energy battery industry" is "industry policy dimension: support strength score 0.92, industry supply and demand dynamics dimension: demand prosperity 0.85, industry price fluctuation dimension: price volatility 0.33, industry competitive landscape dimension: concentration score 0.60, industry development dimension: maturity score 0.68".

[0159] The industry policy dimension can be a dimension in the target industry data that represents the frequency of policy support, regulatory constraints, or policy changes in the industry. The industry policy dimension quantifies the degree of leniency or strictness of the policy environment and is used to assess the direction of the policy's impact on the target company. For example, "Industry policy dimension: Environmental compliance tightening index 0.72" indicates that the industry faces high environmental regulatory pressure.

[0160] The industry supply and demand dynamic dimension can be a dimension in the target industry data that represents the relative relationship and changing trend of the total supply and demand in the industry. The industry supply and demand dynamic dimension quantifies the degree of market gap or surplus and is used to identify market opportunities or pressures faced by target companies. For example, "Industry supply and demand dynamic dimension: talent supply sufficiency 0.38" indicates that the industry's talent supply is tight.

[0161] The industry price volatility dimension can be a dimension in the target industry data that represents the magnitude and frequency of price changes of key products or services in the industry within a time window. The industry price volatility dimension quantifies the price risk of the industry and is used to provide cost warnings or pricing suggestions to target companies in the service. For example, "Industry price volatility dimension: raw material price volatility 0.56" means that the raw material price in the industry is in a moderate to strong volatility.

[0162] The industry competitive landscape dimension can be a dimension in the target industry data that represents the industry concentration, the distribution of the number of competitors, and the market share of leading companies. The industry competitive landscape dimension quantifies the intensity of competition and the degree of monopoly in the industry. For example, "Industry competitive landscape dimension: market share of leading companies 0.73" indicates that the industry has a high degree of concentration.

[0163] The industry development dimension can be a dimension in the target industry data that represents the industry's life cycle position, technological maturity, and infrastructure completeness. The industry development dimension quantifies the overall evolution level of the industry. For example, "Industry Development Dimension: Technological Maturity Score 0.80" means that the industry's technology is relatively mature.

[0164] As an example, the server can obtain the target company's corporate identifier and determine the industry identifier of the target company by querying the company's basic information database.

[0165] The server locates the corresponding industry profile in the industry profile set based on the industry identifier. The industry profile can be a structured data record, which includes six benchmark fields: industry business activity benchmark, industry compliance level benchmark, industry financial status benchmark, industry risk level benchmark, industry development stage benchmark, and industry service demand benchmark. In addition, it can also include quantitative descriptions of industry policy, industry supply and demand dynamics, industry price fluctuations, industry competitive landscape, and industry development level.

[0166] The server can first determine the list of dimensions for the target industry data to be extracted. The list may include one or a combination of industry policy dimensions, industry supply and demand dynamics dimensions, industry price fluctuation dimensions, industry competitive landscape dimensions, and industry development dimensions.

[0167] Based on the dimensions included in the dimension list, the server can retrieve corresponding quantitative values ​​from the industry profile. For the industry policy dimension, the server can retrieve the industry policy leniency score or the industry policy risk score; for the industry supply and demand dynamics dimension, the server can retrieve the industry demand prosperity score and the industry supply sufficiency score; for the industry price volatility dimension, the server can retrieve the industry price volatility value and the industry price trend direction indicator; for the industry competitive landscape dimension, the server can retrieve the industry concentration score and the industry competitor density value; for the industry development dimension, the server can retrieve the industry life cycle stage indicator and the industry technology maturity score.

[0168] The server encapsulates the values ​​read from each dimension to generate target industry data. The target industry data can carry industry identifier, enterprise identifier, dimension name and dimension value, as well as data extraction timestamp. The server outputs the target industry data for subsequent use when providing services to the target enterprise.

[0169] In step 1052, at least one dimension of target enterprise data is extracted from the enterprise profile. The dimensions of the target enterprise data include the dimensions of business activity, compliance level, financial status, risk level, development stage, and service demand.

[0170] As an example, the server can obtain the target company's enterprise identifier, which can be obtained through methods such as carrying it in service requests, dispatching tasks through task queues, or pre-setting a list of target companies.

[0171] The server can search the enterprise profile set based on the enterprise identifier of the target enterprise. The enterprise profile set contains enterprise profiles of multiple enterprises. Each enterprise profile record uses the enterprise identifier as the primary key. The server can locate the enterprise profile that matches the enterprise identifier of the target enterprise.

[0172] The server can obtain dimension extraction instructions for the target enterprise's data. The dimension extraction instructions specify at least one dimension to be extracted. The dimension range is limited to the dimensions of business activity, compliance level, financial status, risk level, development stage, and service needs. The dimension extraction instructions can be generated by the service decision module or determined by the preset service template configuration. The dimension extraction instructions carry a list of dimension identifiers.

[0173] The server reads the list of dimension identifiers and extracts quantitative values ​​one by one from the corresponding fields of the enterprise profile based on the dimension identifiers. The data structure of the enterprise profile includes six fields: enterprise's own business activity value, enterprise's own compliance level value, enterprise's own financial status value, enterprise's own risk level value, enterprise's own development stage value, and enterprise's own service needs value.

[0174] The server encapsulates the extracted quantified values ​​of one or more dimensions into target enterprise data. The structure of the target enterprise data may include enterprise identifier, dimension name field, and dimension value field. For each extracted dimension, the server generates a sub-item containing the dimension name and the corresponding quantified value. If the dimension extraction instruction requires the extraction of multiple dimensions at the same time, the target enterprise data will contain a set of sub-items of multiple dimensions. The server attaches an extraction timestamp to the target enterprise data to mark the time of data extraction.

[0175] In step 1053, services are provided to the target enterprise based on the target industry data and the target enterprise data.

[0176] In some embodiments, providing services to the target enterprise based on the target industry data and the target enterprise data in step 1053 can be implemented through the following technical solution: determining the enterprise type of the target enterprise based on the target industry data and the target enterprise data; when the enterprise type indicates that the target enterprise is a shrinking enterprise, providing risk services to the target enterprise, the risk services including legal consulting services, debt optimization services, operational compliance services, and environmental rectification services; when the enterprise type indicates that the target enterprise is an expanding enterprise, providing expansion services to the target enterprise, the expansion services including subsidy application services and industry fund matching services.

[0177] As an example, the enterprise type can be used to distinguish whether the target enterprise is currently in a contraction or expansion phase. Enterprise types include at least contraction and expansion. For example, the server determines that the enterprise type of enterprise H is "contraction", which indicates that the enterprise is facing industry environmental pressure and its own performance is weak.

[0178] Shrinking enterprises are those whose own dimensions in the target enterprise data are lower than the industry benchmark, while the target industry data shows that multiple industry dimensions indicate an unfavorable industry environment, resulting in the target enterprise's overall performance decline, risk increase, or scale reduction. Shrinking enterprises need to obtain services aimed at preventing and mitigating risks. For example, if a manufacturing enterprise's financial status and operational activity dimensions are both lower than the corresponding benchmark values ​​in the industry profile, and the industry supply and demand dynamics dimension shows shrinking market demand, the server will identify the enterprise as a shrinking enterprise.

[0179] An expansionary enterprise can be defined as one whose own dimensions in the target enterprise data are higher than or equal to the industry benchmark, while the target industry data shows that multiple industry dimensions indicate a favorable industry environment, resulting in the target enterprise's overall performance improvement, expansion of its business scope, or increase in market share. Expansionary enterprises need to acquire services aimed at seizing opportunities and accelerating development. For example, if a technology company's operational activity and service demand dimensions are both higher than the corresponding benchmark values ​​in the industry profile, and the industry policy dimension shows increased support, the server will classify the company as an expansionary enterprise.

[0180] Risk services can be a general term for the service items that a server pushes to shrinking enterprises. Risk services aim to help shrinking enterprises avoid, reduce or resolve various risks they currently face. Risk services may include legal consulting services, debt optimization services, operational compliance services and environmental rectification services. For example, the risk service list that a server pushes to a shrinking enterprise may include accounts receivable collection and connection in debt optimization services and pollution discharge compliance transformation plan in environmental rectification services.

[0181] Extended services can be a general term for the service items that the server pushes to expanding enterprises. Extended services aim to help expanding enterprises acquire resources, expand their advantages, or accelerate market layout. Extended services may include subsidy application services and industry fund matching services. For example, the list of extended services pushed by the server to an expanding enterprise may include matching of special subsidies for high-tech enterprises in the subsidy application service and the application portal for the new energy industry guidance fund in the industry fund matching service.

[0182] As an example, the server can input target industry data and target enterprise data into a preset enterprise type determination model, which includes environmental dimension scoring rules and enterprise dimension scoring rules.

[0183] The environmental dimension scoring rules can score the values ​​of each industry dimension in the target industry data separately and sum them with weights to generate a score for the industry environmental dimension. When the score for industry policy leniency exceeds the leniency threshold, a positive score is awarded to this dimension; when the score for industry demand prosperity exceeds the prosperity threshold, a positive score is awarded to this dimension; when the score for industry supply sufficiency is higher than the sufficiency threshold, a positive score is awarded to this dimension; when the score for industry price volatility is lower than the stability threshold, a positive score is awarded to this dimension; when the score for industry concentration is in a moderate range, a positive score is awarded to this dimension; when the score for industry technology maturity exceeds the maturity threshold, a positive score is awarded to this dimension. The server can map the scores of each industry dimension to sub-scores of the environmental dimension and calculate the total score of the industry environmental dimension according to preset weights.

[0184] The enterprise-level scoring rules can score each enterprise-level value in the target enterprise data separately and sum them in a weighted manner to generate a total score for the enterprise-level dimensions. Specifically, a positive score is awarded when the enterprise's own operational activity value exceeds the activity threshold, when the enterprise's own compliance level value exceeds the compliance threshold, when the enterprise's own financial status value exceeds the sufficiency threshold, when the enterprise's own risk level value is below the risk threshold, when the enterprise's own development stage value exceeds the growth threshold, and when the enterprise's own service demand value is within the reasonable demand range. After the server maps the scores of each enterprise-level dimension to sub-scores of the enterprise-level dimension, it calculates the total score of the enterprise-level dimensions according to the preset weights.

[0185] When the total score of the enterprise dimension is lower than the lower threshold of the enterprise dimension and the total score of the industry environment dimension is lower than the lower threshold of the environment dimension, the server can determine that the target enterprise is a shrinking enterprise. When the total score of the enterprise dimension is higher than the upper threshold of the enterprise dimension and the total score of the industry environment dimension is higher than the upper threshold of the environment dimension, the server can determine that the target enterprise is an expanding enterprise.

[0186] When the enterprise type indicates that the target enterprise is a shrinking enterprise, the server can retrieve a risk service package from the service resource library. The risk service package may include legal consulting services, debt optimization services, operational compliance services, environmental rectification services, service provider information, and service content templates. The server can fill the service content template of the risk service package with the target enterprise's enterprise identifier and the compliance level, financial status, and risk level dimensions that are directly related to the risk from the target enterprise's data, and generate a risk service list for the target enterprise. The risk service list can assign priorities to each risk service sub-item, and the server can push the risk service list to the target enterprise's interactive terminal.

[0187] When the enterprise type indicates that the target enterprise is an expansion-type enterprise, the server can retrieve an expansion service package from the service resource library. The expansion service package may include subsidy application services, industry fund matching services, service provider information, and service content templates. The server can fill the service content template with the target enterprise's enterprise identifier and the values ​​of the business activity dimension, development stage dimension, and service demand dimension related to expansion from the target enterprise's data to generate an expansion service list for the target enterprise. The server can then push the expansion service list to the target enterprise's interactive terminal or service distribution system.

[0188] By jointly determining the enterprise type of the target enterprise based on target industry data and target enterprise data, and providing risk services to shrinking enterprises and expansion services to expanding enterprises according to the enterprise type, the targeting, timeliness and practicality of services in the enterprise profile construction scenario are improved, and resource misallocation and service mismatch are avoided.

[0189] In some embodiments, providing services to the target enterprise based on the industry profile and the enterprise profile in step 105 can be implemented through the following technical solutions: extracting real-time industry information from the industry profile and extracting real-time enterprise information of the target enterprise from the enterprise profile; when the real-time industry information is trend information and the real-time enterprise information indicates that the target enterprise is an enterprise with idle capacity, pushing instant services to the target enterprise, wherein the trend information includes at least one of policy implementation, commodity price fluctuation exceeding a change threshold, and target orders exceeding an order threshold being signed, and the instant services include at least one of subsidy application service, order matching service, and financial credit service; when the real-time industry information is risk information and the real-time enterprise information indicates that the target enterprise is a risky enterprise, pushing risk control services to the target enterprise, wherein the risk control services include industrial transformation consulting services, production reduction suggestion services, and environmental rectification suggestion services.

[0190] As an example, real-time industry information can be structured information that represents recent dynamic events or state changes in the industry to which the target company belongs at a specific moment. For example, real-time industry information could include the extension of the new energy vehicle purchase tax reduction, the spot price of lithium carbonate rising by 12% compared to the previous day, and the confirmation of energy storage project order A with an order volume of 30GWh, exceeding the order threshold.

[0191] Real-time enterprise information can be structured information that represents the actual operating status of a target enterprise at a specific moment. Real-time enterprise information reflects the latest status of the target enterprise in terms of business activity, capacity utilization, capital turnover, and risk exposure. For example, real-time enterprise information could show a capacity utilization rate of 0.43, which is lower than the capacity idle threshold of 0.60, indicating that the target enterprise is a capacity idle enterprise.

[0192] Trend information can be information that represents a sudden favorable event in the industry to which the target company belongs. Trend information can include at least one of the following: policy implementation, commodity price fluctuation exceeding a threshold, and target orders exceeding a threshold being signed. Policy implementation refers to the formal implementation or release of industry-related support policies, subsidy policies, or relaxation policies; commodity price fluctuation exceeding a threshold refers to the price fluctuation of key commodities exceeding a preset percentage value within the monitoring period; target orders exceeding a threshold being signed refers to a large order transaction event in the industry where the single order volume exceeds a preset order volume threshold.

[0193] Risk information can be information that indicates a sudden adverse event in the industry to which the target company belongs. Risk information can include a concentrated outbreak of industry regulatory penalties, a sharp drop in the price of key raw materials leading to inventory impairment risk, and sudden trade restriction policies.

[0194] As an example, the server can extract newly generated dynamic events from the industry profile within the most recent scanning cycle according to a preset scanning cycle, and generate real-time industry information. The real-time industry information can include event type and event quantification parameters. Event type can include policy implementation, commodity price changes, target order signing, sudden industry risks, etc. Event quantification parameters can include policy identification and effective time, commodity price change range, target order order volume, etc.

[0195] The server can extract real-time enterprise information of the target enterprise from the enterprise profile. The real-time enterprise information includes the target enterprise's capacity utilization rate, cash flow, compliance status, and risk level.

[0196] The server can check whether the event type of the real-time industry information is policy implementation. If the event type is policy implementation, the server determines that the real-time industry information is a hot trend. The server can check whether the real-time industry information contains the price change range of commodities. If the price change range of commodities is greater than the change threshold, the server determines that the real-time industry information is a hot trend. The server can check whether the real-time industry information contains the signing of target orders. If the order volume of target orders is greater than the order threshold, the server determines that the real-time industry information is a hot trend.

[0197] The server can also check whether the real-time industry information is related to adverse events such as the release of industry trade restriction policies, embargoes on key raw materials, cluster events of regulatory penalties, or a sharp decline in industry demand. If so, the server determines that the real-time industry information is risky.

[0198] When real-time industry information is identified as trending information, the server can determine whether the target company is a company with idle capacity based on its real-time company information. Specifically, the server can compare the capacity utilization rate in the real-time company information with a preset idle capacity threshold. When the capacity utilization rate is lower than the idle capacity threshold, the server determines that the target company is a company with idle capacity. If the target company is determined to be a company with idle capacity, the server can push instant services to the target company. The specific process of pushing instant services can be as follows: the server retrieves subsidy application services related to the policy implementation in the trending information from the service resource library and generates subsidy information cards; the server matches the capacity parameters of the company with idle capacity with the target orders in the trending information and generates order matching service links; the server matches financial credit service products based on the industry prosperity rating upward adjustment signal corresponding to the trending information and pushes them to the target company. The instant services pushed by the server include at least one of the following: subsidy application service, order matching service, and financial credit service.

[0199] When real-time industry information is identified as risky, the server can determine whether the target company is a risky company based on its real-time company information. Specifically, the server can compare the risk level value in the real-time company information with the high-risk threshold, or the compliance level value with the compliance threshold, or the cash flow value with the safety threshold. When any indicator in the real-time company information triggers the risk assessment condition, the server determines that the target company is a risky company.

[0200] If the target company is identified as a high-risk company, the server can push risk control services to the target company. These services may include industry transformation consulting services, production reduction suggestions, and environmental rectification suggestions.

[0201] By extracting real-time industry information from industry profiles and real-time enterprise information from enterprise profiles, and by pushing instant services when the real-time industry information indicates a trending sector and the real-time enterprise information indicates that the target enterprise is an enterprise with idle capacity, and by pushing risk control services when the real-time industry information indicates a risky sector and the real-time enterprise information indicates that the target enterprise is a risky enterprise, the timeliness of service delivery and the accuracy of matching enterprises with services are improved.

[0202] The following is combined with Figure 5 This invention describes an apparatus for constructing industry and enterprise profiles based on public opinion events. Figure 5 This is a schematic diagram of an apparatus for constructing industry and enterprise profiles based on public opinion events, provided by the present invention. The apparatus for constructing industry and enterprise profiles based on public opinion events described below and the method for constructing industry and enterprise profiles based on public opinion events described above can be referred to in correspondence.

[0203] The identification module 501 is used to perform entity identification on real-time public opinion events, obtain the public opinion entities included in the real-time public opinion events, and extract the relationships between the public opinion entities.

[0204] The correction module 502 is used to correct the public opinion entity and the entity relationship according to the historical event records recorded in the event details table, so as to obtain the corrected public opinion entity and the corrected entity relationship. Each historical event record in the event details table corresponds to a historical public opinion event. The historical event record includes at least one of the following fields: event identifier, company name, full name of associated company, industry four-level classification, product term, event subject, activity, event object, event occurrence time, and event source.

[0205] The construction module 503 is used to construct a real-time event record of the real-time public opinion event based on the corrected public opinion entity and the corrected entity relationship;

[0206] The extraction module 504 is used to extract industry data based on the historical event records and the real-time event records to obtain an industry detail table, and to extract enterprise events based on the historical event records and the real-time event records to obtain an enterprise event table.

[0207] The construction module 503 is also used to construct an industry profile based on the industry details table, construct an enterprise profile based on the enterprise event table, and provide services to the target enterprise based on the industry profile and the enterprise profile.

[0208] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6As shown, the electronic device may include: a processor 610, a communications interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communications interface 620, and the memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a method for constructing industry and enterprise profiles based on public opinion events. This method includes:

[0209] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0210] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a method for constructing industry profiles and enterprise profiles based on public opinion events, as provided by the methods described above. This method includes: performing entity recognition on real-time public opinion events to obtain the public opinion entities included in the real-time public opinion events; extracting relationships from the real-time public opinion events to obtain the entity relationships between the public opinion entities; and correcting the public opinion entities and entity relationships according to historical event records in the event details table to obtain corrected public opinion entities and corrected entity relationships. Each historical event record in the event details table... Each record corresponds to a historical public opinion event. The historical event record includes at least one of the following fields: event identifier, company name, full name of associated company, industry classification (fourth level), product term, event subject, activity, event object, event occurrence time, and event source. Based on the corrected public opinion entities and the corrected entity relationships, a real-time event record for the real-time public opinion event is constructed. Based on the historical event record and the real-time event record, industry data is extracted to obtain an industry detail table. Based on the historical event record and the real-time event record, company events are extracted to obtain a company event table. An industry profile is constructed based on the industry detail table, a company profile is constructed based on the company event table, and services are provided to the target company based on the industry profile and the company profile.

[0211] On another front, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for constructing industry profiles and enterprise profiles based on public opinion events, as provided by the methods described above. This method includes: performing entity recognition on real-time public opinion events to obtain the public opinion entities included in the real-time public opinion events; extracting relationships from the real-time public opinion events to obtain the entity relationships between the public opinion entities; and correcting the public opinion entities and the entity relationships according to historical event records in an event detail table to obtain corrected public opinion entities and corrected entity relationships, wherein each historical event record in the event detail table corresponds to a historical public opinion event. The historical event records include at least one of the following fields: event identifier, company name, full name of associated company, industry level four classification, product term, event subject, activity, event object, event occurrence time, and event source. Based on the corrected public opinion entities and the corrected entity relationships, a real-time event record for the real-time public opinion event is constructed. Based on the historical event records and the real-time event records, industry data is extracted to obtain an industry detail table, and based on the historical event records and the real-time event records, company events are extracted to obtain a company event table. An industry profile is constructed based on the industry detail table, a company profile is constructed based on the company event table, and services are provided to the target company based on the industry profile and the company profile.

[0212] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0213] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0214] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0215] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing industry profiles and enterprise profiles based on public opinion events, characterized in that, include: Entity identification is performed on real-time public opinion events to obtain the public opinion entities included in the real-time public opinion events, and relationship extraction is performed on the real-time public opinion events to obtain the entity relationships between the public opinion entities; Based on the historical event records in the event details table, the public opinion entity and the entity relationship are corrected to obtain the corrected public opinion entity and the corrected entity relationship. Each historical event record in the event details table corresponds to a historical public opinion event. The historical event record includes at least one of the following fields: event identifier, company name, full name of associated company, industry level four classification, product term, event subject, activity, event object, event occurrence time, and event source. Based on the corrected public opinion entities and the corrected entity relationships, a real-time event record of the real-time public opinion event is constructed; Based on the historical event records and the real-time event records, industry data is extracted to obtain an industry detail table, and based on the historical event records and the real-time event records, enterprise events are extracted to obtain an enterprise event table. An industry profile is constructed based on the industry details table, an enterprise profile is constructed based on the enterprise event table, and services are provided to the target enterprise based on the industry profile and the enterprise profile.

2. The method for constructing industry profiles and enterprise profiles based on public opinion events according to claim 1, characterized in that, The process of correcting the public opinion entity and its relationships based on historical event records in the event details table to obtain the corrected public opinion entity and its relationships includes: Based on the public opinion entity, an entity retrieval is performed from the event details table to obtain the entity name corresponding to the public opinion entity in the event details table, and the public opinion entity is updated with the entity name as the corrected public opinion entity; Based on the entity relationship, a relationship retrieval is performed from the event details table to obtain the relationship name of the entity relationship in the event details table, and the entity relationship is updated with the relationship name as the corrected entity relationship.

3. The method for constructing industry profiles and enterprise profiles based on public opinion events according to claim 2, characterized in that, The construction of real-time event records for the real-time public opinion events based on the corrected public opinion entities and the corrected entity relationships includes: When the row sequence corresponding to the entity name in the event details table is the same as the row sequence corresponding to the relationship name, the event record corresponding to the row sequence corresponding to the entity name is taken as the real-time event record of the real-time public opinion event.

4. The method for constructing industry profiles and enterprise profiles based on public opinion events according to claim 1, characterized in that, The process of extracting industry data based on the historical event records and the real-time event records to obtain an industry detail table includes: At least one dimension of historical industry data is extracted from the historical event records, wherein the dimensions of the historical industry data include historical industry policy dimension, historical industry supply and demand dynamics dimension, historical industry price fluctuation dimension, historical industry competitive landscape dimension, and historical industry development dimension. The historical industry data is quantified to obtain the first quantified data; Based on the first quantitative data and the real-time event records, an industry detail table is determined.

5. The method for constructing industry profiles and enterprise profiles based on public opinion events according to claim 4, characterized in that, The step of determining the industry detail table based on the first quantified data and the real-time event records includes: Construct a candidate industry detail table based on the dimension to which the first quantified data belongs; At least one dimension of real-time industry data is extracted from the real-time event records, wherein the dimensions of the real-time industry data include real-time industry policy dimension, real-time industry supply and demand dynamics dimension, real-time industry price fluctuation dimension, real-time industry competitive landscape dimension, and real-time industry development dimension. The real-time industry data is quantized to obtain second quantized data; The first quantitative data in the candidate industry detail table is updated according to the dimension to which the second quantitative data belongs, to obtain the industry detail table.

6. The method for constructing industry profiles and enterprise profiles based on public opinion events according to claim 1, characterized in that, The process of extracting enterprise events based on the historical event records and the real-time event records to obtain an enterprise event table includes: At least one dimension of historical enterprise data is extracted from the historical event records. The dimensions of the historical enterprise data include historical business activity, historical compliance level, historical financial status, historical risk level, historical development stage, and historical service demand. The historical enterprise data is quantified to obtain the third quantified data; Based on the third quantitative data and the real-time event records, the enterprise event table is determined.

7. The method for constructing industry profiles and enterprise profiles based on public opinion events according to claim 6, characterized in that, The process of determining the enterprise event table based on the third quantitative data and the real-time event records includes: Construct a candidate enterprise event table according to the dimension to which the third quantitative data belongs; At least one dimension of real-time enterprise data is extracted from the real-time event records. The dimensions of the real-time enterprise data include real-time business activity, real-time compliance level, real-time financial status, real-time risk level, real-time development stage, and real-time service demand. The real-time enterprise data is quantized to obtain the fourth quantized data; The third quantitative data in the candidate enterprise event table is updated according to the dimension to which the fourth quantitative data belongs, to obtain the enterprise event table.

8. The method for constructing industry profiles and enterprise profiles based on public opinion events according to claim 1, characterized in that, The provision of services to target enterprises based on the industry profile and the enterprise profile includes: Extract at least one dimension of target industry data from the industry profile to which the target company belongs. The dimensions of the target industry data include industry policy dimension, industry supply and demand dynamics dimension, industry price fluctuation dimension, industry competitive landscape dimension, and industry development dimension. Extract at least one dimension of target enterprise data from the enterprise profile. The dimensions of the target enterprise data include the dimensions of business activity, compliance level, financial status, risk level, development stage, and service needs. Services are provided to the target company based on the target industry data and the target company data.

9. The method for constructing industry profiles and enterprise profiles based on public opinion events according to claim 8, characterized in that, The provision of services to the target enterprise based on the target industry data and the target enterprise data includes: Based on the target industry data and the target company data, the company type of the target company is determined; When the enterprise type indicates that the target enterprise is a shrinking enterprise, risk services are provided to the target enterprise, including legal consulting services, debt optimization services, operational compliance services, and environmental rectification services; When the enterprise type indicates that the target enterprise is an expansion-oriented enterprise, expansion services are provided to the target enterprise, including subsidy application services and industry fund matching services.

10. The method for constructing industry profiles and enterprise profiles based on public opinion events according to claim 1, characterized in that, The provision of services to target enterprises based on the industry profile and the enterprise profile includes: Real-time industry information is extracted from the industry profile, and real-time enterprise information of the target enterprise is extracted from the enterprise profile. When the real-time industry information is trending information and the real-time enterprise information indicates that the target enterprise is an enterprise with idle capacity, instant services are pushed to the target enterprise. The trending information includes at least one of policy implementation, commodity price change exceeding a change threshold, and target orders exceeding an order threshold being signed. The instant services include at least one of subsidy application services, order matching services, and financial credit services. When the real-time industry information is risk information, and the real-time enterprise information indicates that the target enterprise is a risky enterprise, risk control services are pushed to the target enterprise. The risk control services include industrial transformation consulting services, production reduction suggestions services, and environmental rectification suggestions services.

Images