Product information auditing method, device and equipment, medium and program product

Abstract

The invention provides a product information auditing method which can be applied to the technical field of artificial intelligence. The method comprises the following steps: acquiring a standard information message of a preset structure corresponding to product information to be audited; according to the standard information message, executing verification processing based on a preset rule to obtain a first auditing result; according to the standard information message, executing natural language processing and logic comparison processing based on a preset knowledge graph to obtain a second auditing result; and inputting the first auditing result and the second auditing result into a correction recommendation engine, so that the correction recommendation engine corrects the auditing result based on a correction rule of a preset error correction strategy mapping map, and obtains and outputs a correction suggestion and an auditing result of the to-be-audited product information. The invention further provides a product information auditing device and equipment, a storage medium and a program product.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically to a product information verification method, apparatus, equipment, medium, and program product. Background Technology

[0002] With the digital transformation of financial asset management, the review of financial product information faces significant challenges. New product launches involve massive amounts of complex information, encompassing the interaction and validation of numerous unstructured documents and structured fields. Currently, the industry primarily employs a "human review + traditional rule engine" work model. However, human review is limited by physiological constraints, exhibiting significant efficiency bottlenecks during peak product launch periods. Furthermore, cognitive fatigue can easily lead to overlooking minor formatting differences or implicit logical errors, making it difficult to ensure consistent review standards. Existing automated assistance technologies largely rely on rigid preset logic or simple regular expression matching, primarily limited to superficial format validity and numerical range checks. They are often ineffective against deep business semantic conflicts within unstructured text. More critically, existing systems typically only passively intercept errors, lacking the ability to proactively provide corrective suggestions based on context and industry standards. This results in high interaction costs for data correction, and the system rules are fragmented, failing to dynamically accumulate knowledge and self-optimize as business evolves. Summary of the Invention

[0003] In view of the above issues, this application provides product information verification methods, apparatus, equipment, media, and procedures.

[0004] According to a first aspect of this application, a product information review method is provided, the method comprising: acquiring a standard information message corresponding to a preset structure of product information to be reviewed; performing verification processing based on preset rules according to the standard information message to obtain a first review result; performing natural language processing and logical comparison processing based on a preset knowledge graph according to the standard information message to obtain a second review result; inputting the first review result and the second review result into a correction recommendation engine, causing the correction recommendation engine to perform correction processing on the review result based on correction rules of a preset error correction strategy mapping graph, and obtaining and outputting correction suggestions and review results for the product information to be reviewed.

[0005] According to an embodiment of this application, obtaining the standard information message corresponding to the preset structure of the product information to be reviewed includes: obtaining the product information to be reviewed; and performing field mapping extraction processing on the product information to be reviewed based on a preset field mapping relationship to obtain the standard information message corresponding to the preset standard structure.

[0006] According to an embodiment of this application, the step of performing a verification process based on preset rules to obtain a first audit result based on the standard information message includes: performing a field verification process based on preset format rules to obtain a format verification result based on the standard information message; performing an association comparison process based on preset logical rules to obtain a logical verification result based on the standard information message; and forming the first audit result based on the format verification result and the logical verification result.

[0007] According to an embodiment of this application, the execution of association comparison processing based on preset logical rules includes at least: executing logical comparison processing based on date order rules to obtain a date logical status; executing logical comparison processing based on business compliance rules to obtain a business compliance status, wherein the business compliance rules are used to verify the constraint relationship between product type and investment ratio in the standard information message; and executing logical comparison processing based on field association consistency rules to obtain an association consistency status, wherein the field association consistency rules are used to verify the dependency relationship between product name field and product type field in the standard information message.

[0008] According to an embodiment of this application, the step of performing natural language processing and logical comparison processing based on a preset knowledge graph to obtain a second audit result based on the standard information message includes: performing natural language processing on the text fields in the standard information message to extract key entities; performing knowledge graph query processing based on the key entities to generate a compliance status; and generating the second audit result based on the compliance status.

[0009] According to an embodiment of this application, the process of querying the knowledge graph includes at least: performing logical comparison processing based on asset type compliance to obtain asset conflict status; and performing logical comparison processing based on a disabled term library to obtain text violation status.

[0010] According to an embodiment of this application, the correction recommendation engine corrects the review results based on correction rules of a preset error correction strategy mapping graph, including: performing modification content generation processing based on the preset error correction strategy mapping graph according to the error marker content and error code in the first review result and the error marker content and error code in the second review result to obtain modification suggestion content; and performing correction processing on the data corresponding to the error marker content based on the modification suggestion content to generate the correction suggestion.

[0011] According to an embodiment of this application, the modification suggestion content includes numerical range suggestions and text completion suggestions. The step of generating modification content based on a preset error correction strategy mapping graph includes at least: performing numerical recommendation processing based on statistical distribution to obtain numerical modification suggestions, wherein the numerical recommendation processing based on statistical distribution includes querying similar product data in the knowledge graph based on the error code, performing statistical calculation processing, and obtaining the numerical range suggestions; performing pattern matching processing based on the knowledge graph to obtain text modification suggestions, wherein the pattern matching processing of the knowledge graph includes querying naming convention templates in the knowledge graph based on the error code, performing pattern matching processing, and obtaining text completion suggestions.

[0012] According to an embodiment of this application, the method further includes: receiving feedback information on the correction suggestions for the product information to be reviewed; and performing processing to update the preset rules and the knowledge graph based on the feedback information.

[0013] A second aspect of this application provides a product information review device, comprising: a first acquisition module configured to acquire a standard information message with a preset structure corresponding to the product information to be reviewed; a first processing module configured to perform verification processing based on preset rules according to the standard information message to obtain a first review result; a second processing module configured to perform natural language processing and logical comparison processing based on a preset knowledge graph according to the standard information message to obtain a second review result; and a third processing module configured to input the first review result and the second review result into a correction recommendation engine, so that the correction recommendation engine performs correction processing on the review result based on the correction rules of a preset error correction strategy mapping graph, and obtains and outputs correction suggestions and review results for the product information to be reviewed.

[0014] A third aspect of this application provides an electronic device comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0015] A fourth aspect of this application also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implement the steps of the above-described method.

[0016] The fifth aspect of this application also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method. Attached Figure Description

[0017] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0018] Figure 1 The illustrations depict application scenarios of product information verification methods, apparatus, devices, media, and programs according to embodiments of this application.

[0019] Figure 2 A flowchart illustrating a product information verification method according to an embodiment of this application is shown schematically.

[0020] Figure 3 A flowchart illustrating step S100 of the product information verification method according to an embodiment of this application is shown schematically.

[0021] Figure 4 A flowchart illustrating step S200 of the product information verification method according to an embodiment of this application is shown schematically.

[0022] Figure 5 A flowchart illustrating step S300 of the product information verification method according to an embodiment of this application is shown schematically.

[0023] Figure 6 A flowchart illustrating step S400 of the product information verification method according to an embodiment of this application is shown schematically.

[0024] Figure 7 A supplementary flowchart illustrating a product information verification method according to an embodiment of this application is shown schematically.

[0025] Figure 8 This schematic diagram illustrates the structural block diagram of a product information verification device according to an embodiment of this application;

[0026] as well as

[0027] Figure 9 A block diagram schematically illustrates an electronic device suitable for implementing a product information verification method according to an embodiment of this application. Detailed Implementation

[0028] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0029] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0030] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0031] When using expressions such as "at least one of A, B, and C," the meaning should generally be interpreted according to the understanding of someone skilled in the art. For example, "a system having at least one of A, B, and C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C. Similarly, when using expressions such as "at least one of A, B, or C," the meaning should generally be interpreted according to the understanding of someone skilled in the art. For example, "a system having at least one of A, B, or C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C.

[0032] The product information verification method, apparatus, equipment, medium, and program products of this application can be used in the field of artificial intelligence technology, and can also be used in any field other than artificial intelligence. The application fields of the product information verification method, apparatus, equipment, medium, and program products of this application are not limited.

[0033] In the technical solution of this application, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, application, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0034] In scenarios involving automated decision-making using personal information, the methods, devices, and systems provided in this application all offer users corresponding entry points for choosing to agree to or reject the automated decision-making results. If the user chooses to reject, the process proceeds to the expert decision-making stage. Here, "automated decision-making" refers to the activity of automatically analyzing and evaluating an individual's behavioral habits, interests, or economic, health, and credit status through computer programs, and then making a decision. Here, "expert decision-making" refers to the activity of making decisions by personnel who specialize in a particular field, possess specialized experience, knowledge, and skills, and have reached a certain level of professional expertise.

[0035] With the development of the financial asset management market and the deepening of digital transformation, the number of wealth management products issued by commercial banks, fund companies, and other financial institutions has exploded, and the complexity and innovation of product design have also been continuously increasing. The launch and distribution process for new products requires processing massive amounts of product information data. This data typically includes dozens or even hundreds of subdivided fields such as product name, code, risk level, fee structure, investment objective, and performance benchmark, and the data source is often unstructured or semi-structured documents (such as prospectuses and legal opinions). Currently, to ensure the accuracy and completeness of product information and compliance with regulatory requirements and internal risk control, financial institutions mainly rely on a large number of business personnel for multi-level manual review. However, faced with the ever-increasing scale of data and timeliness requirements, this traditional manual operation mode has gradually revealed obvious limitations: On the one hand, manual verification of massive fields word by word leads to low review efficiency, with review time increasing linearly with the amount of data, making it difficult to cope with the peak product release at the end of the quarter or year and easily becoming a bottleneck in the business process; on the other hand, limited by human cognitive fatigue and subjective experience differences, manual review cannot guarantee the consistency of standards, and is prone to overlooking minor differences such as "full-width / half-width spaces" and "errors with similar-looking characters," and it is difficult to conduct in-depth verification of complex business logic across fields in a short period of time, resulting in the omission of high-risk compliance errors, which brings potential operational risks and reputational damage to financial institutions.

[0036] While some rule-engine-based assisted auditing systems exist to automate manual checks to some extent, these technologies still have significant technical limitations when dealing with complex financial product information, failing to meet the needs of in-depth intelligent auditing. First, existing auditing technologies are mostly based on rigid preset rules (such as regular expressions and simple database comparisons), capable of only performing superficial format checks (e.g., checking if a code is a 6-digit number or if a date format is correct) or simple numerical range comparisons. They lack the ability to understand the semantics of unstructured text fields and cannot perceive context, making it difficult to detect deep semantic conflicts or hidden business compliance risks, such as "low-risk products being configured with high-risk investment targets." Second, existing systems typically employ a "blocking error reporting" mechanism, which only identifies and intercepts erroneous data but cannot understand the root cause of the error, let alone proactively generate constructive modification suggestions (such as automatically recommending standardized product names or reasonable fee ranges) based on industry standards or business knowledge. This leads to repeated trial and error by operators, resulting in low efficiency. In addition, existing auditing systems often lack knowledge accumulation and self-evolution capabilities. Auditing rules and industry knowledge are in a discrete state, and it is impossible to build a network of relationships between entities (knowledge graph). As a result, the system cannot be dynamically optimized as business rules change or historical data accumulates, making it difficult to adapt to rapidly changing financial regulatory requirements and business innovation scenarios.

[0037] Figure 1 The illustration shows an application scenario of the product information verification method, apparatus, device, medium, and program product according to embodiments of this application.

[0038] like Figure 1 As shown, application scenario 100 according to this embodiment may include a product information verification method, apparatus, device, medium, and program product. Network 104 serves as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0039] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as financial service applications, web browser applications, search applications, email clients, social media platform software, etc. (for example only).

[0040] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0041] Server 105 can be a server that provides various services, such as a backend management server that supports information browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0042] It should be noted that, in the first aspect, the product information verification method provided in this application embodiment can be executed by server 105; in the second aspect, the product information verification method provided in this application embodiment can be executed by first terminal device 101, second terminal device 102, or third terminal device 103. Correspondingly, in the third aspect, the product information verification device provided in this application embodiment can be located in server 105; in the fourth aspect, the product information verification device provided in this application embodiment can be located in first terminal device 101, second terminal device 102, or third terminal device 103. The product information verification method provided in this application embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with first terminal device 101, second terminal device 102, third terminal device 103, and / or server 105. Correspondingly, the product information verification device provided in this application embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with first terminal device 101, second terminal device 102, third terminal device 103, and / or server 105.

[0043] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0044] The following will be based on Figure 1 The described scene, through Figures 2-7 The product information verification method of the application embodiment is described in detail.

[0045] Figure 2 A flowchart illustrating a product information verification method according to an embodiment of this application is shown.

[0046] like Figure 2 As shown, this application provides a product information review method, including operations S100 to S400.

[0047] Specifically, in operation S100, a standard information message corresponding to the preset structure of the product information to be reviewed is obtained. In some specific embodiments, the preset structure is a standardized data model that has been strictly defined and can be directly parsed and processed by a computer program. The preset structure specifies in detail the data attributes and storage specifications of various business elements in the product information. Specifically, this includes, but is not limited to, strict definitions of data types. For example, quantitative indicators such as rates and amounts are defined as numerical data, descriptive content such as product names and investment strategies are defined as string data, and time nodes such as fundraising start date and establishment date are defined as date data. In addition, the preset structure also embeds basic semantic constraints and business logic restrictions for each field, such as the valid range of numerical values, the maximum length limit of text, and specific format requirements. In subsequent processing steps, the standard information message corresponding to the preset structure plays the role of a unified benchmark. As the sole standard input source for subsequent rule verification, semantic analysis, and correction recommendation processes, the standard information message effectively shields the diversity of original data sources and the complexity of formats, ensuring the universality and consistency of subsequent processing logic. From a macroscopic structural perspective, a standard information message can be constructed as a standardized data object containing multiple key-value pairs, where each key corresponds to a unique standard field identifier, and each value corresponds to specific business data, thus forming a structured, organized digital archive containing a full range of business information.

[0048] like Figure 3 As shown, in some exemplary embodiments, obtaining the standard information message with a preset structure corresponding to the product information to be reviewed includes: operation S110 to operation S120.

[0049] Specifically, in operation S110, the product information to be reviewed is obtained. In some specific embodiments, the product information to be reviewed mainly covers various original business data generated during the design, issuance, and ongoing management of financial products. These various original business data usually exist in the form of unstructured or semi-structured electronic files, such as common word processing documents, spreadsheet files, portable document format files, and even scanned image files. For ease of understanding, for example, the product information to be reviewed specifically includes various data items describing the core characteristics of the financial product: in terms of basic information, it involves the full name of the product, product code, operating mode, risk and return characteristic level, etc.; in terms of numerical parameters, it involves various fee structures (such as management fees, custody fees), capital thresholds (such as minimum investment amount, additional investment amount), maximum fundraising scale, etc.; in terms of text description, it involves the description of investment objectives, investment strategies, investment scope, performance benchmark, introduction of management personnel, and legal compliance statements, etc. This information is often scattered across different types of documents such as prospectuses, product information tables, and legal opinions. Furthermore, the arrangement and naming conventions of these different documents may vary significantly, requiring unified extraction and integration through subsequent steps.

[0050] Specifically, in operation S120, the product information to be reviewed is processed by field mapping extraction based on a preset field mapping relationship to obtain the standard information message corresponding to the preset standard structure. In some specific embodiments, during the execution process, the product information to be reviewed first needs to be parsed to identify its internal data organization form, such as identifying the header row of a table or the paragraph tags of a document. Subsequently, key information is extracted based on the preset field mapping relationship. The preset field mapping relationship is a pre-built mapping dictionary containing multiple synonyms or similar words, which defines the correspondence rules between standard field identifiers and multiple possible external field names. In some specific embodiments, when performing field mapping extraction processing, a fuzzy matching algorithm based on character similarity can be used to calculate the similarity distance between the original field names in the product information to be reviewed and the preset standard field names, thereby enabling the identification and association of fields that are semantically identical but have subtle differences in expression (for example, mapping "hosting fee", "hosting fee rate", or English expressions to the same internal standard field). After completing the mapping and association, the extracted data content needs to be further cleaned and normalized. This includes, but is not limited to, removing redundant whitespace characters and invisible characters at the beginning and end of the data, converting full-width characters into half-width characters to eliminate encoding differences, using date parsing logic to convert diverse date writing formats into standard date formats (such as "year-month-day" format), and converting percentage values ​​and monetary values ​​in different units into standard pure decimal or basic currency unit values. Finally, a standard information message that meets the preset structure requirements is assembled.

[0051] Specifically, in operation S200, based on the standard information message, a verification process based on preset rules is executed to obtain the first audit result. In some specific embodiments, operation S200 aims to construct a "rigid constraint layer" in the audit process, utilizing deterministic logic that can be quickly executed by computer programs to perform preliminary quality cleaning and compliance filtering on the data. Product information audits typically face the pressure of processing massive amounts of data. Directly performing deep semantic analysis would consume enormous computing resources and be time-consuming, while most basic errors (such as format errors and explicit logical contradictions) can be quickly intercepted through preset rules. By setting up this rule-based firewall before deep analysis, data that does not conform to basic specifications can be effectively filtered out from the source, ensuring that the data entering the deep processing stage has basic parsability and logical consistency, thereby significantly reducing the overall computational overhead of the system and greatly improving the efficiency and response speed of the entire audit process.

[0052] like Figure 4 As shown, in some exemplary embodiments, the step of performing verification processing based on preset rules according to the standard information message to obtain a first audit result includes: operation S210 to operation S230.

[0053] Specifically, in operation S210, based on the standard information message, field validation processing based on preset format rules is performed to obtain the format validation result. In some specific embodiments, performing field validation processing based on preset format rules involves performing a compliance scan on each smallest data unit (i.e., field) in the standard information message. The principle is to compare the actual data characteristics of each field with the standard feature templates defined in the preset format rule library. The preset format rules define in detail the physical form constraints that different types of fields should meet, such as field length limits, character type composition (e.g., whether only numbers, only Chinese characters, or specific encoding formats are allowed), and format specifications for special data items (e.g., dates must conform to the "YYYY-MM-DD" standard, currency codes must conform to the international standard three-letter code, etc.). In some specific embodiments, when a "product code" field used to identify a product is scanned, the corresponding length and type rules are invoked to verify whether the field value is strictly composed of numeric characters of a specific number of digits; if the field is found to contain non-numeric characters (such as letters or special symbols) or the length does not meet the preset number of digits, the field format validation is determined to have failed, and a corresponding format error status identifier is generated.

[0054] Specifically, in operation S220, based on the standard information message, a correlation comparison process based on preset logical rules is performed to obtain a logical verification result. In some specific embodiments, operation S220 elevates from the morphological verification of a single field to the logical verification level of the relationships between fields. Its core principle is to establish cross-field constraint relationship expressions based on the inherent logic of financial business and the objective laws of the physical world; by extracting the values ​​or attributes of multiple related fields in the standard information message and substituting them into preset logical verification formulas for calculation or comparison, it verifies whether there are contradictions between the data. The correlation comparison process based on preset logical rules does not rely on complex semantic understanding, but is based on deterministic mathematical logic or conditional judgment logic to discover hidden errors that, although the individual field formats are correct, violate business common sense or regulatory provisions when combined, thereby ensuring the consistency and rationality of product information at the logical level.

[0055] In some exemplary embodiments, the execution of the association comparison processing based on preset logical rules includes at least: executing logical comparison processing based on date order rules to obtain a date logical status; executing logical comparison processing based on business compliance rules to obtain a business compliance status, wherein the business compliance rules are used to verify the constraint relationship between product type and investment ratio in the standard information message; and executing logical comparison processing based on field association consistency rules to obtain an association consistency status, wherein the field association consistency rules are used to verify the dependency relationship between product name field and product type field in the standard information message.

[0056] In some specific embodiments, performing logical comparison processing based on date order rules verifies the temporal logic of multiple time nodes involved in the product lifecycle. First, fields with time attributes, such as "fundraising start date" and "fundraising end date," are extracted from standard information messages. These date fields are then converted into comparable timestamp values. Based on the principle of the irreversibility of time flow, a comparison expression is constructed, such as "the start date must be less than or equal to the end date." If the extracted timestamp value of the "fundraising start date" is detected to be greater than the timestamp value of the "fundraising end date," indicating a logical paradox of time reversal, a date logic error is immediately determined, and a corresponding abnormal status identifier is generated. Performing logical comparison processing based on date order rules effectively prevents temporal logic confusion caused by input negligence.

[0057] In some specific embodiments, performing logical comparison processing based on business compliance rules involves transforming regulatory provisions or internal risk control policies into executable conditional statements to constrain the matching relationship between specific product types and their core business parameters. Regarding the constraint relationship between "product type" and "investment ratio," the specific type identifier of the currently reviewed product is first identified (e.g., "bond fund"), and then the corresponding compliance parameter threshold (e.g., "maximum equity investment ratio") is queried. Next, the declared investment ratio value in the product information is extracted and compared with the queried threshold. If the product type is identified as a conservative, low-risk product, but its declared high-risk asset investment ratio exceeds the preset compliance limit, it is determined that the business parameter configuration violates the compliance rules, thus outputting a business compliance verification failure status.

[0058] In some specific embodiments, the logical comparison processing based on field association consistency rules utilizes keyword matching and dependency mapping to verify whether the description of the product name matches the actually defined product attributes. This logical comparison processing involves scanning the "Product Name" field for keywords and simultaneously obtaining the attribute value of the "Product Type" field. A preset rule base defines strong dependencies between name keywords and product types. For example, if a product name contains specific highly recognizable terms (such as "currency" or "index"), its product type field must be selected as the corresponding specific category. If the product name contains the word "currency," but the product type field is marked as "equity" or "hybrid," a consistency conflict is determined, indicating a contradiction between the product's characterization and naming, thus generating an anomalous consistency status.

[0059] Specifically, in operation S230, the first audit result is formed based on the format verification result and the logical verification result. In some specific embodiments, operation S230 is a process of summarizing and structurally encapsulating the status data produced by each of the above verification stages. The collected states of each verification pass or fail are integrated. For items that fail verification, not only is the failure status marker recorded, but also the specific error metadata is associated. This metadata typically includes: the specific field identifier locating the error, the rule ID that triggered the error, and the standardized error code (e.g., "ERR_NAME_FORMAT" represents a name format error, and "ERR_LOGIC_CONFLICT" represents a logical contradiction). The final generated first audit result is a structured data object containing the verification status of all fields and detailed error information. The first audit result clearly distinguishes which data is compliant and which data violates rigid rules, providing precise "targets" for the subsequent intelligent correction recommendation engine, enabling subsequent steps to match corresponding repair strategies to specific error codes.

[0060] Specifically, in operation S300, natural language processing and logical comparison processing based on a preset knowledge graph are performed according to the standard information message to obtain a second review result. In some specific embodiments, operation S300 constructs a "flexible understanding layer" in the review process, aiming to solve the problem of unstructured text semantic understanding and implicit business logic verification that traditional rule engines struggle to handle. Financial product information contains a large amount of unstructured text such as investment strategy descriptions and risk disclosure statements. The business compliance contained in this text cannot be verified by simple numerical comparison or format scanning alone; its context and deep semantics must be understood. By introducing natural language processing technology to transform ambiguous natural language into machine-recognizable structured entities, and combining it with a complex entity relationship network constructed by a preset knowledge graph for deep logical deduction, deep compliance risks such as "semantic conflicts between product type and investment target" or "implicit non-compliant commitments" can be effectively identified, thereby significantly improving the depth and breadth of the review and ensuring the compliance of product information at the business substance level.

[0061] like Figure 5 As shown, in some exemplary embodiments, the step of performing natural language processing and logical comparison processing based on a preset knowledge graph according to the standard information message to obtain a second audit result includes: operations S310 to S330.

[0062] Specifically, in operation S310, natural language processing is performed on the text fields in the standard information message to extract key entities. In some specific embodiments, operation S310 is a crucial step in converting unstructured natural language into structured semantic information. First, for descriptive text fields (such as "investment objectives," "investment scope," etc.) in the standard information message, a pre-trained semantic analysis model is invoked for word segmentation and part-of-speech tagging. Subsequently, named entity recognition processing is performed, scanning the text sequence to locate and extract noun phrases with specific business meanings, i.e., the "key entities." These key entities specifically include, but are not limited to, financial asset classes (such as "stocks," "bonds"), market sectors (such as "Science and Technology Innovation Board"), and descriptive terms of operating methods. While extracting entities, semantic dependency analysis is further performed to parse the grammatical role of the entity in the sentence (such as as the object of "investing in"), thereby eliminating irrelevant words and accurately identifying the core objects that can represent the business essence, providing precise input anchors for subsequent knowledge graph queries.

[0063] Specifically, in operation S320, based on the key entities, the knowledge graph is queried to generate a compliance status. In some specific embodiments, operation S320 is a process of logical reasoning using a pre-constructed digital knowledge network. The knowledge graph pre-stores entities in the financial field (such as product types, asset classes), attribute relationships between entities (such as "belongs to" and "includes"), and business rule constraints (such as "allowed investment" and "prohibited investment"). During processing, each key entity extracted in the previous steps is first mapped to a corresponding node in the knowledge graph. Then, based on the attribute node of the currently reviewed product (such as the "product type" node), a path query is initiated in the graph network to retrieve whether there is a preset association path or attribute constraint between the product type node and the extracted key entity node. Based on the path results or attribute matching returned by the query, it is determined whether the current text description conforms to the preset business logic, and an identifier representing whether the verification passed or failed is generated accordingly, i.e., the "compliance status". This compliance status can intuitively reflect whether the text description meets the requirements of supervision or internal risk control at the business semantic level.

[0064] In some exemplary embodiments, the process of querying the knowledge graph includes at least: performing logical comparison processing based on asset type compliance to obtain asset conflict status; and performing logical comparison processing based on a disabled term library to obtain text violation status.

[0065] In some specific embodiments, performing logical comparison processing based on asset type compliance is a deep verification of whether the investment targets claimed by the product conflict with the product's own type positioning. The first step in performing this process is to determine the type node of the current product (e.g., "bond fund") and obtain the associated "permitted investment scope" or "prohibited investment scope" subgraph from the knowledge graph. Subsequently, the asset entities extracted from the text (e.g., "technology stocks") are hierarchically traced within the knowledge graph. If the graph relationship shows that the asset entity belongs to a sub-node of the "stock" category, and the "stock" category node is located in the "prohibited investment scope" attribute set of the current product type, it indicates that the product description contains an investment target that contradicts the type definition. In this case, a logical conflict is determined, and an asset conflict state representing a compliance risk is generated. This comparison based on the hierarchical relationship of the knowledge graph can effectively identify implicit conflicts between specific asset names and macro asset categories.

[0066] In some specific embodiments, performing logical comparison processing based on a prohibited word library is a compliance scan to check whether there are any prohibited terms in promotional or risk disclosure texts. This logical comparison processing employs a dual-track detection mechanism: on the one hand, it directly matches words in the text against a pre-defined sensitive word entity library (such as a blacklist containing absolute terms like "principal guaranteed" or "risk-free"); on the other hand, it uses a semantic similarity calculation model to analyze the distance between the semantic vector of the text fragment and the semantic model of the prohibited promise. For example, even if the text does not use explicit prohibited words, if its semantic expression (such as "providing stable returns similar to deposits") is highly similar to the concept of "rigid redemption" in the vector space, and the confidence level exceeds a pre-defined threshold, it will still be judged as having implicit violation risk. Based on the above matching results, a text violation status indicating whether the text has compliance issues is generated.

[0067] Specifically, in operation S330, the second audit result is generated based on the compliance status. In some specific embodiments, operation S330 is responsible for converting the semantic-level verification conclusions into structured data that can be output in a standard manner. It collects the compliance statuses (including asset conflict status, text violation status, etc.) generated from the various semantic analysis stages mentioned above. For each status marked as "abnormal" or "violation," the corresponding semantic error metadata is automatically matched. This includes: generating a unique error code (e.g., "ERR_COMPLIANCE_MISMATCH" represents an investment scope conflict, and "ERR_RISK_TERM" represents a violation term), extracting the original text fragment that triggered the error as contextual evidence, and generating a brief error description text. The final generated second audit result aggregates all compliance risk items inferred based on semantics and graphs, and its data structure is consistent with the first audit result, thereby facilitating unified correction strategy matching and report generation in subsequent stages.

[0068] Specifically, in operation S400, the first and second review results are input into the correction recommendation engine. The engine then corrects the review results based on the correction rules of a preset error correction strategy mapping graph, obtaining and outputting correction suggestions and review results for the product information to be reviewed. In some specific embodiments, operation S400 constructs an "intelligent generation layer" in the review process, changing the passive situation of traditional review which can only perform "blocking error reporting," and instead providing constructive proactive repair solutions. The correction recommendation engine is a context-aware decision execution module, while the preset error correction strategy mapping graph connects the logical relationship between "problems" and "solutions." Internally, the error correction strategy mapping graph maintains a mapping relationship from various "error codes" to specific "repair algorithms." Through this step, for format errors, logical contradictions, or semantic compliance risks identified in previous stages, the corresponding knowledge computing strategy can be automatically invoked to generate correct data values ​​that meet industry standards or regulatory requirements, thereby significantly reducing the cognitive burden of manual investigation and correction, and realizing a shift from "pointing out errors" to "assisting in error correction."

[0069] like Figure 6 As shown, in some exemplary embodiments, the correction recommendation engine corrects the review results based on the correction rules of the preset error correction strategy mapping map, including: operations S410 to S420.

[0070] Specifically, in operation S410, based on the error markers and error codes in the first and second audit results, a modification content generation process based on the preset error correction strategy mapping map is performed to obtain the modification suggestion content. In some specific embodiments, operation S410 is essentially a policy routing and knowledge computation process based on error characteristics. First, the input audit results are parsed to extract each data item marked as abnormal and its corresponding unique error code (e.g., "ERR_FEE_DEVIATION" or "ERR_NAME_INCOMPLETE"). Then, using this error code as an index key, a search is performed in the error correction strategy mapping map to locate the preset repair strategy for this specific type of error (e.g., "call the statistical distribution calculation model" or "call the text completion service"). Finally, according to the logic of the strategy definition, combined with the context information of the data item (e.g., product type, affiliated organization), the corresponding data calculation or retrieval task is performed in the background to calculate the theoretically correct and compliant target data value, i.e., the "modification suggestion content". This content is not the final report format, but rather a calculated and derived clean data entity used to replace or correct the original erroneous values.

[0071] In some exemplary embodiments, the modification suggestions include numerical range suggestions and text completion suggestions. The step of generating modification content based on a preset error correction strategy mapping graph includes at least: performing numerical recommendation processing based on statistical distribution to obtain numerical modification suggestions, wherein the statistical distribution numerical recommendation processing includes querying similar product data in the knowledge graph based on the error code, performing statistical calculation processing, and obtaining the numerical range suggestions; performing pattern matching processing based on the knowledge graph to obtain text modification suggestions, wherein the knowledge graph pattern matching processing includes querying naming convention templates in the knowledge graph based on the error code, performing pattern matching processing, and obtaining text completion suggestions.

[0072] In some specific embodiments, performing statistical distribution-based numerical recommendation processing corrects deviations of numerical fields from reasonable industry levels. When an error code indicates an abnormal deviation in a numerical field (such as "management fee rate"), the category attribute of the current product (such as "index fund") is first parsed. Next, this attribute is used to perform a full search in a pre-defined knowledge graph or industry database to extract the corresponding field values ​​of all historical products of the same type. Based on the extracted large sample data, statistical calculations are performed, such as calculating the arithmetic mean, median, and standard deviation, and a reasonable numerical distribution range is constructed according to pre-defined confidence interval rules (such as the mean plus or minus twice the standard deviation). Finally, this range is output as the suggested numerical range. Performing statistical distribution-based numerical recommendation processing provides an objective reference benchmark for abnormal values ​​based on industry big data, rather than relying on fixed and rigid thresholds.

[0073] In some specific embodiments, the pattern matching processing based on the knowledge graph is used to complete missing or non-standard text fields. When an error code indicates that a text field (such as "product name") is incomplete or does not conform to the specifications, the issuing organization entity and product type entity to which the product belongs are first identified. Subsequently, the standard naming rule template for the issuing organization regarding the product type (e.g., "{organization abbreviation}{core strategy}{type suffix}") is queried in the knowledge graph. The currently known attribute values ​​are filled into the corresponding slots of the template, automatically concatenating to generate a complete and standardized text string. If there are ambiguous or uncertain fragments, the most likely words can be matched from the candidate word library of the knowledge graph using a character similarity algorithm for filling. The final generated complete and standardized string is the "text completion suggestion".

[0074] Specifically, in operation S420, based on the modification suggestion content, correction processing is performed on the data corresponding to the error-marked content to generate the correction suggestion. In some specific embodiments, this is a process of applying the calculated correct data to the original data context, especially when multiple errors exist in the same field, a hierarchical pipeline-style superimposed correction processing is performed. For example, if a field simultaneously triggers multiple error codes such as "format error" (e.g., containing extra spaces), "semantic error" (e.g., typos), and "content missing" (e.g., missing prefix), it will not be simply overwritten, but will be processed sequentially according to the priority order of the preset correction strategy: first, a format cleaning strategy is applied to remove redundant characters, and the cleaned result is used as input; then, a semantic correction strategy (e.g., based on a character edit distance algorithm) is applied to correct typos; finally, a knowledge completion strategy is applied to add missing content. After this series of sequential processing, the final correction value is generated, and it is encapsulated with the original error value and the correction reason (e.g., "based on industry average level" or "based on naming conventions") into a structured suggestion object, thereby forming the final correction suggestion displayed to the user, which contains a complete repair path.

[0075] like Figure 7 As shown, in some other exemplary embodiments, the product information verification method further includes operations S500 to S600.

[0076] In operation S500, feedback information regarding correction suggestions for the product information to be reviewed is received. In some specific embodiments, operation S500 is a key step in building a closed loop of human-computer interaction, aiming to capture the reviewer's professional interpretation data of the system output results. After the correction recommendation engine displays the generated correction suggestions (such as recommended numerical ranges or completed text) to the user, it listens for and records the user's actions. The feedback information specifically includes the decision instructions executed by the user for each correction suggestion, such as "adoption instruction" (the user directly accepts the system suggestion), "rejection instruction" (the user rejects the system suggestion and retains the original value), or "modification instruction" (the user rejects the system suggestion and manually enters a new data value). In addition, when the user performs a rejection or modification operation, the feedback information may also include the rejection reason or remarks entered by the user. For example, if the system suggests adjusting the management fee rate of a product to the industry average, but the user chooses to reject it and notes "this product has an innovative structure and a special fee rate", the system will store the "rejection" action and the context label of "innovative structure" as feedback information in a structured manner, and use it as negative sample data for subsequent model optimization.

[0077] In operation S600, based on the feedback information, the process of updating the preset rules and the knowledge graph is executed. In some specific embodiments, operation S600 utilizes accumulated feedback data to drive the system's adaptive iteration, achieving continuous evolution of the review capabilities. In some specific embodiments, for updates to preset rules, the distribution of feedback results for specific rules over a period of time is statistically analyzed. If a suggestion triggered by a certain verification rule or correction strategy is frequently "adopted" by users, the confidence weight or priority of that rule will be increased; conversely, if a suggestion generated by a rule is frequently "rejected" (e.g., the rejection rate exceeds a preset threshold), a de-weighting process will be automatically executed, reducing the confidence weight of that rule, or automatically relaxing the verification threshold of that rule to reduce the interference of false alarms on users. This dynamic weight adjustment mechanism ensures that the system rules can adapt to actual changes in business standards. In other embodiments, for updates to the knowledge graph, the entities and relationships in the graph are expanded or corrected based on the user's "modification instruction" or "confirmation of compliance" operation. Specifically, if a user manually inputs a new term that does not yet exist in the knowledge graph (such as the name of a new type of investment target), and this input is manually verified as compliant, entity extraction and linking processing will be performed. The new term will be automatically added as a new entity node to the corresponding category (such as "permitted investment scope") in the knowledge graph. Simultaneously, using a probabilistic update algorithm (such as a Bayesian update strategy), the confidence level of entity attribute values ​​or relationships between entities in the graph will be dynamically adjusted based on the strength of evidence in the feedback information. In some embodiments, after multiple senior reviewers confirm the reasonableness of the association between a certain type of product and a specific fee structure, the connection weights between the corresponding entity nodes in the knowledge graph will be strengthened, thereby enabling more accurate identification of such special cases in future reviews.

[0078] In this embodiment, the operations S100~S400, S500, S600, S110, S120, S210, S220, S230, S310, S320, S330, S410, and S420 described above can also be executed on the system side, and will not be repeated here.

[0079] Based on the above product information verification method, this application also provides a product information verification device, which will be described below in conjunction with... Figure 8 The device is described in detail.

[0080] Figure 8 A schematic block diagram of a product information verification device according to an embodiment of this application is shown.

[0081] like Figure 8As shown, the product information verification device 200 of this embodiment includes a first acquisition module 210, a first processing module 220, a second processing module 230, and a third processing module 240.

[0082] The first acquisition module 210 is configured to acquire a standard information message with a preset structure corresponding to the product information to be reviewed. In some embodiments, the first acquisition module 210 is mainly responsible for data access and standardized preprocessing. By identifying the original file format of the product information to be reviewed, it uses a preset field mapping dictionary and fuzzy matching logic to map unstructured or semi-structured original business data to a unique standard field identifier within the system, and performs data cleaning and normalization operations to generate a standard information message that conforms to the preset data model specifications and has unified semantic constraints, providing a standardized data input basis for the subsequent automated review process. In one embodiment, the first acquisition module 210 can be used to execute the operation S100, as well as operations S110 and S120 described above, which will not be repeated here.

[0083] The first processing module 220 is configured to perform verification processing based on preset rules according to the standard information message to obtain a first audit result. In some embodiments, the first processing module 220 constitutes a rigid constraint layer of the audit system, aiming to quickly intercept basic errors through a low-computational-cost rule engine. It performs atomic-level format scanning and cross-field numerical logic comparison on the standard information message by calling a preset rule base, verifying whether the data conforms to rigid constraints such as length, type, date sequence, business parameter thresholds, and inter-field dependencies, and summarizes all verification pass or fail states to generate a structured first audit result to identify explicit logical contradictions and format anomalies. The first processing module 220 can be used to execute the previously described operation S200, as well as operations S210, S220, and S230, which will not be elaborated further here.

[0084] The second processing module 230 is configured to perform natural language processing and logical comparison processing based on a preset knowledge graph according to the standard information message to obtain a second audit result. In some embodiments, the second processing module 230 implements the flexible semantic understanding capability of the audit system, aiming to solve the problem of deep compliance verification of unstructured text. It uses natural language processing technology to extract key business entities from descriptive text and combines a preset entity relationship network to perform path query and attribute matching, thereby identifying deep semantic compliance risks such as conflicts between investment targets and product types, and implicit non-compliance commitments, and generating a second audit result containing semantic compliance status. In one embodiment, the second processing module 230 can be used to execute the operation S300, as well as operations S310, S320, and S330 described above, which will not be repeated here.

[0085] The third processing module 240 is configured to input the first review result and the second review result into the correction recommendation engine, so that the correction recommendation engine corrects the review results based on the correction rules of the preset error correction strategy mapping map, and obtains and outputs the correction suggestions and review results of the product information to be reviewed. In some embodiments, the third processing module 240 realizes an intelligent leap from passive error reporting to active error correction. The third processing module 240 matches the corresponding repair algorithm in the strategy mapping map according to the error code output in the previous steps, generates the correct suggestion value by calculating the statistical distribution of industry data or the pattern matching of standard templates, and supports the processing of multiple superimposed errors through a pipeline mechanism, and finally outputs a complete review report containing accurate modification suggestions. In one embodiment, the third processing module 240 can be used to execute the operation S400, as well as operations S410 and S420 described above, which will not be repeated here.

[0086] According to embodiments of this application, any plurality of modules among the first acquisition module 210, the first processing module 220, the second processing module 230, and the third processing module 240 can be combined into one module, or any one of these modules can be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules can be combined with at least part of the functionality of other modules and implemented in one module. According to embodiments of this application, at least one of the first acquisition module 210, the first processing module 220, the second processing module 230, and the third processing module 240 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in a suitable combination of any of these. Alternatively, at least one of the first acquisition module 210, the first processing module 220, the second processing module 230, and the third processing module 240 may be implemented at least partially as a computer program module, which can perform corresponding functions when the computer program module is run.

[0087] Figure 9 A block diagram schematically illustrates an electronic device suitable for implementing a product information verification method according to an embodiment of this application.

[0088] like Figure 9As shown, an electronic device 900 according to an embodiment of this application includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage portion 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include onboard memory for caching purposes. The processor 901 may include a single processing module or multiple processing modules for performing different actions of the method flow according to an embodiment of this application.

[0089] RAM 903 stores various programs and data required for the operation of electronic device 900. Processor 901, ROM 902, and RAM 903 are interconnected via bus 904. Processor 901 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 902 and / or RAM 903. It should be noted that programs may also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in one or more memories.

[0090] According to embodiments of this application, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to a bus 904. The electronic device 900 may also include one or more of the following components connected to the input / output (I / O) interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 910 as needed so that computer programs read from it can be installed into the storage section 908 as needed.

[0091] This application also provides a computer-readable storage medium, which may be included in the apparatus described in the above embodiments; or it may exist independently and not assembled into the apparatus. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0092] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include ROM 902 and / or RAM 903 and / or one or more memories other than ROM 902 and RAM 903 described above.

[0093] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code enables the computer system to implement the product information verification method provided in the embodiments of this application.

[0094] When the computer program is executed by the processor 901, it performs the functions defined in the apparatus of the embodiments of this application. According to the embodiments of this application, the apparatus and the like described above can be implemented by computer program modules.

[0095] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via the communication section 909, and / or installed from a removable medium 911. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0096] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, it performs the functions defined in the system of the embodiments of this application. According to embodiments of this application, the apparatuses described above can be implemented by computer program modules.

[0097] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0098] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0099] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

[0100] The embodiments of this application have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of this application. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Without departing from the scope of this application, those skilled in the art can make various substitutions and modifications, all of which should fall within the scope of this application.

Claims

1. A product information auditing method characterized by comprising: The method comprises: obtaining a standard information message corresponding to a preset structure of the product information to be audited; performing a preset rule-based checking process according to the standard information message to obtain a first audit result; performing natural language processing and performing a preset knowledge graph-based logical comparison process according to the standard information message to obtain a second audit result; inputting the first audit result and the second audit result into a correction recommendation engine, so that the correction recommendation engine corrects the audit result based on a correction rule of a preset error correction strategy mapping graph to obtain and output a correction suggestion and an audit result of the product information to be audited.

2. The method of claim 1, wherein, The method comprises: obtaining the product information to be audited; based on a preset field mapping relationship, performing field mapping extraction processing on the product information to be audited to obtain the standard information message corresponding to the preset standard structure.

3. The method according to claim 1 or 2, characterized in that, The method comprises: performing a field checking process based on a preset format rule according to the standard information message to obtain a format checking result; performing a logical comparison process based on a preset logical rule according to the standard information message to obtain a logical checking result; based on the format checking result and the logical checking result, the first audit result is composed.

4. The method of claim 3, wherein, The method comprises: performing a logical comparison process based on a date sequence rule to obtain a date logical state; performing a logical comparison process based on a business compliance rule to obtain a business compliance state, wherein the business compliance rule is used to check the constraint relationship between the product type and the investment proportion in the standard information message; performing a logical comparison process based on a field association consistency rule to obtain an association consistency state, wherein the field association consistency rule is used to check the dependency relationship between the product name field and the product type field in the standard information message.

5. The method according to claim 1 or 2, characterized in that, The method comprises: performing natural language processing on the text field in the standard information message to extract key entities; based on the key entities, performing a knowledge graph query process to generate a compliance state; based on the compliance state, the second audit result is generated.

6. The method of claim 5, wherein, The method comprises: performing a logical comparison process based on asset type compliance to obtain an asset conflict state; performing a logical comparison process based on a forbidden word library to obtain a text violation state.

7. The method according to claim 1 or 2, characterized in that, The method comprises: based on the error mark content and the error code in the first audit result and the error mark content and the error code in the second audit result, performing a modification content generation process based on the preset error correction strategy mapping graph to obtain a modification suggestion content; Based on the modification suggestion content, a correction process is performed on data corresponding to the error marking content to generate the correction suggestion.

8. The method of claim 7, wherein, The modification suggestion content includes a numerical range suggestion and a text completion suggestion, and the modification content generation process based on the preset error correction strategy mapping atlas includes at least: Performing a numerical recommendation process based on a statistical distribution to obtain a numerical modification suggestion, wherein the numerical recommendation process based on the statistical distribution includes querying the knowledge graph for similar product data based on the error code, performing statistical calculation processing, and obtaining the numerical range suggestion; Performing a pattern matching process based on the knowledge graph to obtain a text modification suggestion, wherein the pattern matching process based on the knowledge graph includes querying the knowledge graph for a naming specification template based on the error code, performing pattern matching processing, and obtaining the text completion suggestion.

9. The method of claim 1 or 2, wherein, The method further includes: Receiving feedback information on the correction suggestion for the product information to be audited; Based on the feedback information, performing a process of updating the preset rules and the knowledge graph.

10. A product information auditing apparatus characterized by comprising: The device includes: A first acquisition module configured to acquire a standard information message corresponding to a preset structure of product information to be audited; A first processing module configured to perform a check process based on preset rules based on the standard information message to obtain a first audit result; A second processing module configured to perform natural language processing and logical comparison processing based on a preset knowledge graph based on the standard information message to obtain a second audit result; A third processing module configured to input the first audit result and the second audit result into a correction recommendation engine, and to cause the correction recommendation engine to perform correction processing on the audit results based on correction rules of a preset error correction strategy mapping atlas to obtain and output a correction suggestion and an audit result for the product information to be audited.

11. An electronic device, comprising: one or more processors; a memory for storing one or more computer programs, characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1-9.

12. A computer readable storage medium having stored thereon a computer program or instructions, characterized in that, The computer program or instructions, when executed by a processor, implement the steps of the method according to any one of claims 1-9.

13. A computer program product comprising computer programs or instructions, characterized in that, The computer program or instructions, when executed by a processor, implement the steps of the method according to any one of claims 1-9. The computer program or instructions, when executed by a processor, implement the steps of the method according to any one of claims 1-9.

Images