Information disclosure text auditing method, computer device and computer storage medium

Abstract

Embodiments of the present application disclose an information disclosure text auditing method, a computer device and a computer storage medium. Embodiments of the present application include: a computer device obtains a target information disclosure text submitted by a requester, constructs a semantic tree of the target information disclosure text, constructs an auditing knowledge graph of the target information disclosure text, determines a comparison object of the requester according to the auditing knowledge graph, compares comparable attributes of the requester and the comparison object of the requester based on the auditing knowledge graph, obtains an attribute comparison result, audits the target information disclosure text according to the semantic tree of the target information disclosure text and the attribute comparison result, and obtains an auditing result of the target information disclosure text. Therefore, by quantitatively calculating the authenticity and accuracy of the information disclosure, the intelligentization and digitalization transformation of information disclosure auditing are promoted, and a foundation is laid for the economic development and prevention of systemic financial risks of service entities.

Description

Technical Field

[0001] This application relates to the field of data processing, specifically to a method for reviewing information disclosure texts, a computer device, and a computer storage medium. Background Technology

[0002] The exchange has released a list of key points regarding the completeness of information disclosure requirements, but there is no specific list or readily available quantitative calculation method for authenticity and accuracy. However, in review inquiries, authenticity and accuracy requirements account for a significant proportion. Inquiries explicitly mentioning "authenticity" and "accuracy" account for 25% of historical inquiries. Other inquiries that do not mention keywords but pertain to authenticity and accuracy, including those related to "qualifications," "penalties," "litigation," "honors," "actual controller," "related parties," and "industry positioning," account for approximately 30%. Inquiries related to "industry recognition," "industry development," "comparable companies," and "market position" also account for approximately 30%. Therefore, authenticity and accuracy constitute a large proportion of actual inquiries. Summary of the Invention

[0003] This application provides a method, computer equipment, and computer storage medium for reviewing information disclosure texts, used to review the authenticity and accuracy of information disclosure texts.

[0004] The first aspect of this application provides a method for reviewing information disclosure text, the method being applied to a computer device, the method comprising:

[0005] Obtain the target information disclosure text submitted by the requester, and construct a semantic tree of the target information disclosure text;

[0006] Construct a review knowledge graph of the target information disclosure text, and determine the comparison object of the requester based on the review knowledge graph;

[0007] Based on the audit knowledge graph, the requester and the comparison object of the requester are compared by comparable attributes to obtain the attribute comparison results;

[0008] The target information disclosure text is reviewed based on the semantic tree of the target information disclosure text and the attribute comparison results to obtain the review result of the target information disclosure text.

[0009] A second aspect of this application provides a computer device, the computer device comprising:

[0010] A construction unit is used to obtain the target information disclosure text submitted by the requester and construct a semantic tree of the target information disclosure text;

[0011] The construction unit is also used to construct an audit knowledge graph of the target information disclosure text, and determine the comparison object of the requester based on the audit knowledge graph;

[0012] The comparison unit is used to compare the comparable attributes of the requester and the comparison object of the requester based on the audit knowledge graph, and obtain the attribute comparison result.

[0013] The review unit is used to review the target information disclosure text based on the semantic tree of the target information disclosure text and the attribute comparison results, and obtain the review result of the target information disclosure text.

[0014] A third aspect of this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method of the first aspect described above.

[0015] A fourth aspect of this application provides a computer storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described in the first aspect.

[0016] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0017] In this embodiment, the computer device acquires the target information disclosure text submitted by the requester, constructs a semantic tree of the target information disclosure text, constructs a review knowledge graph of the target information disclosure text, determines the comparison object of the requester based on the review knowledge graph, compares the comparable attributes of the requester and the comparison object based on the review knowledge graph, obtains the attribute comparison results, and reviews the target information disclosure text based on the semantic tree of the target information disclosure text and the attribute comparison results, thereby obtaining the review result of the target information disclosure text. Therefore, by quantitatively calculating the review of the authenticity and accuracy of information disclosure, the intelligent and digital transformation of information disclosure review is promoted, laying the foundation for serving the economic development of the real economy and preventing systemic financial risks. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the information disclosure text review method in the embodiments of this application;

[0019] Figure 2 This is a schematic diagram illustrating the effect of converting the PDF file structure corresponding to the prospectus into a tree structure in an embodiment of this application.

[0020] Figure 3 This is a schematic diagram of the industry ontology structure in the embodiments of this application;

[0021] Figure 4This is a schematic diagram of the enterprise entity structure in the embodiments of this application;

[0022] Figure 5 This is a schematic diagram of the structure of a computer device in an embodiment of this application;

[0023] Figure 6 This is another schematic diagram of the structure of the computer device in the embodiments of this application. Detailed Implementation

[0024] This application provides a method, computer equipment, and computer storage medium for reviewing information disclosure texts, used to review the authenticity and accuracy of information disclosure texts.

[0025] The information disclosure text review method in this application embodiment is described below:

[0026] Please see Figure 1 One embodiment of the information disclosure text review method in this application includes:

[0027] 101. Obtain the target information disclosure text submitted by the requester, and construct a semantic tree of the target information disclosure text;

[0028] The method of this embodiment can be applied to computer devices, which can be servers, terminals, or other computer devices capable of performing data processing. When the computer device is a terminal, it can be a personal computer (PC), desktop computer, or other terminal device; when the computer device is a server, it can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud databases, cloud computing, big data, and artificial intelligence platforms.

[0029] This embodiment can be applied to any scenario in any field where information needs to be disclosed and the authenticity and accuracy of the disclosed information needs to be verified. For example, it can be applied to scenarios in the securities field where information about securities issuers needs to be disclosed, and the target information disclosure text can be a prospectus or other text.

[0030] One preferred approach to constructing a semantic tree of the target information disclosure text is to extract the directory structure of the target information disclosure text, extract multiple leaf nodes of the directory structure, and construct a semantic tree of the target information disclosure text based on the multiple leaf nodes of the directory structure.

[0031] For example, taking a prospectus as an example, a multi-layered processing framework can be designed based on its semantic characteristics. This framework abstracts common operators and capabilities to support research on data, algorithms, and models in the process of verifying the completeness and accuracy of the prospectus. The main idea is to locate and filter information layer by layer, from coarse-grained to fine-grained, from the table of contents, chapters, paragraphs, tables, sentences to fields, achieving layer-by-layer filtering and extraction from the entire document to the fields. The results of each layer are saved to facilitate processing requirements at various granularities during intelligent review, thereby achieving intelligent review. At the implementation level, rule engines, deep learning, and domain-specific knowledge can be introduced to support multi-layered processing flows.

[0032] Through the above processing, the target information disclosure text can be organized into a knowledge structure. Here, we still use the prospectus as an example. The tree-like directory structure of the prospectus is the backbone of this knowledge organization. Each leaf node's title contains its paragraphs and tables. After extraction and parsing, each paragraph and table corresponds to structured data. This design allows the PDF file structure to be converted into a tree structure, such as... Figure 2 Indication, Figure 2 The PDF file structure shown is converted into a tree structure, with each leaf node in the prospectus serving as a node in the tree structure.

[0033] Furthermore, to facilitate retrieval and calculation, this embodiment can convert the tree structure of the prospectus into a database table for storage. During storage, in addition to considering the aforementioned titles, paragraphs, tables, and corresponding structured data, it is also necessary to map the tree structure to the physical and logical information of elements in the PDF file. Elements in the PDF file are organized in text lines. Text lines include header lines, lines within paragraphs, and lines within tables. The physical information of a text line includes: document ID, text / table ID, chapter ID, page number ID, element coordinates on the page, and block height and width. The logical information of an element includes: table context, number of rows and columns in the table, table index on the current page, paragraph or table content, etc. The knowledge organization structure of the prospectus is shown in Table 1.

[0034] Table 1

[0035]

[0036] Therefore, knowledge organization instances from prospectuses can be stored in search engines. Combining search engines and PDF page front-end display technologies can provide a foundation for subsequent intelligent review, including text positioning, full-text PDF retrieval, and visualization.

[0037] 102. Construct a review knowledge graph of the target information disclosure text, and determine the comparison object of the requester based on the review knowledge graph;

[0038] In the securities sector, historical inquiry letters reveal the approach to industry analysis used in IPO reviews. These inquiries typically require issuers to accurately identify their industry, disclose their current research on industry position and competitive landscape, select accurate and complete comparable companies within the same industry, and compare them across various aspects such as business, technology, and finance. Therefore, in the review process, the analysis of comparable companies within the same industry is a crucial basis for reviewers to assess the truthfulness and accuracy of the issuer's information disclosure.

[0039] Industry elements refer to industry data identified from an industry research perspective that has significant reference value for industry analysis. These mainly include: business models and operations, market size, demand and potential, market competition landscape, industry barriers to entry, technological level and characteristics, typical enterprises, industry development trends, opportunities and challenges, industry cycles and seasonality, industry profit fluctuation trends and reasons, upstream and downstream industries, industry standards and setting companies, industry prosperity level, policies and regulations, and business models. Sources of industry element data include information disclosure documents, publicly released reports from third-party institutions, and news and public opinion published on key industry websites. The selection of industry element data emphasizes timeliness and authoritativeness.

[0040] Comparable companies refer to leading enterprises, domestic and international listed companies, and companies listed on the National Equities Exchange and Quotations (NEEQ) within a specific sub-sector. Comparable attributes include business registration information, industry attributes, related companies / individuals, key actors, financial attributes, and operational attributes. Specifically, business registration information includes company name, registered address, registered capital, and listing date; industry attributes include main business / industry, upstream related companies, downstream related companies, and peer companies; related companies and individuals include parent companies, investee companies, subsidiaries / grandchild companies, shareholders, and other companies controlled by the actual controller; key actors include the actual controller, board members, supervisory board members, and senior management, including attributes such as compensation, shareholding ratio, and equity incentives; financial attributes include the balance sheet, income statement, and cash flow statement. The balance sheet attributes include the debt-to-equity ratio, current ratio, and quick ratio; the income statement attributes include net profit margin and main business revenue; and the cash flow statement attributes include cash flows from operating, financing, and investing activities. Operational attributes include employees, compensation, honors, qualifications, penalties, and litigation.

[0041] In this embodiment, after constructing the review knowledge graph of the target information disclosure text, the review knowledge graph can be pruned. Specifically, the key-value pairs of each level entity in the review knowledge graph can be hashed and concatenated to calculate the hash code value of each level entity, and the review knowledge graph can be pruned according to the hash code value of each level entity.

[0042] The comparison of large amounts of data in the prospectus and knowledge graph involves a balance between performance and efficiency, especially the tree structure of the knowledge graph, such as the issuer's shareholders, holding companies, and affiliated companies. To avoid efficiency losses caused by large-scale Cartesian products, the data structure and key-value pair content of the knowledge graph need to be optimized. Specifically, node encoding and sorting are employed. This involves determining the corresponding nodes in the audit knowledge graph for the requester and the comparison object. For multi-level entities in the knowledge graph, the key-value pairs of each level entity are hashed, sorted, and then concatenated. The hash code value of each node in the audit knowledge graph is calculated. Considering the graph structure, the calculation of the hash code value of each node requires multiple iterations. The final calculation result is saved as an attribute of the semantic key or real-value key. During comparison, the hash code values ​​of each node in the audit knowledge graph are compared first. If they match, the nodes are considered essentially the same; otherwise, the next step of comparison is performed. After determining the hash values ​​of each node in the audit knowledge graph, the audit knowledge graph can be pruned based on the hash values ​​of each node, thereby reducing the comparison and calculation work in subsequent steps.

[0043] To address the issue of inconsistent textual attributes but semantic consistency among nodes, the pruning algorithm described above is further supplemented. First, the hash values ​​are compared for similarity. If the similarity falls within a specified threshold, the nodes are considered identical; otherwise, the subtrees of that node are further matched. For subtree matching, two methods are employed: one is direct subtree matching, which calculates the edit distance between subtrees. Subtrees whose distances meet a specified threshold are considered identical. The calculation method is as follows:

[0044]

[0045] The second method combines edit distance and semantic distance. If the combined distance meets a specified threshold range, the two methods are considered consistent. The calculation method is as follows:

[0046]

[0047] in, =min(

[0048] Among them, subtree semantic similarity The calculation method is as follows:

[0049] a. Sort all nodes in the subtree;

[0050] b. Obtain the distributed representation vector of the content of all nodes, which can be obtained based on the domain language model;

[0051] c. Concatenate the distributed representation vectors of each node to obtain the representation vector of the subtree;

[0052] d. Calculate the similarity of the subtree representation vectors.

[0053] 103. Based on the audit knowledge graph, compare the comparable attributes of the requester and the comparison object of the requester to obtain the attribute comparison results;

[0054] In the securities industry, comparing with comparable companies is a crucial way to describe an issuer's operations during the IPO review process. By comparing the issuer's financial indicators, operating indicators, accounting policies, business technology, products, and other indicators with those of comparable companies, the reviewer can identify the consistency and uniqueness of the issuer's market position and operating performance within the industry, thereby gaining a comprehensive understanding of the issuer. The prospectus includes sections on "Business and Technology" and "Financial Accounting Information" that introduce comparable companies in relevant content. Therefore, the completeness of the comparable companies and the absence of any anomalies in their quantitative indicators are key tasks in the comparable company review.

[0055] Industry classifications in knowledge graphs are generally based on products, technologies, and services, with a fine granularity. However, listed companies typically have multiple sub-sectors in their main business. Therefore, issuers exist in multiple industries, and within each industry, there are several companies in the same industry. Thus, sorting these companies in the same industry according to the similarity of their main business and obtaining the most similar comparable companies is the basis for reviewing the relevant content of comparable companies.

[0056] This embodiment proposes ranking companies in the same industry based on similarity in their main business, in order to accurately obtain the best comparable companies for the issuer. The specific method is as follows:

[0057] Obtain the issuer's financial data, customer data, main business revenue and its proportion, etc.

[0058] Obtain data from the knowledge graph, including financial data, customer data, main business revenue and percentage of revenue for all companies in the same industry;

[0059] The issuer's comparable companies were manually labeled;

[0060] Build a supervised learning model. The input is financial data, customer data, main business revenue and its proportion, etc. If it is text data, it can be encoded by a pre-trained model and output two analogies: "yes" and "no". In the process, feature extraction methods such as neural network models and attention mechanisms can be used.

[0061] Data augmentation and model optimization were performed on the corpus.

[0062] After identifying comparable companies for the issuer, their attributes can be compared, including business registration information, industry attributes, related enterprises / individuals, key actors, financial and operational attributes. There are two methods: for text-based data, a threshold can be calculated using text similarity; for numerical data, the average can be compared. Ultimately, the similarity and ranking of the issuer and comparable companies can be obtained, thus determining the issuer's position within the industry.

[0063] 104. Review the target information disclosure text based on the semantic tree of the target information disclosure text and the attribute comparison results to obtain the review result of the target information disclosure text;

[0064] In this embodiment, the specific method for reviewing the target information disclosure text based on the semantic tree and attribute comparison results of the target information disclosure text may be as follows: determine the industry classification of the requester's industry based on the review knowledge graph, determine the industry classification of the requester's industry disclosed in the target information disclosure text, compare the industry classification determined based on the review knowledge graph with the industry classification disclosed in the target information disclosure text to obtain the attribute comparison results, review the target information disclosure text based on the attribute comparison results, and obtain the review result of the target information disclosure text.

[0065] Therefore, this embodiment determines the issuer's industry positioning through the above method, then verifies the accuracy of the issuer's sector positioning, and compares it with the negative list of industries for the Growth Enterprise Market (GEM). If it belongs to a negative industry, the issuer's compliance with the listing sector requirements should be further evaluated. Issuer attributes, including industry, comparable company industries, regulatory authorities, and policies, are obtained from knowledge graphs and target information disclosure texts to construct an industry positioning calculation model. The specific method is as follows:

[0066] Information such as the issuer's industry, regulatory authorities, policies and regulations, key technologies, intellectual property, standards, sales revenue, and business model can be obtained from the knowledge graph.

[0067] Sales revenue should take into account the proportion of new business formats and technologies, and the use of new business formats and models in the business model.

[0068] Assess the issuer's innovation capabilities, such as whether the authorized patents are invention patents and whether they were acquired through assignment, and calculate the relevance of each patent application to the main business;

[0069] Assess the issuer's industry influence, such as whether the standards set by the issuer are national standards, or whether the standards set by the issuer are industry standards, group standards, or national standards;

[0070] For different listing boards, other characteristics can also be constructed, such as whether it belongs to the list of encouraged industries or whether it is an import substitute;

[0071] Construct a classification model with the above features as input and output the issuer's industry positioning status, including two labels: "compliant" and "non-compliant". The corpus can be manually labeled.

[0072] Based on the verification method described above in this embodiment, issues related to the authenticity and accuracy of the target information disclosure text can be identified, thereby prompting improvements in the quality of information disclosure and accurately alerting investors to risks.

[0073] Specifically, determining the industry classification of the requester's industry in the target information disclosure text can be achieved by: determining the business description and main business of the requester in the target information disclosure text, and predicting the industry classification of the requester's industry based on the business description and main business; or by determining the industry classification of the requester's comparison object's industry based on the pre-defined correspondence between industry objects and industries, and using the industry classification of the requester's comparison object's industry as the industry classification of the requester's industry.

[0074] For example, in the prospectus, the issuer's industry classification is clearly defined by the issuer itself. In addition, the issuer's industry classification can be obtained based on a knowledge graph and compared with the industry classification disclosed by the issuer. If the industry classification differs significantly, it can prompt a correction to the issuer's industry positioning. The main methods include the following:

[0075] The issuer's business description and main business can be obtained from the target information disclosure text. The business description can be predicted using an industry classification model, while the main business can be calculated using similarity and mapped to the industry classification in a knowledge graph.

[0076] Obtain the industry classification of the issuer of the knowledge graph and compare it with the industry classification disclosed in the prospectus;

[0077] The prospectus discloses comparable listed companies. The China Securities Regulatory Commission (CSRC) discloses the industry classification of all listed companies on its official website every quarter, which can be compared with the comparable companies disclosed in the prospectus, and thus with the industry classification disclosed by the issuer in the prospectus.

[0078] The prospectus typically discloses the relevant regulatory authority and industry policies. These can be compared with the industry information in the audit knowledge graph corresponding to the prospectus to determine its approximate industry.

[0079] For example, such as Figure 3 and Figure 4As shown, when analyzing the industry of the requester, you can refer to... Figure 3 The industry structure shown is broken down to analyze the requester's position and status within its industry, identifying its strengths and weaknesses. For example, if the requester belongs to an upstream industry group, its current situation can be analyzed from a micro perspective, considering factors such as market size, business model, industry barriers, and financing events, to determine its position, strengths, and weaknesses within the industry. Alternatively, it can be analyzed from a macro perspective, considering factors such as opportunities and challenges, policies and regulations, industry standards, industry prosperity, and industry trends. Similarly, benchmarking analysis can be conducted, examining its financial situation, technological R&D capabilities, inventory capacity, and other aspects, ultimately revealing the requester's position, strengths, and weaknesses within the industry based on multiple different dimensions.

[0080] Similarly, the analysis of the requester can be based on the enterprise itself. The enterprise's development capabilities can be analyzed from multiple aspects such as its development stage, business registration information, operating conditions, related parties, and innovation capabilities, thereby determining the enterprise's competitive position, competitive advantages, and disadvantages in the industry.

[0081] This embodiment proposes for the first time a process and paradigm for quantitative calculation of information disclosure, as well as a knowledge graph for information disclosure review and a quantitative calculation method for the authenticity of information disclosure. This promotes the intelligent and digital transformation of information disclosure review and lays the foundation for serving the economic development of the real economy and preventing systemic financial risks.

[0082] The information disclosure text review method in the embodiments of this application has been described above. The computer equipment in the embodiments of this application is described below. Please refer to [link / reference]. Figure 5 One embodiment of the computer device in this application includes:

[0083] Construction unit 301 is used to obtain the target information disclosure text submitted by the requester and construct the semantic tree of the target information disclosure text;

[0084] The construction unit 301 is also used to construct an audit knowledge graph of the target information disclosure text, and determine the comparison object of the requester based on the audit knowledge graph;

[0085] The comparison unit 302 is used to compare the comparable attributes of the requester and the comparison object of the requester based on the audit knowledge graph, and obtain the attribute comparison result.

[0086] The review unit 303 is used to review the target information disclosure text based on the semantic tree of the target information disclosure text and the attribute comparison results, and obtain the review result of the target information disclosure text.

[0087] In a preferred embodiment of this example, the review unit 303 is specifically used to determine the industry classification of the industry in which the requester is located based on the review knowledge graph; determine the industry classification of the industry in which the requester is located as disclosed in the target information disclosure text; compare the industry attributes based on the review knowledge graph with the industry attributes disclosed in the target information disclosure text to obtain the attribute comparison result; and review the target information disclosure text according to the attribute comparison result to obtain the review result of the target information disclosure text.

[0088] In a preferred embodiment of this example, the review unit 303 is specifically used to determine the business description and main business of the requester in the target information disclosure text, and predict the industry classification of the industry in which the requester operates based on the business description and main business; or, determine the industry classification of the industry in which the requester's comparison object operates based on a preset correspondence between industry objects and industries, and use the industry classification of the industry in which the requester's comparison object operates as the industry classification of the requester's industry; or, determine the industry classification of the requester's industry based on one or more factors among the industry regulatory agencies, intellectual property rights, industry policies, and laws in which the requester operates.

[0089] In a preferred embodiment of this example, the construction unit 301 is specifically used to extract the directory structure of the target information disclosure text from the target information disclosure text, extract multiple leaf nodes of the directory structure, and construct a semantic tree of the target information disclosure text based on the multiple leaf nodes of the directory structure.

[0090] In a preferred embodiment of this invention, the computer device further includes:

[0091] The processing unit 304 is used to perform hash encoding and concatenation on the key-value pairs of each level entity in the audit knowledge graph, calculate the hash code value corresponding to each node in the audit knowledge graph, and prune the audit knowledge graph according to the hash code value corresponding to each node in the audit knowledge graph.

[0092] In this embodiment, the operations performed by each unit in the computer device are the same as described above. Figure 1 The embodiments shown are similar and will not be repeated here.

[0093] This embodiment proposes for the first time a process and paradigm for quantitative calculation of information disclosure, as well as a knowledge graph for information disclosure review and a quantitative calculation method for the authenticity of information disclosure. This promotes the intelligent and digital transformation of information disclosure review and lays the foundation for serving the economic development of the real economy and preventing systemic financial risks.

[0094] The computer device in the embodiments of this application is described below. Please refer to [link / reference]. Figure 6 One embodiment of the computer device in this application includes:

[0095] The computer device 400 may include one or more central processing units (CPUs) 401 and a memory 405, in which one or more applications or data are stored.

[0096] The memory 405 can be volatile or persistent storage. The program stored in the memory 405 can include one or more modules, each module including a series of instruction operations on the computer device. Furthermore, the central processing unit 401 can be configured to communicate with the memory 405 and execute the series of instruction operations stored in the memory 405 on the computer device 400.

[0097] Computer device 400 may also include one or more power supplies 402, one or more wired or wireless network interfaces 403, one or more input / output interfaces 404, and / or one or more operating systems, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0098] The central processing unit 401 can perform the aforementioned... Figure 1 The specific operations performed by the computer device in the illustrated embodiment will not be described in detail here.

[0099] This application also provides a computer storage medium, one embodiment of which includes: the computer storage medium storing instructions, which, when executed on a computer, cause the computer to perform the aforementioned... Figure 1 The operations performed by the computer device in the illustrated embodiment.

[0100] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0101] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0102] 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0103] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0104] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or 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 this application. 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.

Claims

1. A method for reviewing information disclosure texts, characterized in that, The method is applied to a computer device, and the method includes: Obtain the target information disclosure text submitted by the requester, and construct a semantic tree of the target information disclosure text; the semantic tree is constructed based on multiple leaf nodes in the directory structure of the target information disclosure text; A review knowledge graph of the target information disclosure text is constructed, and the comparison object of the requester is determined based on the review knowledge graph. The review knowledge graph is optimized by pruning based on the hash code values ​​of its entities at each level. The hash code values ​​of the entities at each level of the review knowledge graph are obtained by hash encoding and concatenating the key-value pairs of each entity at each level. The specific steps are as follows: determine the nodes in the review knowledge graph corresponding to the requester and the comparison object of the requester, respectively; for the multi-level entities in the knowledge graph, hash encode and sort the key-value pairs of each entity at each level and then concatenate them; calculate the hash code value of each node in the review knowledge graph; the calculation of the hash code value of each node is iterated multiple times, and the calculation result is saved as an attribute of the semantic key or real value key. Based on the audit knowledge graph, the requester and the comparison object of the requester are compared by comparable attributes to obtain the attribute comparison results; The target information disclosure text is reviewed based on the semantic tree of the target information disclosure text and the attribute comparison results to obtain the review result of the target information disclosure text; The step of reviewing the target information disclosure text based on the semantic tree of the target information disclosure text and the attribute comparison results to obtain the review result of the target information disclosure text includes: The industry classification of the requester's industry is determined based on the aforementioned audit knowledge graph; Determine the industry classification of the industry in which the requesting party is located in the target information disclosure text; The industry classification determined based on the audit knowledge graph is compared with the industry classification disclosed in the target information disclosure text to obtain the attribute comparison result; The target information disclosure text is reviewed based on the semantic tree of the target information disclosure text and the attribute comparison results to obtain the review result of the target information disclosure text.

2. The method according to claim 1, characterized in that, The determination of the industry classification of the requesting party in the target information disclosure text includes: Determine the business description and main business of the requesting party in the target information disclosure text, and predict the industry classification of the industry in which the requesting party operates based on the business description and the main business; or, The industry classification of the industry in which the comparison object of the requester is located is determined based on the preset correspondence between industry objects and industries, and the industry classification of the industry in which the comparison object of the requester is located is taken as the industry classification of the industry in which the requester is located. Alternatively, the industry classification of the industry in which the requester operates may be determined based on one or more of the following factors: industry regulatory agencies, intellectual property rights, industry policies, and laws.

3. A computer device, characterized in that, The computer device includes: A construction unit is used to obtain the target information disclosure text submitted by the requester and construct a semantic tree of the target information disclosure text; the semantic tree is constructed based on multiple leaf nodes in the directory structure of the target information disclosure text. The construction unit is also used to construct the review knowledge graph of the target information disclosure text, and determine the comparison object of the requester based on the review knowledge graph; the review knowledge graph is optimized by pruning based on the hash code values ​​of its various level entities. The hash code values ​​of the various level entities of the review knowledge graph are obtained by hash encoding and concatenating the key-value pairs of each level entity. The specific steps are as follows: determine the nodes in the review knowledge graph corresponding to the requester and the comparison object of the requester; for the multi-level entities in the knowledge graph, hash encode and sort the key-value pairs of each level entity and then concatenate them; calculate the hash code value of each node in the review knowledge graph; the calculation of the hash code value of each node is iterated multiple times, and the calculation result is saved as an attribute of the semantic key or real value key; The comparison unit is used to compare the comparable attributes of the requester and the comparison object of the requester based on the audit knowledge graph, and obtain the attribute comparison result. The review unit is used to review the target information disclosure text based on the semantic tree of the target information disclosure text and the attribute comparison results, and obtain the review result of the target information disclosure text. The review unit is specifically used to determine the industry classification of the requester's industry based on the review knowledge graph; determine the industry classification of the requester's industry disclosed in the target information disclosure text; compare the industry classification determined based on the review knowledge graph with the industry classification disclosed in the target information disclosure text to obtain the attribute comparison result; and review the target information disclosure text according to the semantic tree of the target information disclosure text and the attribute comparison result to obtain the review result of the target information disclosure text.

4. The computer device according to claim 3, characterized in that, The review unit is specifically used to determine the business description and main business of the requester in the target information disclosure text, and predict the industry classification of the industry in which the requester operates based on the business description and main business; or, determine the industry classification of the industry in which the requester's comparison object operates based on a preset correspondence between industry objects and industries, and use the industry classification of the industry in which the requester's comparison object operates as the industry classification of the requester's industry; or, determine the industry classification of the requester's industry based on one or more factors among the industry regulatory agencies, intellectual property rights, industry policies, and laws in which the requester operates.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 2.

6. A computer storage medium, characterized in that, The computer storage medium stores instructions that, when executed on the computer, cause the computer to perform the method as described in any one of claims 1 to 2.

Images